IP-Probe
User Manual
1 Introduction
1.1 Overview
IP-Probe is designed to monitor, visualize, and diagnose streaming IP traffic in real-time. It supports a range of streaming protocols and is built for broadcasters, network operators, and IT professionals who need detailed visibility into their stream quality and performance.
In live streaming - whether for contribution, production, or distribution - reliability is everything. Any issue in the signal path can cause disruptions that quickly spread through the workflow.
That's why continuous monitoring and visualization are essential. By tracking key parameters such as bitrate, jitter, packet loss, and codec performance, you gain real-time insight into the health of every stream. Clear visual dashboards make it easy to spot anomalies, pinpoint the source of problems, and take action before they affect downstream delivery.
What is happening on your Network?
With powerful monitoring in place, complex multistream environments become manageable. Engineers can oversee contribution feeds, safeguard production stability, and verify distribution quality - ensuring streams remain robust, reliable, and ready for the most demanding broadcast and media applications.
You don't want to float your network
1.2 Use Cases
- Monitoring RTP/UDP streams in broadcast environments
- Detecting packet loss, IAT and jitter in real-time
- Triggering alerts based on quality thresholds
- Integration into larger observability stacks
1.3 Supported Protocols
- RTP (Real-time Transport Protocol)
- ST 2110-20/30/40
- ST 2022-6
- PTP (Precision Time Protocol)
- MPTS (Transport Streams)
- SRT
- RIST
1.4 Licensing Overview
IP-Probe is licensed per machine. A license file (license.slc) is made by the vendor for the Unique System ID of your probe and says which options are included, until when the license is valid and until when software updates are covered. It is installed on the License page (chapter 32) and takes effect at once, without a restart.
The Base option provides the base toolset and is required. Seven optional options extend it, depending on what you need to monitor: MPTS, Streaming, 2110, 2022-6, Audio, Video and Ancillary. Options can be combined freely on one probe. The license determines which pages, data and functions are available, so a probe can be tailored to a specific use case.
What an option does not include is not only hidden: the probe does not analyze that data, and the menu shows the entry grayed out with a lock ("Requires the Streaming option"). Opening such a page shows which option it needs. To add an option, contact sales with the Unique System ID shown on the License page; you receive a new license file.
Combining options
Some pages combine a protocol with audio, video or ancillary data. They need both: the protocol option opens the page, and the audio, video or ancillary parts need their own option. For example, the ST 2110-30 page needs 2110, and its meters, waveform and listening buttons also need Audio.
Without a valid license
Without a valid license the probe runs restricted: you can log in, open the License page and install a license, but the probe does not capture or analyze traffic. A banner at the top of every page says so. This also applies to an expired license and to a probe software version released after the update expiration date of the license. The analysis pages (Protocols, Flows, Endpoints, Devices, Network Map, Packet Timing, Compare, Reports and Recordings) are grayed out with a lock ("Requires a valid license") and open a page that says the probe has no valid license. Dashboard, Inputs, Logs, Settings, License and About stay available.
Evaluation license
An evaluation license (also called a demo license) works like a full license for its options, but the header shows "Evaluation license" and the pictures and video carry a DEMO mark.
1.4.1 Option: Base
The Base option enables the base toolset and is required for all installations. It covers:
- Flows, Endpoints, Devices, traffic and the network map
- Inputs, protocol overview and jitter
- Logs, incidents and the dashboards
- Reports, comparing and recordings
- Configuration, users and the License page
Without the Base option the license is not accepted and the probe stays restricted. All other options depend on it.
1.4.2 Option: MPTS
The MPTS (Multiple Program Transport Stream) option enables analysis of MPEG transport streams. The MPTS page shows the programs of each stream with a thumbnail, codec, resolution, bitrate, the number of audio and subtitle tracks, and warnings for continuity counter errors, PCR interval problems and scrambled services. The PIDs, services and tables of a stream are on the PIDs, Services and Tables pages under MPTS.
Playing the audio of a program also needs the Audio option, and the pictures and video playback also need the Video option.
1.4.3 Option: Streaming
The Streaming option enables SRT and RIST inputs and stream analysis: the SRT and RIST pages and their connection monitors.
1.4.4 Option: 2110
The 2110 option enables SMPTE ST 2110 and its timing: the PTP page, the ST 2110-10, ST 2110-20, ST 2110-30 and ST 2110-40 pages and their flow details, and the test stream generator.
Parts of these pages need a second option: the audio parts of ST 2110-30 (meters, waveform, loudness, listening) need Audio, the picture parts of ST 2110-20 need Video, and the ST 2110-40 page needs Ancillary.
1.4.5 Option: 2022-6
The 2022-6 option enables the ST 2022-6 page: analysis of uncompressed video over IP in the SDI-in-IP format.
1.4.6 Option: Audio
The Audio option enables audio monitoring and processing wherever it appears: the Audio page, the audio meters, waveform and loudness (EBU R128) measurements, and listening in the browser. It is used together with the protocol option of the stream (MPTS or 2110) for the audio of that stream. The AV Monitor opens with Audio or Video and shows only the part that is included.
1.4.7 Option: Video
The Video option enables video monitoring wherever it appears: the Video page, thumbnails and live pictures, video playback and the scopes. It is used together with the protocol option of the stream (MPTS or 2110) for the picture of that stream. The AV Monitor opens with Audio or Video and shows only the part that is included.
1.4.8 Option: Ancillary
The Ancillary option enables ancillary data: the analysis of ST 2110-40 streams, and the Subtitles page with captions and subtitles.
2 Getting Started
2.1 Setup hardware
• Starting up the IP-Probe machine • Connecting to a stream/network
2.2 User Interface Software
... first steps to see that IP-Probe is giving network information ...
• Verifying stream/network activity
2.2.1 Initial Setup
- Open the web UI at http://localhost:8080
- Add your first stream
- Review stream metrics in the dashboard
2.2.2 Connecting to a Stream
• Input stream IP and port • Optionally set protocol and bitrate expectations
2.2.3 Verifying Stream Activity
• Green indicator for active stream • Metrics (bitrate, packets, errors) populate in real-time
3 User Interface
The User Interface (UI) is designed to provide clear navigation, consistent system status visibility, and efficient access to all features. The interface is divided into several key areas, each with a specific role in supporting user interaction and workflow.
3.1 Header
The Header is located at the top of the interface and remains visible across all screens. It provides:
- Application identification and branding
- Global status indicators and system notifications
- Access to user-related actions, such as account or session controls
The Header ensures that critical system information and global actions are always accessible.
3.2 Footer
The Footer (status bar) is located at the bottom of every page and provides real-time system status at a glance. It contains the following elements from left to right:
| Element | Description |
|---|---|
| Following / PAUSED | Indicates whether the log view is automatically scrolling to follow new entries. A green indicator means the system is actively following live log updates; a yellow indicator means auto-scroll is paused. |
| Visible / Total lines | Shows the number of currently visible log lines versus the total number of log lines received from the backend since the WebSocket connection was established (e.g. 0 / 359 lines). |
| Date & time | The current date and time as reported by the probe, displayed on the right side of the footer. |
The footer is always visible regardless of which page is active, providing persistent visibility into connection health and log activity.
3.3 Navigation sidebar
The Navigation Sidebar is positioned along the side of the interface and provides structured access to the main sections of the application. It includes:
- A grouped menu of available features
- Quick navigation between functional areas
- Visual indication of the currently active section
Some entries have sub items, which are always visible and indented under their entry. Under Sources, ST 2110 (PTP, 2110-10, 2110-20, 2110-30, 2110-40) and MPTS (Overview, ETR 290, PIDs, Services, Tables) are labels only: they open nothing, their sub items are the links. A sub item whose option the license does not include is shown with a lock, like any other entry. On the page that is open, its sub item is highlighted.
The PIDs, Services and Tables items under MPTS open that page for the MPTS stream you looked at last (the browser remembers the stream). Without a remembered stream they open the first MPTS stream in the order of the MPTS page; with no MPTS stream at all they open the MPTS overview.
When the sidebar is narrow (icons only) the sub items show as icons of their own, without indent and without the label, and a tooltip names each one. On a phone the same menu opens as a drawer. With the keyboard, Tab moves over the links only; the labels are skipped.
The Navigation Sidebar allows users to move efficiently between different parts of the system.
3.4 Filter bar
The Filter bar sits above every flow page (Dashboard, Flows, Endpoints, Devices, MPTS, AV Monitor and the protocol pages). Its settings apply to all of these pages at once and stay as they are while you move between them. It is one row:
| Control | What it does |
|---|---|
| Inputs | Which inputs are shown. The menu lists each input with its flow count (active/total) and bitrate. |
| Categories | Which categories are shown: Media, Clock and Network. Only offered when more than one category is present. |
| Protocols | Which protocols are shown, with the number of flows of each protocol. Lists the protocols of the shown categories. |
| Active flows | When on (highlighted), only flows that currently receive packets are shown. |
| Filter ... | Text filter on flow alias, input alias, source and destination address and protocol stack. Esc clears it. |
Each group button shows how much of the group is shown: all 3 (or 3 on a narrower window) when everything is shown, a highlighted 5/7 when 5 of 7 are shown, and a red none when nothing is shown. Click the button to open its menu:
- Tick or untick the items; the menu stays open, so you can change several at once.
- The top row (for example All Protocols) ticks every item; when everything is already ticked, it unticks them all.
- Only this shows only the item you last ticked or moved to with the arrow keys.
- Close the menu with Esc or by clicking outside it.
On a narrow window (below about 700 pixels of bar width) the groups fold into one Filters button. Its number counts the filters that hide something: each group that does not show everything, plus Active flows when it is on. The button opens the same groups in a panel from the bottom of the screen, with Only per item, All/Clear per group and Show all to show every input, category and protocol again. The text filter stays next to the button.
The Filter bar can be hidden and shown again with the filter button in the header or with Alt+P.
3.5 Content area
The Content Area is the central part of the interface and displays the main information and functionality related to the selected section. This area is used for:
- Viewing and interacting with data
- Configuration and management tasks
- Dashboards, tables, forms, and visualizations
The Content Area dynamically updates based on user navigation and actions.
3.6 Summary sidebar
The Summary Sidebar is a fixed panel on the right side of the content area. It provides a persistent, at-a-glance overview of the monitored network - including active flow counts, total bitrate, status breakdown by severity, and protocol distribution. The Summary Panel is described in detail in chapter 4.
3.7 Help sidebar
The Help Sidebar provides contextual assistance related to the current screen or feature. It typically includes:
- Explanations of displayed data and controls
- Usage guidance and best practices
- Links to relevant documentation or support resources
The Help Sidebar helps users understand features and complete tasks without leaving the current workflow. The built-in user manual is described in detail in chapter 5.
Open it with the ? button in the top row, the H key, or a help icon next to a value (for example Help on 1.4 on the ETR 290 page, which opens the manual at that check). Close it with the X at the top right of the panel, with Esc, with H, or with the ? button again.
3.8 Login and Connection
The login page puts the cursor in the username field, so you can type the user name and password right away.
Sign in with a passkey. A passkey is a login saved on your computer, phone or security key, which you unlock with your fingerprint, face, device PIN or the key itself. Where the probe has passkeys set up and the page is opened at its web address (an address with a host name and a trusted HTTPS certificate), the login page shows Sign in with a passkey under the password form. Select it and confirm in the dialog of your browser. In browsers that support it, the Username field also lists your saved passkeys: click the field and pick the passkey instead of typing. Either way you are signed in with the role of the passkey's owner.
If you cancel the dialog, or the passkey is not accepted, a short message appears under the Login button and the password form stays ready, so you can always sign in with your user name and password. A passkey that no longer works can be removed and added again in your profile.
The button does not appear when passkeys cannot work on the page: when it is opened by an IP address (for example https://172.16.0.52:2300), over plain HTTP, in a browser without passkey support, or when no web address is set up on the probe. When passkeys are set up for an address of the probe and you opened the page in another way, including by its IP address, the login page shows one line, for example Passkeys: use https://probe1.example.com:2300, with a link to that address.
Live data (the WebSocket connection) is only delivered after you have logged in. When you log out, or when your session ends, the data stops within a second.
User menu. The button with your initials at the right end of the top bar (with your name on wider screens) opens the user menu: your name and role, Profile... and Log out. It works with the keyboard too (Enter or Space opens it, the arrow keys move, Esc closes it).
Profile. The Profile dialog shows your name, your role and what that role may do, and your last login (time and address). Under Change password enter the current password and the new one twice. For users other than Admin the new password needs at least 8 characters and at least one digit, and may not be the same as the current one; the rules tick off while you type. When the password changes, your other sessions are logged out; this one stays.
Passkeys in the Profile. The Passkeys section lists the passkeys you have made, up to 10. Each row shows the name, when it was added and last used (point at the time for the full date), and tags: Synced (the passkey is backed up by your device or browser account) or This device, whether your device checked your PIN or biometrics when you made it, For another address (the passkey was made for another web address of the probe, the address is shown, and it cannot sign in here), and This passkey is blocked (the use counter of the passkey went back, which can mean it was copied: the passkey can no longer sign in and its sessions were ended; remove it, then add it again). In Settings > Users such a passkey is marked Blocked.
- Add a passkey: select Add a passkey, type your current password, check the name (it starts with your browser and system, for example "Edge on Windows") and select Create passkey. Confirm in the dialog of your browser or device. Adding is only possible when the page is opened at the web address of the probe; otherwise the section says why and, when passkeys are set up for another address, links to it. If the probe has no web address yet, an Admin is pointed to Settings > Web server (31.12). A wrong password keeps you signed in and you can try again; after five wrong tries within a minute the probe asks you to wait.
- Rename: the pencil icon changes the name in the row.
- Remove: the bin icon asks in the row first. Sessions that were signed in with that passkey end.
Renaming and removing also work on an address where adding is not possible.
An administrator can have open pages logged out after a time without activity (Browser Page Timeout, see chapter 31.1.2; the default is Never). The login page then shows the reason, for example "Logged out after 2 h without activity". The login page also shows why a session ended in other cases, for example after a probe restart.
When the backend cannot be reached for more than 3 seconds, a Backend not reachable screen appears. It shows a 10 second countdown and retries automatically. Click Retry now to retry at once. When the backend is back, the page reloads. After a backend restart you must log in again.
While the probe restarts. When an Admin restarts IPProbe, reboots or powers off the probe (Settings > System > Power, 31.13), every logged-in user sees a screen that says what is happening instead of Backend not reachable:
| Screen | Shown after | What it says |
|---|---|---|
| IPProbe is restarting | Restart IPProbe | About 30 s. |
| The probe is restarting | Reboot | About 3 minutes. |
| The probe is switched off | Power off | Switch it on with its power button. |
The screen shows who started it, the time since then and, for a restart or reboot, the expected time with a progress bar. Every 3 seconds it checks whether the probe is back; a probe that only accepts a connection again does not count, only the newly started IPProbe does. When it is back, the page opens the login with the note The probe was restarted. Please log in again.: sessions do not survive a restart, so everyone logs in again. After twice the expected time the screen adds This takes longer than usual. If the probe does not come back, check the screen connected to it for its address. After a power off the screen checks for one minute, then shows Stopped checking and a Check again button: switch the probe on, wait until it has started, and click it.
Embedded widgets (the widget API) need a login token (authToken, obtained from POST /api/login) or credentials.
4 Summary Panel
The Summary Panel is a persistent sidebar positioned on the right side of the interface. It provides a real-time, condensed overview of the monitored network, visible across all pages. While the main content area changes depending on which feature is selected, the Summary Panel remains constant - giving operators a consistent reference point for key metrics and system health.
4.1 KPI Statistics
The Overview section at the top of the Summary Panel lists the key performance indicators as rows, the label on the left and the value on the right, each on one line, so the panel keeps its height when a value changes:
| KPI | Description |
|---|---|
| Active Flows | The number of flows currently receiving packets. |
| Total Flows | The total number of flows detected across all inputs and protocols, including inactive flows. |
| Total Bitrate | The aggregate bitrate of all active flows, formatted in bps, kbps, Mbps, or Gbps as appropriate. |
| Networks | The number of distinct network inputs on which traffic has been detected. |
| Alerts | The number of flows with an active warning or error condition. A non-zero value warrants investigation. |
These KPIs update in real-time as traffic conditions change. They provide a quick health check without requiring navigation to any specific page.
4.2 Status Breakdown
Below the KPI grid, the Status section lists every status with a count of how many flows are currently in that state. Each row is marked with a color-coded left border. The three incident levels are the severities of the checks (Settings > Checks); a flow counts by the highest level of its open incidents. Incidents of the level Info and incidents that stay below the trigger of their check are not counted.
| Color | Status | Meaning |
|---|---|---|
| Red | Critical | At least one Critical incident is open on the flow. |
| Amber | Major | The worst open incident is Major. |
| Blue | Minor | The worst open incident is Minor. |
| Green | OK | The flow is monitored, at least one check is on for it and no incident is open. |
| Gray | No checks enabled | Every check is switched off for the flow. |
| Gray | Not active | Flow has stopped receiving packets. |
| Gray | Not Monitored | Monitoring of the flow is switched off (see 31.11); only packets and bitrate are counted. |
The Status section always shows all levels, even when the count is zero, with Not Monitored last. This makes it easy to confirm at a glance that no incident is open. The Alerts count is the number of active flows with a Critical, Major or Minor incident.
4.3 Protocol Distribution
The Protocols section lists every protocol type detected on the network, each with a color-coded icon and the total number of flows of that type. Protocols are sorted by display order and then alphabetically.
Each row shows:
- A protocol icon with a category-specific color for quick visual recognition
- The protocol name (e.g. ST 2110-20, MPTS, PTP, SRT)
- The flow count for that protocol type
This section provides a quick inventory of what is on the network without leaving the current page.
4.4 Visibility
The Summary Panel is visible by default on all pages. It is positioned as a fixed sidebar on the right side of the content area, ensuring it does not interfere with the main page content.
To toggle the Summary Panel on or off, press Alt+S on the keyboard. The visibility preference is saved per user and persists across sessions.
The panel also has a small chevron button at its top right (tooltip "Hide summary"). It collapses the panel to a thin strip at the right edge of the content area, and the page content takes the space back. Click the strip, or focus it and press Enter or Space, to show the panel again (tooltip "Show summary"). The button, the strip, Alt+S and the Summary switch in Settings > System > Display all change the same per-user setting. On a phone-width screen no strip is shown.
The Summary Panel respects the current input filter - if an input filter is active, only flows matching that filter are included in the KPI statistics, status breakdown, and protocol distribution.
The Summary Panel width is 220 pixels. When the Help Sidebar is also open, both panels are visible side by side, with the Help Sidebar appearing to the far right.
5 User Manual
IP-Probe includes a built-in user manual accessible directly from the application interface. The manual provides comprehensive documentation for all features, organized into chapters that follow the structure of the application.
5.1 Opening the Manual
The manual can be opened and closed in several ways:
- Press the H key on the keyboard to toggle the manual panel
- Click the help icon in the application header
The manual opens as a sidebar on the right side of the screen. It slides in with a smooth transition and can be used alongside the main application - there is no need to leave the current page.
5.2 Layout
The manual panel is divided into two areas:
- Table of Contents (left) - A scrollable sidebar listing all chapters and sections
- Content area (right) - The full text of the manual, rendered from Markdown with tables, images, callout boxes, and formatted code
The Table of Contents and the content area each have their own independent scroll position, so navigating the TOC does not affect the reading position and vice versa.
5.3 Table of Contents
The Table of Contents (TOC) is organized hierarchically:
- Chapters (top level) - shown as expandable sections
- Sections (second level) - shown as indented items within a chapter
- Subsections (third level) - shown as further indented items within a section
Each chapter can be expanded or collapsed by clicking its heading. Two convenience buttons are provided at the top of the TOC:
| Button | Action |
|---|---|
| Open all | Expands all chapters to show every section. |
| Close all | Collapses all chapters to headers only. |
5.3.1 Scrollspy
As the reader scrolls through the manual content, the TOC automatically highlights the current chapter and section. The active heading is marked with a blue left border and background highlight. This scrollspy feature keeps the TOC synchronized with the reading position at all times.
5.3.2 Navigation
Clicking any item in the TOC smoothly scrolls the content area to that section. The target heading is briefly highlighted with a yellow flash to help the reader locate it visually.
5.4 Contextual Help
Pages and features throughout the application can link directly to a specific section of the manual. When activated, the manual opens (if not already open) and scrolls directly to the relevant section. This contextual help mechanism allows operators to get immediate, targeted documentation without searching.
5.5 Expand and Shrink
The manual panel supports two width modes:
| Mode | Width | Description |
|---|---|---|
| Normal | 768 px | Default width, suitable for reading alongside the main content. |
| Full | 960 px | Expanded width for comfortable reading of wide tables or images. |
Toggle between modes by clicking the Expand / Shrink button in the manual header, or by pressing the F key on the keyboard.
5.6 Manual in a New Tab
The Manual button in the manual header opens the whole manual as one page in a new browser tab, at the section you are reading. The page has its own contents list, follows the light or dark theme, and needs no network connection once it is loaded.
To get a paper or PDF copy, open the page and use the browser's Print command (for example Print > Save as PDF). The printout has a title page, a contents page and each chapter on a new page.
The built-in manual and the single-page manual are generated from the same source document, ensuring they are always in sync.
6 Dashboard
The Dashboard is the default landing page after login. It provides a customizable, widget-based overview of the monitored network, allowing operators to assemble the exact combination of status indicators, charts, and data feeds that matches their operational needs.
The Dashboard uses a 12-column grid layout. Widgets can be positioned and sized freely within this grid. Each user's dashboards are stored on the probe with their user account, so multiple operators can maintain their own preferred views, and a user sees the same dashboards in every browser and on every PC. Changes are saved about a second after they are made. Dashboards that an older version kept only in the browser are moved to the probe the first time the user logs in from that browser.
6.1 Dashboard Templates
A user who has no dashboard yet starts with the Overview dashboard: the KPI Summary bar on top, then the Input Status Grid beside Active Incidents, then the Live Bitrate Timeline beside the Recent Events Feed, and Host Health across the bottom. Change it like any other dashboard; the other templates stay available under New and Manage.
A template pre-populates the grid with a curated set of widgets when a dashboard is created. Templates provide a quick starting point that can then be customized further.
| Template | Description |
|---|---|
| Overview | The default layout: KPI summary, input status grid, active incidents, live bitrate timeline, events feed and host health. |
| Blank | A single KPI bar. Start here to build a fully custom layout. |
| NOC Operations | Full-width layout for a network operations center. Includes KPI summary, active incidents, bitrate timeline, flow heatmap, protocol distribution, input status grid, events feed, and top talkers. |
| Media / Broadcast | Optimized for SMPTE 2110 stream health monitoring. Includes KPI summary, SMPTE 2110 monitor, input status grid, bitrate timeline, jitter analysis, and active incidents. |
| Executive Summary | High-level KPIs and trends without operational noise. Includes KPI summary, protocol distribution, bitrate timeline, top talkers, and flow heatmap. |
| Network Analyst | IP/MAC deep-dive layout. Includes KPI summary, flow map, top talkers, protocol distribution, jitter analysis, active incidents, and events feed. |
Selecting a template replaces the current layout. Widgets can be added, removed, repositioned, or resized after applying a template.
6.2 Adding Widgets
Click Edit Layout in the Dashboard header to change the layout, then click Add Widget to open the Widget Catalog. The catalog lists all available widget types grouped by category. Click a widget to add it to the grid at the next available position. On an empty dashboard, Edit Layout opens the Widget Catalog straight away.
A new dashboard (New in the header, or Create New in Manage) opens in layout editing with the Widget Catalog right after its name is entered. Click Save to keep the layout, or Cancel to leave it empty.
6.2.1 Managing Dashboards
Click Manage in the Dashboard header to rename, clone, export, import or delete dashboards. The Show switch in front of each dashboard hides it from the dashboard list in the header without deleting it; switch it on again to bring it back. A hidden dashboard stays in Manage and keeps its layout. When the dashboard on screen is hidden, the Dashboard switches to the first one that is still shown. At least one dashboard always stays shown, and the dashboard whose layout is being edited cannot be hidden until the edit is saved or canceled.
6.3 Available Widgets
The following widgets are available for placement on the Dashboard:
6.3.1 KPI Summary
A compact, full-width bar displaying key performance indicators at a glance. Typical KPIs include total flows, active flows, aggregate bitrate, and active incident count. This widget is included in every template and is recommended as the first row of any layout.
6.3.2 Bitrate Timeline
A time-series chart showing aggregate or per-flow bitrate over time. Useful for spotting traffic spikes, drops, or trends. The chart updates in real-time as new data arrives from the probe.
6.3.3 Active Incidents
A scrollable list of the flows that are in warning or error now. Each line shows the flow, its protocol and input, the Incident types of its current episodes, and Since: how long the incident has been going on. The start time comes from the incident history of the probe, so it stays the same after a reload of the page and in every browser; hover over it to see the date and time. Episodes of a flow less than 10 minutes apart count as one incident. The start times and types follow the probe live: a flow that goes into warning or error shows its incident type within about a second. When the probe has no history database, Since is the time this page first saw the flow in warning or error, and the type shows -. Clicking a line opens the flow. A colored dot in front of each line shows the level of the incident (Critical, Major or Minor; hover over it for the word); the worst level is listed first. The settings choose the minimum severity (Minor or worse or Major or worse), the number of lines and whether the type column is shown. The stored history of all incidents, also the ones that ended, is on the Incidents page (chapter 27.1).
6.3.4 Flow Heatmap
A grid of small colored cells, one per flow. Each cell is colored according to the flow's current severity:
- Green - OK
- Yellow - Warning
- Red - Error
- Gray - Inactive or no checks
Hovering over a cell shows the flow name and protocol. Clicking a cell navigates to the flow detail view.
6.3.5 Top Talkers
A ranked list of the flows or IP endpoints with the highest bitrate. Useful for identifying bandwidth-intensive sources and verifying that expected high-bitrate streams (such as uncompressed video) are present and at the correct rate.
6.3.6 Protocol Distribution
A breakdown of all detected flows by protocol type. Each protocol is shown with a color-coded bar indicating its share of total bandwidth, along with the flow count and aggregate bitrate. Protocols are sorted by total bitrate in descending order.
6.3.7 Input Status Grid
Shows the status of each monitored network input (network port), including link state, speed, and current receive/transmit utilization. Useful for confirming that all physical connections are active and that no input is approaching saturation.
6.3.8 Events Feed
A chronological log of recent events, including flow state changes, incident triggers, and system notifications. The feed scrolls automatically to show the latest entries.
6.3.9 Flow Map
A Sankey-style diagram showing traffic flows between source and destination endpoints. The width of each connection is proportional to its bitrate. This widget provides an intuitive visual representation of who is talking to whom and how much bandwidth each conversation uses.
6.3.10 SMPTE 2110 Monitor
A summary panel showing the health of all detected SMPTE 2110 streams (video, audio, and ancillary data). Streams are listed with their format, bitrate, and current status. This widget is designed for broadcast engineers who need a quick overview of all 2110 essences without navigating to individual source pages.
6.3.11 Jitter / IAT
A real-time chart showing inter-arrival time (IAT) statistics across monitored flows. High jitter values are highlighted, helping operators identify network paths that may be introducing timing instability.
6.3.12 Host Health
CPU, GPU, NVMe, the hottest NIC chip and memory of the probe at a glance, with the worst state. Click the NIC chip row to open that port on the Inputs page. A row is left out when the probe has no such sensor.
6.3.13 NIC Health
Link, chip and module temperature and Rx loss for each port, one row per port. Click a row to open the port on the Inputs page.
6.3.14 Temperature Trend
CPU, GPU and NIC chip temperatures over the last hour or the last day, with the warning limits as lines. Choose the range and the sensors in the widget settings.
6.3.15 Sensor Gauge
One chosen sensor as a gauge with its warning and critical limits. Add the widget several times for several sensors, and choose the sensor in the widget settings.
The four health widgets show Not available with the reason when the probe cannot read health data, and Waiting for data during the first seconds after a start. The limits are the ones set under Settings > System health.
The Cores widget of earlier versions is now on the Diagnostics page (chapter 30). A saved dashboard that still holds it shows an Admin a small Moved to Diagnostics box with a link (remove it in Edit Layout); other users do not see it.
6.4 Customizing the Layout
Widgets can be repositioned by dragging their header area to a new location on the grid. Widgets can be resized by dragging their edges or corners. To remove a widget, click the close button in the widget header.
Layout changes are saved automatically per user. Switching between templates discards the current layout and replaces it with the selected template's default arrangement.
7 Inputs
The Inputs page displays the physical and virtual network interfaces that IP-Probe is monitoring. Each input corresponds to a capture port through which packets are captured and fed into the processing pipeline.
The page shows one card per input, presenting link-level status and real-time traffic statistics. This is the starting point for verifying that all network connections are operational and that traffic is flowing at expected rates.
7.1 Input Card
Each input card displays the following information:
| Field | Description | Example |
|---|---|---|
| Name | The user-assigned name of the input (configurable in Settings > Inputs), with its description below it when one is set. Point at the name to see the input ID. | Main |
| Status | The link state of the port: UP, No link or Link lost (see below). | UP |
| Speed | The negotiated link speed, with the fastest speed of the card when the link runs slower. | 10G (25G max) |
| Duplex | The duplex mode of the connection. | FDX |
| Autoneg | Whether auto-negotiation is enabled. | Fixed |
| Chip | The NIC chip temperature with its state icon; left out when the card has no chip sensor. | 78 C |
| MAC | The hardware (MAC) address of the network interface. | 68:05:ca:a3:12:f0 |
| Card | The network card (highlighted) and its PCI address. | 0000:07:00.0 |
The cards follow the order set on Settings > Inputs. The values of Bitrate, Usage and Packets/s are shown in the brand color.
Link states. A port that has had no link since it started shows a gray No link with a short note: nothing is monitored on it and no incident is raised (most installations cable only one port of a card). It does not count in the page banner or in Problems only; health starts when the first link comes up. A link that was up and went down shows an amber Link lost with an incident and a note that says when it went down. While a port has no link (gray No link or Link lost), the module values (supply voltage, Tx bias, Tx and Rx power) are shown but not judged: a module without fiber reads a very low Rx power, and the missing link is already shown. They count again as soon as the link is up. An Admin or NOC user can click Reset in that note to turn the port back to gray No link, for example after a cable was removed on purpose.
Total received. The health details (Details) show the bytes received on the port under Rx counters since start, for example "1.32 TB since 10 Oct 2026 07:18". An Admin or NOC user can reset the count with the small button next to it, to count from that moment; the reset is written to the audit log.
7.2 Receive and Transmit Panels
Each input card has a Receive panel and, when it is switched on, a Transmit panel. A capture port does not transmit, so Transmit is hidden by default; an Admin shows it for all users with Show transmit on Settings > Inputs (31.1.8). Without it, the Receive panel uses the full width of the card. Both panels display the same set of metrics for their respective traffic direction:
| Metric | Description |
|---|---|
| Bitrate | The current data rate in the given direction, formatted in bps, kbps, Mbps, or Gbps as appropriate. |
| Usage | The percentage of the link's capacity currently in use. Calculated as (current bitrate / link speed) x 100. |
| Packets/s | The number of packets per second in the given direction. |
Chart scale. The chart of each panel shows the last minutes of the bitrate. Its scale starts at 0 and ends a little above the highest value in the shown minutes (about 20 % more, rounded), so a steady stream of 2 Gbps on a 25G port is a line you can read and not a flat line at the bottom. The scale therefore changes with the traffic. The link speed is written at the top right of the chart (for example "of 25 Gb/s"); Usage shows how full the link is.
Usage monitoring is particularly important for detecting links that are approaching saturation. A receive usage consistently above 80% on a 10G link may indicate that the link is at risk of packet drops during traffic bursts.
Input names, descriptions, and display order can be configured on the Settings > Inputs page (31.1.8). See chapter 31 for details.
7.3 Port Health
The top row of an input card is plain: it names the port and shows the facts that identify it (status, speed, duplex, autoneg, chip temperature, MAC and card). Everything about the condition of the hardware is in the Health strip directly below it. Each item in the strip is a small button with a value and a state icon (a green check for OK, an amber triangle for a warning, a red cross for critical):
| Item | What it shows |
|---|---|
| Link | The link speed, with the number of link flaps in the last 24 hours when there were any. |
| NIC chip | The temperature of the network chip. Not shown when the card has no readable chip sensor. |
| Module | The temperature and the receive power of the SFP or QSFP module, read from its built-in diagnostics (DDM). |
| PCIe | The negotiated PCIe link (for example x8 Gen 4) and the count of PCIe errors. |
| Rx loss | Packets the NIC dropped before IP-Probe could read them (missed packets and receive errors). |
A strip item is only shown when the port has that kind of sensor. A replay input has no NIC, module or PCIe link and shows a note that health does not apply.
Click an item to open the details at that item, or click Details at the right end of the strip to open or close them. The details show:
- NIC and PCIe - the controller, the driver and its firmware, the chip temperature with its warning and critical limits, the PCIe link and the PCIe errors (corrected and uncorrected), and the link time.
- Module (DDM) - the module type and wavelength, and a table with temperature, supply voltage, Tx bias, Tx power and Rx power: the value now, the warning limits and the alarm limits (low and high). A value outside its warning range is amber, outside its alarm range red.
- Rx counters since start - the total bytes received on the port (Total received, for example "1.51 TB since Oct 11, 2026, 11:53"), receive errors, missed packets and packets without a buffer. An Admin or NOC user can reset the total with the small button next to it; the reset is written to the audit log.
- Last hour - small graphs of the chip temperature, the module temperature and the Rx power.
- Health events - the health events of this port, with a link to the log (All in Logs).
The chip temperature is read from the network card; an info icon next to the value says how. When a chip cannot be read, the details say why.
The buttons at the top right of the page choose how many cards show their details: All details, Problems only (the default: only ports with a warning or a critical state are opened) or None. A banner above the cards counts the warnings, for example "2 health warnings on 1 port", with one line per warning that says what to do (clean or replace the fiber, reseat the card, improve the airflow). A port with no link since the start (gray No link) is not counted.
The limits that turn an item amber or red are set under Settings > System health (see chapter 31.8). A warning also raises an incident and a line in the log; while the warning stays, one reminder line is logged per interval.
7.4 Capture Stopped by Health
A NIC chip or module that gets too hot can be damaged. When the NIC chip temperature or the module temperature reaches its critical limit, and Stop port on critical is switched on for that check, IP-Probe stops capture on that port. A chip with two ports stops both. The card then shows a red box: Capture stopped by health at the time, the sensor and its value, how long the port has delivered no data, and what is needed to start again. The flows of the port show the gap as "capture stopped by health" instead of a packet loss.
The sensors keep being read while the port is stopped. The port restarts by itself when the sensor has stayed below the warning limit, minus the clear margin, for the restart hold time (5 minutes by default). An Admin can also click Restart now and confirm; if the sensor is still critical, capture stops again. Other users see the note Only an Admin can restart it now. Only the NIC chip and the module temperature can stop a port; the host sensors (CPU, NVMe, GPU, memory) never stop capture.
7.5 Receive Mode and Port Warnings
On a probe started with the receive mode rss (a startup option, see the Linux setup guide), the network card spreads each port over several receive queues, and one Worker analyzes each flow. Use rss for many flows with small packets; use the default pipeline mode when single flows are above about 1 million packets per second (a 2160p50 or 2160p59.94 ST 2110-20 video flow with standard 1200-byte packets is about 0.9 to 1.05 million; 2160p25/30 about half that).
IP-Probe warns about two cases on an rss port. Both show in the log and as an event in the port's health events:
- A flow near the limit of one Worker: a flow above 1.2 million packets per second for 3 seconds. The warning appears once per flow and returns only after the flow has dropped below 1.0 million. Packets of such a flow are lost at the network card; switch the port to pipeline mode.
- The card cannot include UDP ports in the hash: flows to one destination that differ only in UDP port then land on one Worker. Use pipeline mode for that port.
7.6 Help Balloons
Values with a small i icon have a help balloon. Click the icon to open it: it names the value and explains what it counts. Click the icon again, or click elsewhere, to close it. The balloon follows the light or dark theme.
7.7 Reset Input
After a test on the IP network you can start fresh on one input, or on the whole probe. An Admin or NOC user sees a Reset input button at the bottom of each input card and a Reset all inputs button at the top right of the page. Other users do not see them.
A reset first asks for a confirmation that names the input (or says how many inputs are reset); Cancel is selected, and the red button confirms. The reset cannot be undone, and a new reset is accepted a few seconds after the last one at the earliest. While it runs the button says Resetting...; afterwards a line on the card (or under the page title) says how many flows were forgotten and how many history rows were deleted, or why the reset was refused, for example when another reset is still running.
What a reset clears, for the input (or for every input):
- Flows - every flow seen on the input is forgotten: it disappears from the flow lists with its counters, error counters, graphs and analysis (ST 2110, transport stream and ETR 290, SRT and RIST, loudness, thumbnails, picture and audio checks). Traffic that is still running shows up again within a second as new flows, with new flow IDs and counters that start at the reset.
- Incidents - the open incidents of those flows end with the reason "reset" (the log shows a CLEAR line ending in "(input reset)", or "(reset)" for an ETR 290 episode). An alarm that the alarm export had raised for them gets its normal clear message, and no further reminders are sent.
- History - the stored history of the input is deleted: incidents and their episodes, black, freeze and audio check episodes with their pictures, ETR 290 episodes and statistics, and SCTE-104 messages. When the history database is not written at that moment (for example after it was damaged), the history is not deleted and the result line says so; reset the input again once the database is written. The same applies when another history database is opened while the deletion runs (Start fresh, a recovery or a restore).
- Network port - the counters of the network port (total received, receive errors, missed packets, packets without a buffer) count from the reset, the link flap count starts over, and the charts of the card (bitrate, chip and module temperature, Rx power) and the port's health events start empty.
What a reset keeps: recordings (their files, and recordings that are running keep running), the configuration (aliases, groups, checks, users and all settings) and everything of the other inputs. The reset is written to the audit log and to the log with the user name and the counts.
A reset is meant to start fresh after a test, not to hide a fault: a fault that is still there is detected again on the new flows and raises its incident again.
8 Protocols
The Protocols page provides a high-level summary of all detected network traffic, organized by protocol type. It is designed for operators who need a quick answer to the question: What is on this network?
The page has three sections: a KPI summary bar, a protocol breakdown table, and a flow health grid.
8.1 KPI Summary Bar
The KPI summary bar displays four key performance indicators across the top of the page:
| KPI | Description |
|---|---|
| Total Flows | The total number of flows detected across all inputs and protocols. |
| Active Flows | The number of flows currently receiving packets. |
| Total Bitrate | The aggregate bitrate of all active flows. |
| Active Incidents | The number of flows with an active warning or error condition. A non-zero value warrants investigation. |
8.2 Protocol Breakdown
The Protocol Breakdown section lists every protocol type detected on the network. Each row shows:
| Element | Description |
|---|---|
| Color dot | A color-coded indicator identifying the protocol category. |
| Icon | A protocol-specific icon for quick visual recognition. |
| Label | The human-readable protocol name (e.g. ST 2110-20, MPTS, PTP). |
| Bar | A horizontal bar showing the protocol's share of total bandwidth relative to the highest-bandwidth protocol. |
| Flow count | The number of flows of this protocol type. |
| Bitrate | The aggregate bitrate of all flows of this protocol type. |
Protocols are sorted by total bitrate in descending order, so the most bandwidth-intensive protocol types appear at the top.
Clicking any protocol row navigates to the Flows page, where the full list of flows can be inspected.
8.3 Flow Health Grid
The Flow Health grid displays a small colored cell for each detected flow. The color of each cell reflects the flow's current status:
| Color | Status | Meaning |
|---|---|---|
| Gray | Not active | Flow has stopped receiving packets. |
| Green | Active | Flow is receiving packets and has no open incident. |
| Blue | Minor | The worst open incident is Minor. |
| Amber | Major | The worst open incident is Major. |
| Red | Critical | At least one Critical incident is open. |
Hovering over a cell displays a tooltip with the flow name and protocol type. Clicking a cell navigates to the detail view for that flow.
The health grid is especially useful in large environments with hundreds of flows. At a glance, an operator can confirm that all cells are green (or blue) and immediately spot any red or yellow outliers.
If no flows have been detected, the page displays a placeholder message: "No streams detected. Waiting for data..."
9 Flows
The Flows page is the central inventory of all network flows detected by IP-Probe. A flow represents a unique combination of source address, destination address, ports, and protocol that identifies a distinct stream of packets on the network.
The page supports two view modes (Grid and List), multi-level grouping, and a slide-out detail drawer for inspecting individual flows without leaving the page.
9.1 View Options
The view options bar at the top of the page provides controls for organizing and displaying flows:
9.1.1 Group By
Flows can be grouped by one or more dimensions using the Group by chip selector:
| Dimension | Effect |
|---|---|
| Input | Groups flows by the network input on which they were captured. |
| Protocol | Groups flows by protocol type (e.g. ST 2110-20, MPTS, PTP). |
| Severity | Groups flows by their current status (Critical, Major, Minor, OK, No checks enabled, Not active, Not Monitored). |
Multiple dimensions can be selected simultaneously to create nested groups (e.g. Input > Protocol). Each group header shows the group name and the number of flows it contains.
9.1.2 Layout
The Layout switch (the same control on the Flows, Endpoints, Devices and ETR 290 pages, in the view bar) switches between two display modes:
- Grid - Flows are displayed as cards in a responsive grid. Each card shows the flow name, protocol icon, and key metrics. Cards have a colored left border indicating the flow's severity.
- List - Flows are displayed as compact rows in a vertical list, showing the same information in a denser format suitable for environments with many flows.
9.1.3 Show
The Show chip selector controls which detail fields are visible on each flow card or row.
9.2 Group Tabs
When flows are grouped, a tab bar appears above the view options. The tabs allow quick filtering:
- All - Shows all groups.
- [Group name] - Shows only the flows in the selected group.
- Ungrouped - Shows flows that do not belong to any group.
The group you choose here applies to the whole application: the Dashboard, MPTS, ETR 290, Packet Timing, PTP, ST 2110, Subtitles, Protocols, Reports, Topology and the other pages that list flows show only that group. Each of those pages shows a chip Group: name in its toolbar; click the x on the chip, or choose All on the Flows page, to show all groups again.
9.3 Flow Card
Each flow card (or list row) displays the following information:
| Element | Description |
|---|---|
| Activity dot | Green when the flow is currently receiving packets; gray when inactive. |
| Protocol icon | An icon representing the detected protocol type. |
| Flow name | The flow's alias (if configured) or a default label such as "Flow 42". |
| Protocol label | The human-readable name of the protocol (e.g. ST 2110-20, MPTS). |
| Severity border | A colored left border indicating the flow's current health status. |
When the detail level is set to show additional fields, the card also displays:
| Field | Description |
|---|---|
| Input | The name of the input on which the flow was captured. |
| VLAN | The VLAN ID, if the flow uses VLAN tagging. |
| Source | The source IP address and port. |
| Destination | The destination IP address and port. |
| Bitrate | The current data rate. |
| Packets/s | The current packet rate. |
For SRT and RIST flows, a badge is displayed showing the monitor name.
9.4 Flow Detail Drawer
Clicking a flow card opens a detail drawer on the right side of the page. The drawer shows comprehensive information about the selected flow without navigating away from the Flows page.
The drawer content is the same as the flow detail page described in section 9.5, with the Summary, Network, payload, Log, History and Hex Viewer tabs. The History tab shows the stored incidents of the flow for the last 24 hours, 7 days or 30 days (chapter 27.1).
The drawer can be closed by clicking the same flow card again or by selecting a different flow.
9.5 Flow Details
The flow details show everything the probe knows about one flow. They open in the drawer of the Flows page (section 9.4) and as a page of their own, for example when you follow a flow link from another page. The page has a header with the flow name (the alias, or "Flow" and the ID), the protocol, the brand of the sender (the manufacturer of the source MAC address, left out for a router MAC), the source and destination address, the recording tag and Record button (chapter 29), the bitrate and the state Active or Inactive.
The details are organized in tabs. Which tabs you see depends on the flow:
| Tab | Content |
|---|---|
| Summary | Identity, interface, key metrics, media parameters and packet statistics (9.5.1). |
| Network | The protocol layers of the flow: Ethernet, VLAN, IP, UDP and RTP (9.5.2). |
| Analysis | Only for a flow that is not monitored: it says that the flow has no analysis. |
| ST 2110-20 Video, ST 2110-22 Video, ST 2110-30 Audio, ST 2110-31 Audio (AES3), ST 2110-40 ANC or ST 2022-6 | The payload tab, named after the type of the flow (9.5.3). |
| MPTS | Only for an MPTS flow: the programs and PIDs of the transport stream (chapter 18). |
| Log | The log lines of this flow, with time, level and text. |
| History | The stored incidents of the flow for the last 24 hours, 7 days or 30 days (section 27.1). |
| Hex Viewer | A captured packet of the flow, decoded layer by layer (9.5.4). |
The Summary, Network, payload and Log tabs have a Save to file button that saves the tab as a text file named after the flow ID, the tab and the time. The payload tab and the MPTS tab need the license option of their type (2110, 2022-6, Ancillary or MPTS); without it the tab says which option is missing.
A flow that is switched off in Settings > Flow Monitoring (section 31.11) shows a bar with its Not Monitored tag and the note that only packets and bitrate are counted. It has the Summary, Network, Analysis, Log and History tabs, but no payload, MPTS or Hex Viewer tab.
9.5.1 Summary Tab
The Summary tab is a stack of cards. A card is shown only when it applies to the flow.
Flow. The identity of the flow.
| Field | Description |
|---|---|
| ID | The number IP-Probe gives the flow. It is the same in all views. |
| Alias | The name you gave the flow (for example "Channel-5-HD"). Set it on the Aliases page (section 31.2). |
| Description | Your note about the flow, such as the codec, the program or the transport type. Set it on the Aliases page. |
| First seen, Last seen | When the first and the latest packet of the flow arrived. They show when a flow started and when it stopped. |
| Analysis | Only for a flow that has stopped: Paused, no traffic since the time of the last packet. |
Interface. The input on which the flow was captured: its ID, Alias and Description. Name an input on the Aliases page.
SRT Origin and RIST Origin. Only for SRT and RIST flows. The monitor that serves the flow and its latest values: RTT, Loss, Buffer and Retrans (SRT) or Quality, RTT, Loss and Recovered (RIST), and the Score. See chapters 19 and 20.
Key Metrics. The values to judge the flow at a glance.
| Field | Description |
|---|---|
| Active | Yes while packets arrive, otherwise Last seen: and the time since the last packet. |
| Severity | The health of the flow, from the checks that apply to it (see below). |
| Bitrate | The current data rate of the flow. |
| Packet rate | The current number of packets per second. |
| Type | The type of the flow, for example 2110-20. |
| Protocol Stack | The protocols from the lowest to the highest layer, for example ETH:IPv4:UDP:RTP:2110-20 or ETH:VLAN:IPv4:UDP:RTP:2110-30. |
| Category | The group of the type, for example 2110 or MPTS. |
| Source, Destination | The source and destination address and port. The destination is a multicast group for most media flows. |
The Severity values are:
- Not active - the flow is not receiving packets.
- No checks enabled - the flow is active, but every check is switched off for it.
- OK - the flow is within its limits: at least one check is on and no incident is open.
- Minor, Major, Critical - the highest level of the incidents that are open on the flow now, as set per check in Settings > Checks.
ST 2022-7 Protection. Only for the legs of a protected pair.
SMPTE ST 2022-7 sends the same RTP stream twice, over two network paths (usually called red and blue), so a receiver can rebuild the stream when one path loses packets. The probe finds the two flows (legs) of such a pair by itself: the same RTP stream (SSRC and payload type) from another source address, on another destination, port, input or VLAN. It then compares the payloads of the two legs; only when they carry the same packets is the pair confirmed (Copies verified). Flows that turn out to carry different packets, or whose packets never line up within 10 seconds, are not paired. One stream seen twice from the same source (for example one multicast stream reaching two probe ports through the same switch) is not paired automatically. A pair can also be set up by hand in the aliases configuration, with a label and the receiver class to check against.
The ST 2022-7 Protection panel appears on the Summary tab of both legs and shows:
- The class the pair needs: the path differential (PD) is the arrival time difference of the two copies of a packet. ST 2022-7 defines receiver classes by how much PD they can absorb: class D 150 us, A 10 ms, B 50 ms and C 450 ms (150 ms for streams of 270 Mbit/s or more). The panel names the smallest class that copes with the PD of the last 10 seconds, for example "Needs class B: PD max 23.0 ms".
- PD now, PD max and p99: the latest value, the largest value and the 99th percentile of the last 10 seconds, and the largest value of the whole run. The probe can measure a PD of up to about 600 ms for ST 2110-20 video and audio; for other high-rate streams (ST 2022-6, ST 2110-22) about 120 ms at HD rates. When one leg's copies arrive later than that, the class shows Not measurable with the limit, instead of a class.
- Configured class: the class set for the pair (in its aliases entry, or by a check rule). When the PD goes over it, the tag turns red and an incident is raised.
- Loss after merge: packets that neither leg delivered, so a receiver could not rebuild them. Rescued: packets that one leg lost and the other delivered.
- Loss after merge per receiver class: what a class D, A, B and C receiver would have lost on the same arrivals. A packet that only the later leg delivered is lost for every class whose limit is smaller than that leg's delay.
- Protected time: the share of the time in which both legs were received.
- One row per leg: its destination and input, its state (Live, Late when its copies arrive too late to be used, Silent when nothing arrived for 300 ms), the packets received, lost on that leg, rescued by that leg, the late copies, and how far it is behind the earlier leg (Lag). The leg that arrives first is marked "first".
Incidents. A confirmed pair raises these incident types (see 27.1 for the Incidents page):
- ST 2022-7 Loss After Merge (
st2022_7_loss_after_merge, an error, on the red leg): packets that no leg delivered. - ST 2022-7 Path Loss (
st2022_7_path_loss, a warning, on the leg that lost them): packets one leg lost. The other leg usually delivered them; the RTP sequence gap of that leg is reported as well. - ST 2022-7 Path Differential over the receiver class (
st2022_7_class_exceeded, on the red leg, only when a class is configured): the PD of a second went over the class. It is raised again only after 5 seconds within the class. - ST 2022-7 Path Differential Change (
st2022_7_skew_change, a warning, on the leg that arrives later): the PD moved to another value (by more than 0.5 ms and more than a quarter) for two seconds in a row, for example after a network path changed.
Video Parameters. For ST 2110-20 flows: Type, Resolution, Scan format (progressive or interlaced), Frame rate and Component depth (bits per color component). For ST 2110-22 (JPEG XS) flows the card is called Video Parameters (JPEG XS) and shows Type, Resolution, Scan format, Frame rate and the Packetization. The color sampling is on the payload tab.
Audio Parameters. For ST 2110-30 and ST 2110-31 flows: Type, Sample depth, Sample rate and Channel count.
MPTS. For MPTS flows: the number of Programs and Streams, the TS ID, the Network name, the Total bitrate and the Null stuffing. CC errors and PCR violation are shown in red when they occur. The programs are on the MPTS tab.
Packet Stats. The jitter of the flow (Jitter (RFC 3550)) and the packet inter-arrival time (IAT Average, IAT Min, IAT Max and the number of IAT Samples). Flows that have a payload tab show these values there.
9.5.2 Network Tab
The Network tab shows the protocol layers of the flow, one card per layer: Ethernet, VLAN, IP, UDP and RTP. A card appears only when the flow has that layer.
Ethernet. One block for the source and one for the destination MAC address.
| Field | Description |
|---|---|
| Address | The MAC address of the endpoint, for example 00:1A:2B:3C:4D:5E. |
| Alias | The name you gave the endpoint on the Aliases page. |
| Description | Your note about the endpoint. |
| Manufacturer (OUI) | The vendor, looked up from the first 24 bits of the address. Some systems use random MAC addresses, which hide the real vendor. |
| Usage Short, Usage Long | What the address is for, when it is a well-known one (for example "STP" and "Spanning Tree Protocol"). |
A MAC address has 48 bits. The first 24 bits are the OUI that IEEE assigns to the manufacturer, the last 24 bits identify the interface. Two bits in the first byte have a meaning: the I/G bit (0 is a unicast address, 1 a multicast or broadcast address) and the U/L bit (0 is assigned by the manufacturer, 1 is set locally by software or by hand).
VLAN. For a VLAN tagged flow: the VLAN ID and the Priority.
IP. The IP layer, with the source and the destination address as sub blocks.
| Field | Description |
|---|---|
| TTL | Time to Live, 1 to 255. Each router lowers it by one. A flow that arrives with a lower TTL than before has taken another path. |
| Checksum | How many packets had the IP checksum value 0xffff. Receivers may accept it as valid, but the value is easily damaged when a device changes the header (TTL decrement, address translation). |
| Flags | Which of the MF, DF and reserved bits were set in the IPv4 header. |
| Padding | Whether the Ethernet frame is longer than the IP packet, or the IP packet is longer than the UDP datagram. This is allowed but unexpected: it can confuse receivers that take the payload length from the IP length, or point to a sender that writes wrong length fields. Frames shorter than 64 bytes must be padded. |
| Address, Addr Extended | The IP address. For IPv6 the fully expanded form is shown as well. |
| Alias, Description | Your name and note for the IP address, from the Aliases page. |
| Range, Usage Short, Usage Long | The special-purpose range (private, loopback, link-local, multicast) the address belongs to, as set aside by IANA. |
UDP. The Checksum (whether the sender computes the UDP checksum, which is optional for IPv4 and required for IPv6) and the source and destination Port. The destination port tells where receivers listen, and is shared by all receivers of a multicast group.
RTP. For RTP flows: the Payload type, the SSRC (shown in hexadecimal), the number of Extensions, the CSRC count (contributing sources) and Padding.
9.5.3 Payload Tab
The payload tab is named after the type of the flow and shows what the probe reads from the media packets.
9.5.3.1 ST 2110-20 Video
The ST 2110-20 (Uncompressed video) card shows the packetization of the video, as measured from the packets. Together with the sender type it gives the traffic profile that ST 2110-21 judges the sender on. A second card, Video parameters, shows the Type, Resolution, Scan format, Frame rate, Component depth and Color sampling.
| Field | Description |
|---|---|
| Packing Mode | How the pixel components are arranged in the RTP payload (see below). |
| Sender type | The pacing behavior of the sender (see below). |
| SRD count | The number of Sample Row Data headers in the packets. |
| Bits per pixel | The number of bits per pixel, inferred from the SRD headers. |
| Pgroup size | The smallest block of pixels that is packetized together. It follows from the pixel format and the chroma subsampling. |
| Pgroup construction | The order of the sample components in the payload of a pgroup. |
| Pixels per packet | The number of pixels in a packet. |
| Packets per frame | The number of RTP packets that carry one video frame. |
Packing Mode.
| Mode | Name | Description |
|---|---|---|
| GPM | General Packing Mode | Bit-packed sample groups. A packet may start at any bit position, even in the middle of a sample group. Most efficient, but the hardest for receivers. |
| GPMC0 | General Packing Mode with C0 alignment | GPM where every packet starts at a sample group boundary (the first chroma sample, C0). Pixels are spread evenly over a fixed number of packets per line and no packet holds pixels of two lines, so all packets have the same size. Easier for receivers. |
| BPM | Block Packing Mode | A legacy byte-aligned mode with whole blocks only (180 bytes for 4:2:2). Easy for old hardware, but wasteful for 10 bit and more. Not recommended for new designs. |
The packing mode sets the payload size and so the packet rate and the spacing of the packets. For GPMC0 and 10 bit 4:2:2 a packet carries 480 pixels for 2160p (8 packets per line) and 1080 (4 packets per line), 320 for 720p (4 packets per line) and 360 for 480i and 576i (2 packets per line).
Sender type. Receivers and switches need it to size their jitter buffers and queues.
| Sender type | Burstiness | Typical use |
|---|---|---|
| N (Narrow) | Very low, very even spacing | Deterministic broadcast senders (FPGA, ASIC). |
| NL (Narrow Linear) | Low, allows slow drift | Software and virtualized senders. |
| W (Wide) | Highest, tolerates bursts | Standard network cards, non-deterministic networks. |
SRD count. The ST 2110-20 header holds one, two or three SRD headers. 1 is the simple case, 2 improves the packet efficiency and 3 is rare. The value shows how many SRDs one or more packets had.
Bits per pixel. Receivers can derive the possible component depth and subsampling from it.
| Bits per pixel | Format |
|---|---|
| 8, 10 | Key or alpha signal of 8 or 10 bit |
| 12 | 4:2:0 8 bit, or key 12 bit |
| 15 | 4:2:0 10 bit |
| 16 | 4:2:2 8 bit, or key 16 bit |
| 18 | 4:2:0 12 bit |
| 20 | 4:2:2 10 bit |
| 24 | 4:4:4 8 bit, or 4:2:2 12 bit |
| 30 | 4:4:4 10 bit |
| 32 | 4:2:2 16 bit |
| 36 | 4:4:4 12 bit |
| 48 | 4:4:4 16 bit |
Pixels per packet. The number may be the same in every packet or vary. A fixed number gives predictable packet sizes and rates and is typical for hardware senders. A varying number is common for software senders: most packets carry a nominal number of pixels and some carry fewer, for example at the end of a line. The number is always a multiple of the pgroup size and fits in the MTU after the RTP, UDP and IP headers. Receivers must take the pixel count from the payload length and not assume a fixed value.
9.5.3.2 ST 2110-22 Video (JPEG XS)
For an ST 2110-22 flow the tab shows two cards. ST 2110-22 (JPEG XS) lists the resolution, profile, sampling, slice height, frame rate, compression ratio, packetization and transmission mode, declared against measured bitrate, packets per frame and the size of the last codestream. JPEG XS errors lists the frame counter gaps, missing markers and codestream size mismatches, and any other packet problem that occurred. Until the first JPEG XS picture header has arrived the first card says it is waiting for it. These flows are also described in section 14.12.
9.5.3.3 ST 2110-30 and ST 2110-31 Audio
The card is called Uncompressed Audio for ST 2110-30 and AES3 Audio for ST 2110-31. It shows the Sample depth, Sample rate, Channel count, Packets per second, Packet time (the audio duration of one packet), Packet size and the Samples/Channel/Packet. The levels, the waveform and listening are on the ST 2110-30 page (chapter 15), where section 15.9 explains the AES3 checks of ST 2110-31.
9.5.3.4 ST 2110-40 Ancillary Data
The ST 2110-40 Ancillary Data card shows the timecode when the flow carries one (with drop frame or not and the number of timecode packets), the RTP Packets, ANC Packets, Sequence Gaps and Duplicates, a table of the DID/SDID pairs found with their description and count, and the AFD code and aspect ratio when present. See chapter 16.
9.5.3.5 ST 2022-6
For an SDI over IP flow the tab shows the SDI Format (format, resolution, frame rate and scan), the HBRMT Header (VSID, reference clock, scrambling, FEC, clock frequency, MAP, frame code, frame rate code and sample code), the Embedded Audio (audio groups, total channels, sample rate and bit depth), the VANC Ancillary Data and the Statistics. See chapter 17.
9.5.4 Other Tabs
Analysis. Shown instead of the payload and MPTS tabs for a flow that is not monitored. It says that the flow has no analysis, and an Admin gets a link to Settings > Flow Monitoring (section 31.11) to switch monitoring on.
MPTS. The programs, PIDs and tables of the transport stream, as described in chapter 18.
Log. The log lines that concern this flow, with time, level and text. The tab says "No log entries for this flow" when there are none.
History. The incidents of this flow that were stored, for the last 24 hours, 7 days or 30 days (section 27.1).
Hex Viewer. A packet of the flow, decoded layer by layer. The upper part shows the packet as a Tree of layers and fields or as a Trace; click a field to mark its bytes in the lower part. The lower part shows the bytes in Hex, Dec, Binary and Ascii, as you choose with the Display boxes, and the Address Size setting sets the bytes per row. Refresh takes a new packet and Save to file saves the decoded packet as text. The splitter between both parts can be dragged.
9.6 Flow Picker
The flow picker is the control that chooses flows on the ST 2110-20, ST 2110-30, ST 2110-40 and ST 2022-6 pages, on the Packet Timing page (the Flow and Compare views) and, for several streams at once, on the MPTS page. It looks and works the same everywhere; only the value it shows next to each flow differs by page.
The button. The closed picker shows the health dot of the chosen flow, its type, name and destination address, and the main value of the page (for example the IAT p-p on Packet Timing). Next to it are the arrows that step to the previous or next flow, and a button that opens the flow's own details. On a narrow bar the address and the type are left out. The chosen flow is in the page address (for example ?flow=12) and the browser remembers it, so a link or a reload opens the same flow.
The list. Click the button (or press Alt+Down) to open the list. Each row shows a star for pinning, the health dot, the name with its type and source below it, the address, the bitrate, the page's own columns and a status word (the color is never the only sign). A flow that is not monitored is listed dimmed with Not Monitored and cannot be chosen. A note in the footer says how many flows the filter bar of the page hides.
| Control | What it does |
|---|---|
| Search | Finds flows by name, alias, address, type, input or group while you type; every word you type must match. Typing anywhere in the list goes to the search. |
| All / Active / Problems | All lists every flow, Active the flows that send now, Problems the flows with a warning or an error. Each button shows how many flows it holds. |
| Sort | Worst first (errors, then warnings, then the rest by name), Name, Bitrate (highest first) or Address. |
| Group | Groups the rows by Type, Input or flow Group, or None. |
| Pins | The star of a row pins the flow. Pinned flows are listed first under Pinned, in the chosen sort order. A pin follows the flow (its source and destination address and input), so it stays when the probe restarts. |
The sort and the grouping are remembered per page, the pins for all pages, in this browser.
Choosing several flows. Where a page works with more than one flow (the Compare view, the MPTS page) the picker is in multiple mode: a check box in each row, the button says how many flows are chosen ("2 of max. 4 flows"), and the chosen flows show as chips next to it, each with a button to remove it. The limit is the number of flows the page can show; one more flow is refused with a message in the footer. Select shown and Clear in a group heading choose or drop the whole group.
Keys.
| Key | Action |
|---|---|
| Up, Down | Move the highlight; the arrows on the closed button step to the previous or next flow. |
| Enter | Single mode: show the highlighted flow and close. Multiple mode: close. |
| Space | Multiple mode: add or remove the highlighted flow (a space in a search that holds text). |
| Ctrl+A, Ctrl+Backspace | Multiple mode: choose or clear all flows of the highlighted group. |
| Alt+P, Insert | Pin or unpin the highlighted flow. |
| Esc | Close the list. |
On a phone. Below 600 pixels wide the list opens as a sheet from the bottom of the screen. The Group control and the Active button are left out, and a touch on the star of a row pins the flow. The sheet closes with Esc, a touch outside it or a flow choice.
10 Endpoints
The Endpoints page shows every IP address seen on the monitored network as a source (it sends) or a destination (it receives), and which source feeds which destination. While the Flows page lists each flow, the Endpoints page adds up the flows per address, so you can see at a glance what a device sends, where it goes, and whether something is wrong.
An address that both sends and receives appears in both lists.
10.1 Group Tabs and View Options
- Group tabs - All, one tab per group from Settings > Groups, and Ungrouped. Each tab shows how many sources and destinations it holds. The count turns amber with a warning icon when an address in that group has a problem, so you can see where to look without opening every tab.
- Group by Input - Adds a section row per network input ("Input Backup - 18 destinations - 4.18 Gbps") instead of repeating the input on every row.
- Layout - The Layout switch: Table shows the sources as a list and the destinations as a table; Cards shows compact cards. The choice is saved in your own preferences, separate from the Flows and Devices pages; a new user starts with Table.
- Find name or address - Narrows both lists on this page to the addresses whose name or address contains the text. It works on top of the filter bar at the top of the window. Esc in the box clears it.
10.2 States
Every address has a state, shown as a color, an icon and a word:
| State | Meaning |
|---|---|
| Error | A flow of this address fails a check (red). |
| Warning | A flow is in warning, or its CC errors or RTP gaps went up in the last minute. The word names the cause, for example CC errors (amber). |
| No traffic | None of its flows has carried packets lately (gray). The bitrate drops to 0 and the sparkline falls to zero. |
| OK | Traffic and no problems (green). |
The lists are sorted by state by default: errors first, then warnings, then no traffic, then OK, and within a state by name.
10.3 Table Layout
The Sources list shows per source its name, address, state, bitrate with a sparkline of about the last 40 seconds, packets per second, active and total flows, and how many destinations it sends to. The sort button in its header sorts by state, name, address, bitrate or type.
The Destinations table has these columns:
| Column | Description | Example |
|---|---|---|
| State | Color, icon and word, see 10.2. | CC errors |
| Name | The alias (if configured) or the address. | TS 17 |
| Type | The kind of stream: 2110-20, 2110-30, 2110-40, MPTS, PTP, or the protocol. | MPTS |
| Address | The IP address. | 239.120.125.1 |
| From | The source that sends the most to it, with +N when more sources send to it. | Prod 124 +1 |
| Bitrate | Total bitrate received, with a sparkline. | 44.1 Mbps |
| ETR 290 | MPTS destinations only, with the MPTS option: the P1, P2 and P3 lamps of the stream, see 10.8. | P1 P2 P3 |
| Packets/s | Total packet rate. | 4,027 |
| Flows | Active flows / all flows. | 2/2 |
| Group | The first group the address belongs to. | MPTS |
Click a column header to sort by it; click it again to reverse the order. When the window is too narrow, columns drop out (Group, From and Type first). The column button at the right of the header shows or hides any column.
10.4 Selecting a Source or a Destination
- Click a source to select it. Its details open on the right and the destination table shows only the destinations this source sends to. A chip From: name in the view bar shows the filter; the x on the chip shows all destinations again and keeps the details open. Another source that feeds the same destinations (for example the backup of a main / backup pair) says so: "Also sends to these 8". The filter works together with the group tabs, the find box and the filter bar.
- Click a destination to open its details. The table is not filtered; the sources that send to it are marked ("Sends to TS 17") and the others are dimmed.
- Click the selected source again, or press Esc, to clear the selection and the filter.
- The selection is part of the page address (
?ip=and?from=), so a bookmark, a shared link or a reload brings it back. - With the keyboard: Tab to a row, Enter opens it, Esc clears.
- On a narrower window, while the details are open, the source list becomes a row of chips above the table so the table keeps its room.
10.5 Detail Panel
The panel shows, from top to bottom:
- Header - name, address, state, Source or Destination, the stream types and the address range. The buttons open the flows of this address on the Flows page, and close the panel.
- Summary - bitrate with a sparkline, packets per second, flows (active / all) and the number of destinations or sources.
- Sends to (source) or Received from (destination) - each address on the other side with the bitrate between the two. The bar shows which share of the destination's bitrate comes from this source. A problem on one of those flows is shown on that line (for example "CC errors on this flow"), so the cause is visible at once. Click a line to open that address.
- MPTS - transport stream ID, programs, PIDs, null packet share and the CC errors with the PID that has the most, with a link to the MPTS page. PTP - domain, delay mechanism, clock class and grandmaster, with a link to the PTP page.
- Counters - RTP gaps, CC errors and flows without traffic, counted since the probe started.
- Flows - the flows of this address with their bitrate; click one to open it on the Flows page.
- Details - input, MAC address and manufacturer, VLAN, DSCP / TTL, groups and first seen.
A list of the incidents of one address with their times is not available yet; the panel shows the counters from the flows instead.
10.6 Cards Layout
Cards show the same information in a compact grid: the type badge and the state on top, the name and address, the bitrate large with a sparkline, and at the bottom the flows and "Sends to N" (source) or packets per second, flows and the sending source (destination). With a source selected, each destination card shows how much it receives from that source and how many other sources feed it. Use Sort to change the order.
10.7 Empty Group and Phone
- A group tab without any traffic shows "No traffic in" and the group name, with Show all endpoints and Edit group. For a group that should stay empty (for example forbidden multicast addresses) this is the good state.
- On a phone the sources are a row of chips that scrolls sideways; tap one to select it. The destinations are a compact list, and the details open as a sheet at the bottom: tap Show all details to see the rest.
IP aliases and descriptions can be configured in the Settings > Aliases section. See chapter 31 for details.
10.8 ETR 290 on the Endpoints Page
With the MPTS option, the page shows the ETR 290 (TR 101 290) result next to the MPTS streams:
- Lamps - An MPTS destination shows the P1, P2 and P3 lamps, as on the MPTS page: a red lamp with an exclamation mark is in error now, an amber outline lamp with a clock was in error within the hold time, a green lamp is OK, and a gray lamp is not checked. The lamps are in the ETR 290 column of the table, on the cards, in the phone list and in the details panel. Point at them to see the worst check, for example "Priority 1: error (1.4 Continuity_count_error)". Click them, or press Enter, to open that stream on the ETR 290 page. The details panel also has an Open in ETR 290 link.
- State - The priority of the check decides, not the severity you set in the ETR 290 settings. A Priority 1 error makes the address an Error ("ETR 290 P1"), a Priority 2 error a Warning ("ETR 290 P2"). A Priority 3 error and a past error only show on the lamps and do not change the state. The severity per check only sets the color and the alarm level on the ETR 290 pages, in the episodes and in the alarm messages.
- Source badge - A source that sends MPTS streams in ETR 290 error shows a badge such as ETR 290: 2 P1 ERROR: the number of its streams in error and the worst priority. When the streams only were in error, the badge says WAS. Click it to open the ETR 290 overview for that source.
11 Devices
The Devices page provides a hardware-level view of all Ethernet (MAC) addresses observed on the monitored network. This page is useful for device identification, vendor analysis, and detecting unexpected or unauthorized devices.
Each MAC address appears as either a source endpoint, a destination endpoint, or both.
11.1 View Options
The view options bar provides the same controls as the Endpoints page:
- Group by Input - Groups MAC endpoints by the network input on which they were observed.
- Layout - The same Layout switch: Grid (cards) or List (compact rows).
11.2 MAC Card
Each MAC endpoint is displayed as a card (or list row) with the following information:
| Element | Description | Example |
|---|---|---|
| Direction icon | Upload icon for source MACs; download icon for destination MACs. | Upload |
| Name | The alias (if configured) or the formatted MAC address. | Switch-Core-Eth0 |
| Address | The MAC address in colon-separated hex notation. | 00:1a:2b:3c:4d:5e |
| Activity dot | Green when at least one associated flow is active. | Active |
| Severity border | A colored left border reflecting the worst severity across associated flows. | Green |
The card body shows:
| Metric | Description | Example |
|---|---|---|
| Vendor | The hardware manufacturer derived from the OUI (first 24 bits of the MAC address). Resolved automatically from the IEEE OUI database. | Cisco Systems |
| Usage | A brief description of the MAC address's typical role (e.g. STP, LLDP). Shown when applicable. | Spanning Tree Protocol |
| Bitrate | The total bitrate across all flows involving this MAC endpoint. | 1.49 Gbps |
| Pkts/s | The total packet rate. | 129,594 pps |
| Flows | Active versus total flow count (active / total). | 2/3 |
11.3 MAC Detail Drawer
Clicking a MAC card opens a detail drawer showing:
- All flows associated with this MAC address
- The full MAC address, alias, and description
- Manufacturer information (OUI lookup)
- Aggregate traffic statistics
11.4 Identifying Orphan Flows
The Devices page is particularly useful for identifying orphan flows - traffic from unknown or unexpected MAC addresses. Orphan flows may indicate:
- Unauthorized devices connected to the network
- Misconfigured switches or routers
- MAC address spoofing
- Network loops generating traffic with unexpected source MACs
Sort or filter by vendor to quickly spot MAC addresses from unexpected manufacturers.
MAC aliases and descriptions can be configured in the Settings > Aliases section.
12 Network Map
The Network Map page provides a visual representation of the relationships between network endpoints and the traffic flowing between them. It contains two tabs: Topology and Flow Map.
12.1 Topology
The Topology tab displays an interactive network diagram showing the discovered devices (IP and MAC endpoints) and their interconnections. Nodes represent endpoints; edges represent observed traffic between them.
The diagram is rendered using D3.js force-directed layout, which automatically arranges nodes to minimize edge crossings and overlap. Users can:
- Drag nodes to manually reposition them for clarity.
- Zoom and pan using mouse scroll and drag on the background.
- Hover over a node to highlight all its connections and display a tooltip with endpoint details (address, alias, bitrate).
- Click a node to select it and view its associated flows.
The topology view is especially valuable for understanding the overall structure of a broadcast IP network - which encoders feed which decoders, how multicast groups are distributed, and where redundant paths exist.
12.2 Flow Map
The Flow Map tab (the former Traffic page) displays a Sankey diagram showing traffic flows between source and destination endpoints. The width of each band is proportional to the flow's bitrate, providing an intuitive sense of bandwidth distribution.
Source endpoints are arranged on the left side of the diagram; destination endpoints are arranged on the right. Flows connect the two sides, with band width representing data rate.
The Flow Map is useful for:
- Identifying the highest-bandwidth connections at a glance
- Verifying that expected multicast groups are being received by the correct destinations
- Detecting unexpected traffic patterns or imbalanced load distribution
Both the Topology and Flow Map views update in real-time as new flows are detected or existing flows change status.
13 PTP
The PTP (Precision Time Protocol) page analyzes the IEEE 1588 PTPv2 traffic on the monitored network. PTP is the timing backbone of SMPTE 2110 networks: when the clock is wrong, every media stream that follows it is wrong too. The page shows one view per clock, that is one domain with one grandmaster, with all inputs together, and says at a glance whether that clock is good. Every value is shown once.
This page requires the 2110 license option.
From top to bottom the page has:
- Toolbar - the title with the number of clocks and PTP flows, and the ST 2110-10 timing button.
- Clock tabs - one tab per clock, and an Overview tab.
- Header - the grandmaster, its live time, its state and its warnings.
- Panels - Grandmaster, Sync, Probe and path, Messages, ST 2059-2 profile, Grandmaster changes, Flows, and a Masters table when several masters compete.
- Sync over time - three charts of the Sync interval, the jitter and the correction field.
- Message log - the latest PTP messages of the chosen master.
PTP analysis needs PTP packets (UDP ports 319 and 320) on at least one monitored input. The filter bar applies to this page, as on the other source pages. When no PTP flow is shown, the page tells how many PTP flows the filter bar hides and offers Show all.
If no PTP flows appear at all, check that the network switch forwards PTP traffic to the monitoring port.
13.1 Clock Tabs and Overview
A clock is a PTP domain together with the grandmaster that serves it. A grandmaster that reaches the probe on two inputs (a red and a blue network) is still one clock. Each tab shows a state dot, the domain, the time scale (TAI or ARB), the grandmaster brand and identity, and a short line such as "Input 1, 2 - locked to GNSS". The tabs are sorted with the worst clock first, and the page opens on the clock you used last, or on the worst one.
- When the tabs do not fit, the rest goes into a +n more menu.
- On a phone the tabs are replaced by a Clock picker field with the same information.
- The choice is kept in the address (
/ptp?clock=...), so a link opens the same clock.
The Overview tab lists every clock on every input in one table, alarms first: state, domain, grandmaster, time scale, clock class, time source, inputs, number of masters and slaves, and the Sync rate. A row opens its clock. On a phone each row is a small card.
A domain that only has slaves (devices sending Delay_Req and no master) is listed as "no master seen".
13.2 The Header
The header belongs to the clock you picked.
- Domain and state tag - One word that sums up the clock: OK, WARNING (or "2 WARNINGS"), ALARM, NO MASTER or STARTING (no Announce received yet). Sequence gaps and a grandmaster that is not traceable are information only and do not change the state. The same problem on several inputs of one grandmaster counts once.
- Identity - The 8-byte clock identity of the grandmaster and its brand (from the manufacturer part of the identity). Under it: the inputs, the delay mechanism (E2E or P2P), one-step or two-step, and the number of masters and slaves. Behind a boundary clock the master port is shown as well.
- Master - When the grandmaster reaches the probe through more than one master (for example on input 1 and on input 2), choose the master here. The choice applies to the Sync panel, the charts, the messages and the log.
- Live time - The time the grandmaster sends, ticking live with milliseconds. The tag LIVE shows that Sync messages arrive. TAI is the PTP time scale. A second, smaller time shows UTC, but only for the PTP time scale (UTC is TAI minus the UTC offset the grandmaster announces). With the ARB time scale (arbitrary, for example a free-running grandmaster at a date in 1970) there is no UTC time, because the time has no relation to UTC.
- NO SYNC - When a master stops sending Sync messages, the time freezes and turns gray, the tag changes to NO SYNC n S and an alarm banner appears. The time does not run on by itself.
- Incidents row - Counts since the probe started: GM changes, Sync losses, Sequence gaps, Several GMs and High correction. A count above zero is a link: it opens the Logs page with the lines of that incident for this master.
- Banners - One banner per problem, worst first (red alarm, amber warning, blue information). The same problem on several inputs of one grandmaster gives one banner that names the inputs.
| Banner | Meaning |
|---|---|
| BMCA conflict (alarm) | Two or more masters announce at the same time on one domain and input. It names the best master, the field that decided (priority 1, class, accuracy, variance, priority 2 or identity) and the master that still sends Announce. Two boundary clocks that announce the same grandmaster count as a conflict too. Devices may disagree on the time, which breaks media synchronization. |
| No Sync (alarm) | The master stopped sending Sync messages. |
| Free-running grandmaster / ARB timescale (warning) | The grandmaster has no time reference or sends arbitrary time. Media can still lock to it, but the offsets against it are as far off as its time. |
| Clock class (warning) | The class is not 6 (locked), for example 7 (holdover) or 248 (no time reference). |
| Sync or Announce rate differs from the declared one (warning) | For example "sends Sync every 125 ms but declares 250 ms" (more than 10 % apart). |
| No Announce (warning) | No Announce for longer than expected; the grandmaster properties may be out of date. |
| Grandmaster changes (warning) | One or more grandmaster changes in the last 15 minutes. |
| Not traceable, Sequence gaps (information) | The time is not traceable to a primary reference; sequence numbers were missed since the probe started. |
13.3 Grandmaster, Sync, and Probe and Path
Three panels sit side by side on a wide screen (1150 px and wider), in two columns on a laptop, and under each other on a phone. Each value appears in one place only, and the values start at the same offset in every panel.
Grandmaster - Everything the grandmaster announces:
| Field | Description |
|---|---|
| Clock class | The ITU-T class with its meaning (6 locked to a primary reference, 7 holdover, 248 default, free-running). |
| Accuracy, Variance | The declared accuracy (for example within 100 ns) and the offset scaled log variance. |
| Time source | GNSS, NTP, atomic clock, internal oscillator and so on. |
| Traceable | Whether time and frequency are traceable to a primary reference. |
| Timescale | PTP (TAI) or ARB. |
| UTC offset | Seconds between TAI and UTC, with VALID or NOT VALID and a note on a leap second. |
| Priority 1 / 2, Steps removed | The BMCA priorities; 0 steps means the master is the grandmaster. |
| Announce | The measured Announce interval against the declared one, the age of the last Announce (it turns amber when overdue) and the sequence number with its gaps. |
The subtitle names where the data comes from (for example "from Announce, 172.16.0.10 on Input 1").
Sync - How regular the Sync messages are:
| Field | Description |
|---|---|
| Interval | The measured mean interval between Sync messages. |
| Declared | The interval the master declares (logMessageInterval) with AS DECLARED or MISMATCH - SENT EVERY n MS. |
| Jitter | The peak deviation from the mean interval, per second. A few microseconds to tens of microseconds is normal. |
| Std dev, Min / max, Samples | The spread of the interval and the number of Sync messages analyzed. |
| Mode | One-step or two-step (Follow_Up) and the Follow_Up rate. |
| Sequence | The last sequence number, the gaps (lost messages, CLEAN when none) and duplicates. |
| Correction field | The latest, mean and maximum correction field of the Sync messages. A correction above zero means a transparent clock in the path; only values above 1 us raise the High correction warning, which does not turn the clock red. |
Probe and path - What the probe itself measures:
| Field | Description |
|---|---|
| Receive time | HARDWARE or SOFTWARE timestamps on the network card. Shown here only. |
| Probe clock fit | Whether the probe's own clock follows the grandmaster: LOCKED, drifting, acquiring or stale, with the drift in ppm and its uncertainty. |
| Probe - GM | How far the probe clock (set by NTP or PTP) is from the grandmaster time, with its range. Normally well below a few milliseconds. For an ARB grandmaster it is the distance to a date in 1970, which is meaningless. |
| Path delay | The end-to-end path delay measured from Delay_Req and Delay_Resp (not shown when no pair is seen). |
| Media clocks on this grandmaster | The number of ST 2110 flows locked to this grandmaster, and a link to the ST 2110-10 page. |
13.4 Masters and BMCA
The Masters announcing on domain n, Input m table appears only when more than one master announces on the same domain and input. It lists the masters in BMCA order (best first): the rank (BEST, 2ND), grandmaster and brand, the IP address it sends from, priority 1, class, accuracy, variance, priority 2, steps removed and the Announce rate. The value that decided the election is marked red, and a line below the table names it ("Decided by class: 6 before 248"). Use it to find the device that should not be a master, and change its priority or switch it to slave only.
On a phone every master is a small card with labeled values instead of a wide table.
13.5 Messages and the 10 Percent Rule
The Messages panel lists each message type (Sync, Follow_Up, Announce, Delay_Req, Delay_Resp) with its count, its measured rate, its declared rate and a status. One rule applies to Sync, Announce and Delay_Resp: a rate more than 10 % away from the declared rate is marked (for example "2x declared"). Delay_Req has no declared rate. Types that are not present (Pdelay, Signaling, Management) are named in one line below the table. Statuses carry a word as well as a color: OK, a warning, missing or inactive.
13.6 ST 2059-2 Profile
The ST 2059-2 profile panel checks the clock against the SMPTE ST 2059-2 default profile, one line per item with OK, Check or Fail:
- Domain (0 to 127, with 127 the profile default),
- Sync interval (125 ms to 0.5 s) and Announce interval (125 ms to 2 s), measured and declared,
- Timescale (PTP, not ARB),
- Clock class (6 is good, 7 holdover, 248 and others "check"),
- Delay mechanism (E2E is the default; P2P is "check").
Fail means the clock is outside the profile; Check means worth a look. The profile is shown for information and does not set the clock state.
13.7 Grandmaster Changes
The Grandmaster changes panel records every time the grandmaster of the domain changed: time (UTC), old and new grandmaster with their clock class, and the input. A stable network shows "No grandmaster change since the probe started". Frequent changes mean an unstable election, which disturbs all devices on the domain. When changes happened on one input only, a line names the inputs without a change. Changes are also drawn as red markers in the charts. On a phone each change is a small card.
13.8 Flows of This Domain
The Flows panel lists the PTP flows of the clock per input, with the masters first and their slaves under them: state dot, role (Master event, Master general, Slave Delay_Req), source address and brand, messages, rate, and on a wide page the port and the bitrate (these two columns are left out below 1150 px). The arrow at the end of a row opens the flow detail. A gray row is a flow without traffic.
13.9 Sync Over Time
Three charts, each with one axis, show the history of the chosen master: the Sync interval (with the declared interval as a dashed line and a band of 10 % around it), the Jitter around the mean (peak per second) and the Correction field (maximum per second). They stand next to each other on a wide page and under each other on a phone.
- Windows - Choose 1 min, 10 min or 60 min.
- Markers - A red line marks a grandmaster change, a dashed red line a Sync loss. Move the pointer over a chart to see the time and the values.
- CSV - The download button saves the samples of the shown window as a CSV file.
- History - One sample per second is kept in the browser, per master, but only while the PTP page is open: when you leave the page no samples are taken, so the charts show a gap for that time. At most the last hour is kept per master, for up to 64 masters. The history survives switching tabs and masters and is lost when the page is reloaded.
13.10 Message Log
The Message log shows the newest messages of the chosen master and of its slaves on the same input, merged, newest first (the last 50 per flow). Columns: arrival time on the probe clock (UTC, with microseconds), type, sequence number (with a GAP badge when numbers are missing), source (master or slave and address), correction field and a short content text.
- Filter - Chips for All, Sync, Follow_Up, Announce, Delay_Req and Delay_Resp. On a phone a type menu replaces the chips.
- Pause - Freezes the list so you can read it; the button then shows how many new messages arrived and becomes Resume.
- Row details - Click a row (or press Enter) to open the decoded fields: flags, source port identity, timestamps and so on.
13.11 The Phone Layout
Below about 640 px the page is built for one hand:
- A Clock picker replaces the tabs and the header stacks: identity, master choice, time card, incidents and banners.
- The panels are folded sections. Each shows its title and a one-line summary (for example Sync: "125.002 ms, +/- 3.8 us, 0 gaps"). Tap a header to open it. Grandmaster is open at first; on a conflict clock the Masters table is open too.
- Tables become cards, so nothing needs sideways scrolling; the message log keeps four columns (time, type, sequence, source).
13.12 ST 2110-10 Links
Older links to a grandmaster from the ST 2110-10 page (/ptp#...) still work: they open the clock tab and the master of that grandmaster. The ST 2110-10 timing button in the toolbar opens the page described below.
13.13 ST 2110-10 System Timing
The ST 2110-10 page (navigation: Sources, directly under PTP) shows the ST 2110 system as a whole: which devices send, whether all their flows run on the PTP grandmaster's clock, and whether the flows follow the network rules of SMPTE ST 2110-10. The filter bar applies, as on the other source pages.
Timing. Tiles at the top count the media clocks of all ST 2110-20, -22, -30, -40 and AES67 flows (locked, drifting, acquiring, stale, not known yet), the sender devices and how many of them run on another clock, and the grandmaster and domain most flows use, with the number of flows on another grandmaster or domain. A table lists per input the grandmaster(s) and domain(s) seen, how many flows use each, the probe's own receive clock against the grandmaster, the number of PTP masters and the IGMP versions seen (ST 2110-10 receivers join with IGMPv3; IGMPv1 or v2 shows as a warning).
"Probe receive clock: drifting" means the probe's own clock does not follow the grandmaster (for example a probe without phc2sys). The media clocks are still measured against the grandmaster, so this is not a fault of the senders.
Senders. One panel per sender device. Flows are grouped by source IP; source IPs that share a MAC address are one device, unless that MAC belongs to a router (many source IPs behind it). Each flow shows its type, format, name (a link to the flow detail), media clock state, drift against the grandmaster, offset and reference (PTP or the probe receive clock). The panel title starts with the brand of the sender ("Nevion - 10.10.4.11"), the manufacturer of the source MAC address from the IEEE OUI list; it is left out when the MAC belongs to a router, as a router shows its own brand for every sender behind it.
| Mark | Meaning |
|---|---|
| differs (red) on a flow | The flow runs on another clock than the other flows of the same device: lip sync between them is not possible. |
| other clock than the other senders (red) on a device | The device runs on another clock than most devices: free-running, on another grandmaster, or on another epoch (an offset more than 50 ms away). |
| mixed packet sizes (amber) | The device's video flows use both standard and jumbo packets. |
The offset is only compared when it is below 1 second. A larger offset usually means the sender adds an RTP offset (signaled in its SDP); it is shown in gray with "RTP offset?" and is informational. The play button on a device with video opens it in the AV Monitor; the PTP button in the toolbar opens the PTP page.
Network rules. One row per flow: the smallest and largest UDP payload, that is the RTP header and media without the 8-byte UDP header (standard up to 1460 bytes, the ST 2110-10 Standard UDP Size Limit; jumbo up to 8960 bytes, the Extended UDP Size Limit; larger is an error; a flow within 8 bytes of a limit is borderline, as the probe does not know whether the limit counts the UDP header), size class (standard or jumbo), destination, multicast or unicast, the DSCP marking of the newest packet (not marked is a warning), the RTP payload type (ST 2110 uses 96-127) and the RTP sequence gaps and time jumps since the flow started.
SDP and NMOS conformance. A placeholder for now: a later version compares each flow with its SDP file (format, clock rate, packet size, multicast, RTP offset, ts-refclk) and shows the NMOS labels and groups.
13.14 Why PTP Matters in Broadcast
In an IP-based broadcast facility, every device (cameras, switchers, routers, audio consoles, graphics engines) must share a common sense of time. Without it, video frames from different sources cannot be combined without visible artifacts, audio drifts out of lip-sync, and ancillary data is misaligned to its essence.
PTP (IEEE 1588-2019 Precision Time Protocol) provides this common timebase. A grandmaster clock distributes time across the network, and every endpoint locks its internal clock to that reference. In SMPTE ST 2110 environments PTP is mandatory: ST 2110-10 defines how the media clocks for video (ST 2110-20), audio (ST 2110-30) and ancillary data (ST 2110-40) are derived from the PTP time.
When PTP works, it is invisible. When it breaks, the consequences are immediate:
| Problem | Visible symptom |
|---|---|
| Grandmaster disappears | All clocks free-run; frames go out of sync within seconds. |
| Multiple grandmasters | Devices fight over the time reference; random glitches. |
| High sync offset | Video tearing, audio clicks, dropped frames. |
| Sync message loss | Slave clocks drift until the next Sync arrives. |
| High correction field | A transparent or boundary clock adds a lot of latency. |
These conditions can only be found by watching the PTP traffic itself, which is what the page does. The probe captures the PTP messages on the Event port (UDP 319: Sync, Delay_Req and Pdelay messages) and the General port (UDP 320: Announce, Follow_Up, Management and Signaling messages) without adding any load to the network or touching the timing chain.
13.15 Typical Fault Scenarios
No grandmaster. The page shows the state NO MASTER, or STARTING while no Announce has been received yet. This happens when the grandmaster does not send Announce, or when UDP port 320 is filtered between the grandmaster and the probe. Check that this traffic reaches the monitoring port (section 13.8 lists the PTP flows that do arrive).
Several grandmasters. The BMCA conflict banner appears and the Grandmaster changes panel (13.7) shows the grandmaster switching between two or more identities. The BMCA does not converge: usually two devices have the same priority 1, or the Announce messages of one grandmaster do not reach all devices in time. Compare the Masters table (13.4) and set a clear priority order.
Sync interval differs from the declared one. The Sync panel shows MISMATCH. Common causes are an overloaded grandmaster that delays its messages, a transparent clock that queues the messages, or a network device that rate-limits UDP port 319.
High correction field. A transparent or boundary clock adds a residence-time correction. A stable value is normal, a value that keeps growing means that the clock of the intermediate device drifts and its measurement collects error. The High correction warning appears above 1 us.
Sync gaps on an otherwise healthy clock. Short bursts of missing Sync messages that recover by themselves usually come from congestion that drops packets on the way to the probe, or from a short CPU overload of the grandmaster. Persistent gaps are a serious fault. In both cases the slave clocks are in holdover for the length of the gap.
14 ST 2110-20
The ST 2110-20 page provides real-time analysis of SMPTE ST 2110-20 uncompressed video streams. It is the primary tool for verifying that a video-over-IP flow conforms to the standard and is arriving correctly at the network level.
The page follows the Source page layout pattern:
- Flow picker - Chooses one of the detected ST 2110-20 flows.
- Detail panel - Full analysis of the currently selected video flow.
This page requires the 2110 license option. The picture parts also need the Video option.
ST 2110-22 flows (JPEG XS video, RFC 9134) are listed on this page as well, with a JPEG XS tag in the flow picker. Their detail panel is described in section 14.12, JPEG XS (ST 2110-22) Flows.
14.1 Flow Picker
Choose the flow with the flow picker in the toolbar. Click it to open the list: search by name or address, filter, group and sort the flows, and pick one. Every row shows a health dot, the flow name and its bitrate; the counts of active and inactive flows are in the picker. The chosen flow is in the page address (?flow=<id>) and the browser remembers it, so a link opens the same flow again. The arrows next to the picker step to the previous or next flow (the picker is described in 9.6). For video flows the rows also show the video thumbnail (when available), the resolution and frame rate (for example 1920x1080p30) and the number of sequence gaps.
14.2 What to Monitor in an ST 2110-20 Stream
Carrying uncompressed video over IP introduces several failure modes that do not exist in traditional SDI infrastructure. Because there is no forward-error correction (FEC) in a basic ST 2110-20 deployment, a single lost or reordered packet is visible in the output picture as a corrupted line or flash. The following areas are therefore critical to monitor continuously.
Stream Identity Before analyzing quality metrics, confirm that the stream is carrying the expected format. A misconfigured encoder or a wrong multicast subscription can silently deliver the wrong resolution, frame rate, or pixel format. IP-Probe reads the SRD (Sample Row Data) headers in every packet to derive the actual format and displays it without relying on any out-of-band signaling.
Packet Loss and Sequence Integrity RTP sequence numbers advance by one for every packet. A gap in the sequence means one or more packets were lost in transit. In an uncompressed video context this maps directly to missing lines in the picture. Duplicate RTP sequence numbers indicate that the same packet arrived twice, which can occur with SMPTE 2022-7 protection switching or from network loops. Both conditions must be zero in a healthy stream.
Frame Completeness ST 2110-20 uses the RTP marker bit to signal the last packet of every video frame. If the marker bit fires at the wrong time - or not at all - the receiver cannot reconstruct the frame boundary correctly. Missing frames are counted when the expected marker bit does not appear within the expected interval.
Timing and Receiver Buffer (VRX) ST 2110-20 senders are governed by ST 2110-21, which defines how packets must be paced in time. The standard defines a virtual receiver model with a buffer of N packets. If packets arrive too early or too late they exceed the buffer limits and cause a compliance violation. This is the most common source of interoperability problems between different vendor implementations.
- TRO (Transmission Reference Offset) measures how far in time each packet deviates from its ideal scheduled arrival. A tight, symmetric distribution around zero is expected. A systematic offset indicates a misconfigured sender. A wide distribution indicates jitter introduced by the network.
- VRX (Virtual Receive Buffer) models the receiver buffer fill level defined in ST 2110-21. If the peak VRX value exceeds the N_MAX limit for the detected sender category, the stream is marked non-compliant. This will cause picture problems on strict receivers.
- PIT (Packet Inter-arrival Time) measures the time between consecutive packets at the monitoring point. High PIT values reveal momentary bursts or congestion events that may not yet show up as lost packets.
Frame Interval Stability The time between consecutive frame boundaries (marker bits) should be constant and equal to one frame period (e.g. 33.333 ms for 30 fps). Deviation from this value - frame jitter - indicates instability in the sender's clock or in the PTP synchronization it relies on.
14.3 Stream Identity Panel
The Stream Identity panel confirms what format is actually flowing on the wire, independently of any configuration or SDP (Session Description Protocol) metadata.
| Field | What it tells you |
|---|---|
| Resolution | The active picture width x height derived from SRD row/offset data. Verify this matches the expected format. |
| Frame Rate | The frame rate type detected from the RTP timestamp cadence. |
| Color Sampling | The chroma sampling scheme (e.g. YCbCr 4:2:2) derived from the pgroup geometry. |
| Bit Depth | Bits per sample, typically 10-bit for broadcast. |
| PGroup | The pgroup size (bytes) and coverage (pixels) define the pixel packing. Verify these match the agreed format. |
| Packing Mode | Type N (Narrow) or Type W / NL. Narrow is the most common and switch-friendly mode. |
| Sender Type | Narrow, Linear, or Wide - the ST 2110-21 traffic-shaping category declared by the sender. This determines the VRX compliance limit. |
| Packets / Frame | The number of RTP packets that make up one complete video frame. A 1080p59.94 10-bit stream typically uses around 4320 packets per frame. |
| SRD Length | Minimum and maximum SRD payload sizes seen. A wide spread can indicate a non-standard packetization. |
IP-Probe derives all identity fields directly from the packet stream. No SDP or NMOS registration is required.
14.4 Sequence Analysis Panel
The Sequence Analysis panel tracks RTP and extended sequence number continuity. Any non-zero error count here means data is missing from the stream.
| Field | What it tells you |
|---|---|
| Total Packets | Total RTP packets received since the last reset. Useful for confirming the stream is active. |
| RTP Seq Gaps | Count of jumps in the 16-bit RTP sequence number. Each gap represents lost packets. Should be zero. |
| RTP Duplicates | Count of repeated RTP sequence numbers. A small number may be expected with 2022-7 seamless protection switching; a large number may indicate a loop. |
| Ext Seq Gaps | Gaps in the 32-bit extended sequence number (if used by the sender). Provides a larger sequence space for long-running streams. Should be zero. |
| Last RTP Seq | The most recently received 16-bit RTP sequence number (decimal and hex). |
| Last Ext Seq | The most recently received 32-bit extended sequence number (decimal and hex). |
Frame Integrity (shown within this panel when frame data is available):
| Field | What it tells you |
|---|---|
| Total Frames | Number of complete video frames received. |
| Missing Frames | Frames for which the marker bit was never received. Should be zero. |
| Marker Bit | OK / ERROR - indicates whether the marker bit pattern matches the expected frame rate. A persistent error means the sender is not correctly signaling frame boundaries. |
14.5 Video Preview
The Video Preview shows a decoded thumbnail of the actual picture content. Thumbnails are updated approximately once per second.
The preview confirms:
- The content is visually correct (not frozen, not corrupted, not black)
- The correct source is being received on the expected multicast address
- Gross encoding or routing errors are immediately visible without needing SDI monitoring equipment
Thumbnails are decoded in software, within a decode budget. The preview does not affect the accuracy of any other measurements.
14.6 Timing and VRX Compliance Panel
This is the most technically detailed panel and is central to ST 2110-21 compliance verification.
14.6.1 TRO Distribution
The TRO (Transmission Reference Offset) histogram plots how far in time each packet arrived from its ideal scheduled position, binned at 200 ns resolution. The x-axis is in microseconds; the y-axis is packet count on a logarithmic scale.
What a healthy histogram looks like: A tight, roughly symmetric distribution centered near zero. The vast majority of packets should fall within +/-1 ms. A long tail to the right indicates network delay or queuing. A systematic shift to one side indicates a sender clock offset.
TRO Statistics:
| Field | What it tells you |
|---|---|
| Min / Max | The extreme TRO values seen. A large asymmetry between min and max indicates the stream is not well-centered in the receiver buffer. |
| Mean | The average TRO. A non-zero mean indicates the sender is transmitting ahead of or behind the ideal schedule. |
| Std Dev | The spread of packet timing. A high standard deviation indicates network jitter that could exceed the VRX buffer. |
14.6.2 VRX Compliance
The VRX (Virtual Receive Buffer) model simulates the behavior of a receiver buffer as defined in ST 2110-21. The buffer fill level rises when packets arrive early and drains as the receiver consumes them.
| Field | What it tells you |
|---|---|
| Status | PASS if the peak buffer fill has not exceeded the N_MAX limit; FAIL if it has. A failing stream may cause picture artefacts on strict receivers. |
| VRX Peak | The maximum buffer fill level observed. |
| Limit (est.) | IP-Probe estimates N_MAX from the detected sender type and vertical line count. For a Narrow sender at 1080 lines, N_MAX is approximately 864. |
| VRX Fill Bar | Visual representation of how close the stream is running to the buffer limit. Green = well within limit; Orange = approaching limit; Red = exceeding limit. |
When VRX fails: First check whether the sender is correctly classified as Narrow or Wide - a Wide sender has a higher N_MAX limit and a stream classified as Narrow may pass when reclassified. If the sender category is correct, the root cause is typically either excessive network jitter or a sender that does not correctly implement ST 2110-21 pacing.
14.6.3 PIT Statistics
PIT (Packet Inter-arrival Time) is the elapsed time between consecutive packet arrivals at the probe. Unlike TRO (which is measured against the ideal schedule), PIT measures what the network actually delivered.
| Field | What it tells you |
|---|---|
| Min | The shortest gap between any two consecutive packets. Very small values indicate micro-bursts. |
| Max | The longest gap. A large Max PIT can indicate transient congestion - the sender queued packets while waiting for bandwidth. |
| Mean | The average inter-arrival time. For a 1080p60 stream with approximately 4320 packets per frame, the mean PIT should be approximately 16.7 ms / 4320 = 3.9 us. |
| Std Dev | The jitter. A high standard deviation relative to the mean indicates unpredictable delivery, which stresses receiver buffers. |
14.6.4 Frame Interval
Measures the time between consecutive video frame boundaries (RTP marker bits).
| Field | What it tells you |
|---|---|
| Current | The most recently measured frame interval in milliseconds. |
| Min / Max | The extreme frame intervals seen. Any variation beyond +/-1 us for a locked sender may indicate a PTP synchronization problem. |
| Jitter | Peak-to-peak variation in frame interval (us). Should be below 500 us for a well-locked sender. Values above 2 ms indicate a serious timing problem. |
14.7 Frame Analysis Panel
The Frame Analysis panel provides packet-level statistics aggregated per frame.
| Field | What it tells you |
|---|---|
| Total Frames | Total complete frames received. |
| Packets / Frame | Current, minimum, and maximum packet counts per frame. The current value should match the expected packets-per-frame for the format. Variation in min/max indicates dropped or extra packets per frame. |
| Frame Interval | Confirms the frame period at this level, redundant with the Timing panel. |
| Frame Jitter | Peak-to-peak frame interval deviation. |
| Marker Bit | OK or ERROR - whether marker bits are arriving at the correct periodicity. |
SRD Header Distribution: Each RTP packet in ST 2110-20 carries one, two, or three SRD (Sample Row Data) headers, each describing one contiguous run of pixels. The bar chart shows what fraction of packets carry 1, 2, or 3 SRD headers. For most standard packetizations, nearly all packets carry a single SRD header. A significant proportion of multi-SRD packets is not unusual for wide-packing or non-standard implementations, but a sudden change in this distribution can indicate a sender reconfiguration.
Field Distribution (interlaced streams only): For interlaced formats (1080i, 576i, etc.), shows the packet count for Field 1 and Field 2 separately. The two counts should be equal. An imbalance may indicate that one field is being lost or that the sender is not correctly setting the field identification bit in the SRD header.
14.8 Recent Packets Timeline
The Recent Packets table shows the last 100 RTP packets in arrival order. It provides packet-level detail for debugging.
| Column | Description |
|---|---|
| # | Packet index within the displayed window. |
| HW Delta | Inter-arrival time from the previous packet, measured using hardware timestamps when available. Spikes in this column reveal micro-bursts or gaps. |
| RTP TS | The 90 kHz RTP timestamp. All packets belonging to the same frame carry the same timestamp. |
| Ext Seq | The 32-bit extended sequence number (if present). |
| Row | The SRD first active row - the video line this packet covers. |
| Offset | The SRD first active sample offset within the line. |
| Length | The SRD data length in bytes for this packet. |
| Field | For interlaced streams: 1 or 2. |
| M | A blue M badge marks the last packet of a frame (marker bit set). |
Using this table: Click any row to pause automatic scrolling; click Follow latest to resume. The table is most useful for isolating whether packet loss occurs at a specific row or offset (suggesting a regular pattern from the sender) or randomly (suggesting network congestion).
14.9 Error History Graph
The Error History graph shows the per-second rate of each error type over the last 60 seconds. This allows operators to see whether errors are intermittent or sustained, and to correlate them with events such as a network switch failover or a PTP lock loss.
| Series | Description |
|---|---|
| RTP Gaps | Packets missing from the RTP sequence per second. |
| Ext Seq Gaps | Gaps in the extended sequence number per second. |
| Missing Frames | Frames that did not complete (marker bit never arrived) per second. |
| Marker Errors | Packets with unexpected marker bit values per second. |
Reset Errors: The Reset Errors button in the page header resets all cumulative counters to zero on both the probe and in the graph history. Use this after correcting a fault to obtain a clean baseline.
14.10 Hardware vs. Software Timestamps
IP-Probe indicates whether the current measurements are based on hardware timestamps (captured by the NIC at the moment a packet is received) or software timestamps (recorded by the operating system after the packet has been processed).
| Mode | Badge | Impact on measurements |
|---|---|---|
| HW Timestamps | Green "HW Timestamps" badge | TRO, PIT, and VRX measurements are accurate to within nanoseconds. Recommended for compliance testing. |
| SW Timestamps | Orange "SW Timestamps" badge | Timestamps include OS scheduling jitter, typically +/-10-100 us. Sufficient for coarse fault detection but not for strict ST 2110-21 compliance verification. |
Hardware timestamping requires a network card with hardware timestamp support. Refer to the hardware installation guide for supported cards.
14.11 Typical Fault Scenarios
| Symptom | Likely Cause | Where to look in IP-Probe |
|---|---|---|
| Occasional picture freeze or flash | Packet loss | Sequence Analysis - RTP Seq Gaps > 0 |
| Picture tears at specific lines | Micro-bursts dropping packets on congested links | Recent Packets - spike in HW Delta; PIT - high Max |
| VRX FAIL on a well-known sender | Sender classified as Narrow but actually Wide | Identity - Sender Type |
| Frame rate appears wrong | Sender clock drift or PTP unlock | Timing - Frame Jitter > 2 ms |
| Sporadic missing frames | Marker bit never arrives for some frames | Frame Analysis - Missing Frames counter; Frame Integrity - Marker Bit ERROR |
| Image has wrong format / appears green | Wrong multicast joined or wrong flow selected | Identity - Resolution, Color Sampling |
| All errors suddenly appear together | PTP grandmaster failover / network reconfiguration | Error History - correlated spike across all series |
14.12 JPEG XS (ST 2110-22) Flows
An ST 2110-22 flow carries JPEG XS compressed video. When you select one, the detail panel shows a JPEG XS panel set instead of the ST 2110-20 panels (there is no pgroup, sender type or VRX for compressed video):
- JPEG XS Stream - resolution, profile, level and sublevel, sampling and bit depth, slice height, frame rate, compression ratio, packetization (codestream or slice) and transmission mode, and the declared bitrate of the stream next to the measured one. Below them the flow ID, input, sender and addresses.
- Video Preview - the thumbnail and the live view. The probe decodes JPEG XS itself and sends it to the browser as H.264.
- Packets and Frames - packets, frames (and fields for interlaced video), packets per frame, the size of the last codestream, and the RFC 9134 header of the last packet.
- JPEG XS Errors - frame counter gaps, missing markers and codestream size mismatches (the three JPEG XS incidents), plus any other packet problem that occurred. A counter above 0 is shown in the warning or error color with the word "warning" or "error".
The probe cannot decode RGB JPEG XS. Such a flow shows Not decodable (RGB JPEG XS) in place of its thumbnail; its values and counters are still measured.
15 ST 2110-30
The ST 2110-30 page provides real-time analysis of SMPTE ST 2110-30 uncompressed audio streams (AES 67). It monitors packet integrity, audio levels, and timing characteristics for each detected audio flow.
The page follows the Source page layout pattern:
- Page title - The audio icon and the title.
- Flow picker - Chooses one of the detected ST 2110-30 flows.
- Detail panel - Full analysis of the currently selected audio flow.
This page requires the 2110 license option. The meters, waveform, loudness and listening also need the Audio option.
15.1 Flow Picker
Choose the flow with the flow picker in the toolbar. Click it to open the list: search by name or address, filter, group and sort the flows, and pick one. Every row shows a health dot, the flow name and its bitrate; the counts of active and inactive flows are in the picker. The chosen flow is in the page address (?flow=<id>) and the browser remembers it, so a link opens the same flow again. The arrows next to the picker step to the previous or next flow (the picker is described in 9.6). For audio flows the rows also show the channel count and sample rate (for example 8ch 48kHz) and the number of sequence gaps.
15.2 Stream Identity Panel
The Stream Identity panel confirms the audio format detected from the packet stream:
| Field | Description |
|---|---|
| Channel Count | The number of audio channels carried in the flow (e.g. 2, 8, 16, 64). |
| Sample Rate | The audio sample rate in Hz (typically 48,000 Hz or 96,000 Hz). |
| Bit Depth | Bits per sample (typically 24-bit for broadcast). |
| Packet Time | The duration of audio contained in each RTP packet (e.g. 1 ms for AES 67 Class A, 125 us for Class B). |
| Samples per Channel per Packet | The number of audio samples per channel in each RTP packet, derived from sample rate and packet time. |
| Nominal RTP Timestamp Increment | The expected RTP timestamp difference between consecutive packets. |
15.3 Sequence Analysis Panel
The Sequence Analysis panel tracks RTP sequence continuity for the audio stream:
| Field | Description |
|---|---|
| Total Audio Units | Total audio packets (Audio Units) received since the last reset. |
| RTP Seq Gaps | Count of sequence number discontinuities. Each gap represents lost audio packets, which cause audible clicks or dropouts. Should be zero. |
| RTP Duplicates | Count of duplicate sequence numbers. |
| Last RTP Seq | The most recently received sequence number. |
15.4 Level Meters
The Level Meters display per-channel audio levels in real-time. Each channel is shown as a vertical bar meter with:
- Bar - The sample peak in dBFS (decibels relative to full scale), as a digital peak program meter per IEC 60268-18. The attack is instant, the release is 20 dB in 1.7 s.
- Peak hold marker - Holds the peak for 2 s, then falls at the same rate as the bar.
- Alignment tick - A tick mark at -18 dBFS (EBU R 68 alignment level).
- Colors - Green below -18 dBFS, yellow up to -9 dBFS, red above.
- Readout - Shows the held peak. The RMS level is shown in the tooltip.
The meter shows exactly the number of channels the ST 2110-30 stream carries (detected from the packets, maximum 64). Channel numbers are shown up to 32 channels, dB values up to 16 channels.
When channel groups are configured (see chapter 23 for details), the meters display:
- Group brackets - A labeled bar spanning each group's channels, showing the layout name (e.g. "5.1") and group name.
- Speaker labels - Standard speaker abbreviations (L, R, C, LFE, Ls, Rs, etc.) below each meter.
- Group health indicators - A thin colored bar below each group: green (all channels active), amber (some silent), red (all silent).
Silence detection: When all channels have been below -60 dBFS for 3 seconds or more, a red alert banner appears showing the elapsed silence duration.
15.5 Timing Panel
The Timing panel shows inter-arrival time statistics relative to the nominal packet interval:
| Metric | Description |
|---|---|
| Average | The mean inter-arrival time. Should match the declared packet time. |
| Max | The longest gap between consecutive packets. |
| Jitter | The variation in inter-arrival time. High jitter can cause audio glitches on strict receivers. |
15.6 Audio Playback Controls
The detail panel includes playback controls for monitoring the selected audio stream:
| Control | Description |
|---|---|
| Audio On/Off | Starts or stops audio playback. The button glows green when playback is active. |
| DIM | Reduces playback volume by 10 dB for quick level comparison. The button glows amber when active. |
| Reset Errors | Resets all cumulative error counters to zero. |
| AAC / PCM | While audio plays: the format it comes in and its measured rate (see 23.2.1). |
15.7 Waveform Display
The waveform display shows the last 3 seconds of audio of the selected flow, one row per channel, scrolling in real time. The rows are the channels the level meters show, each with its channel number; a channel that is not in use draws a flat line. The button at the top right switches between a dB scale (-60 to 0 dBFS, as the peak meters) and a linear scale. The display shows the lowest and highest sample of every 4 ms with a resolution of about 0.6 dB from -72 to 0 dBFS, so quiet audio is drawn in fine steps; on the dB scale anything below -60 dBFS sits on the center line, and on the linear scale audio below about -40 dBFS is too small to see.
15.8 Packet Timeline
The packet timeline visualizes recent packet arrival times, helping operators identify timing irregularities, bursts, or gaps in the audio stream.
15.9 ST 2110-31 AES3 Audio
An ST 2110-31 flow carries each AES3 subframe whole: the 24 audio or data bits plus the validity, user, channel status and parity bits, and the flags that mark the start of every frame and of every 192-frame channel status block. The probe recognizes these flows by their packets (no SDP needed) and lists them as ST 2110-31 Audio, with the sample depth 24-bit (AM824). They open on the ST 2110-30 page, where the meters, the waveform, listening and the Dolby E, AC-3 and E-AC-3 pairs work as for ST 2110-30.
The probe also checks the AES3 layer of every pair and counts:
- Parity errors: subframes whose parity bit is wrong.
- Flag errors: frame or block start flags in the wrong place.
- Block errors: a channel status block that starts too early or whose start is missing. A lost packet is counted as an RTP gap, not as a block error.
- CRC errors: channel status blocks with a wrong checksum. The probe keeps showing what the last good block said.
- Channel status changes: what a pair's channel status says (professional or consumer, audio or non-audio, sample rate, word length) changed.
- Validity bit set: samples the source marks as not fit for playback.
The probe does not need the parity bits to be right to recognize a flow: a sender that sets them wrong or not at all is still listed as ST 2110-31, and the parity errors are counted. A sender may mark the first frame of a block with the block flag only, which is accepted too. A flow is listed as ST 2110-31 after two packets in a row that fit and show a block start.
For every pair the probe shows what its channel status says: professional or consumer, audio or non-audio, the sample rate and the word length, and whether the checksum is good (CRC ok, CRC error, or no CRC for a consumer block or a minimum professional block, which carry none). Until a block with a good checksum was read the line says no status yet. A pair is shown as non-audio when either of its two subframes says so. A non-audio pair is not metered or played as PCM.
Incidents. The AES3 layer raises two incident types, both on by default (see 27.1 for the Incidents page and 31.4 for the check settings):
- ST 2110-31 AES3 errors (
aes3_errors, an error): parity, flag, block start and channel status checksum errors. The detail names the kinds and how many, for exampleparity 3, channel status CRC 2. - ST 2110-31 AES3 channel status change (
aes3_status_changed, a warning): what a pair's channel status says changed. The detail is the pair number.
Both checks are in the Audio card of Settings > Checks (section 31.4): there they are switched off, given another severity, or changed for a group or a flow with an override.
Sample rates. ST 2110-30 and ST 2110-31 flows at 44.1 kHz, 48 kHz and 96 kHz are recognized without an SDP. A 44.1 kHz sender keeps the sample counts of 48 kHz, so 48 samples are 1.088 ms. A flow that fits both starts as 48 kHz; after about 50 ms the probe compares the RTP clock with the arrival time and changes it to 44.1 kHz when that is what the sender runs at.
16 ST 2110-40
The ST 2110-40 page provides real-time analysis of SMPTE ST 2110-40 ancillary data streams. Ancillary data includes timecode, closed captions, AFD (Active Format Description), and other metadata carried alongside video and audio in a broadcast workflow.
The page follows the Source page layout pattern.
This page requires the 2110 and the Ancillary license options.
16.1 Flow Picker
Choose the flow with the flow picker in the toolbar. Click it to open the list: search by name or address, filter, group and sort the flows, and pick one. Every row shows a health dot, the flow name and its bitrate; the counts of active and inactive flows are in the picker. The chosen flow is in the page address (?flow=<id>) and the browser remembers it, so a link opens the same flow again. The arrows next to the picker step to the previous or next flow (the picker is described in 9.6). For ancillary flows the rows also show the ANC type summary (for example TC + CC + AFD) and the number of sequence gaps.
16.2 Timecode Panel
The Timecode panel displays a live SMPTE 12M-2 timecode readout:
- Live timecode display - A large, ticking timecode display showing hours, minutes, seconds, and frames.
- Drop Frame / Non-Drop Frame badge - Indicates the timecode mode.
- Packet count - The number of timecode packets received.
If no timecode is detected in the ancillary stream, a "No timecode detected" message is shown.
Timecode checks. Below the Timecode panel, the Timecode Checks panel follows each timecode type the flow carries (LTC, VITC1, VITC2) on its own, because they can differ in the same frame. For each type it shows:
- The running timecode, with a semicolon before the frames when it is drop frame, and the state: Counting, Stalled (the same timecode again while the video moves on), Missing (the type stopped), Invalid digits or Measuring the frame rate (the first frames after the flow started).
- The frame rate the probe measured from the RTP timestamps, and Drop frame yes or no. Above 30 fps the timecode counts frame pairs, so the page says "frame pairs".
- Problems: counters for jumps (including going backwards), stalls, missing and invalid timecode since the flow started.
- Last bad: the last bad timecode with the one that was expected, for example
Jump: 10:24:33:18, expected 10:24:33:13 (1.2 s ago). A lost packet is not a jump: the probe looks at the RTP timestamp to see how many frames passed.
Offset from PTP compares the timecode with PTP time and shows the difference in frames (+3 frames means the timecode is 3 frames ahead), with the timecode PTP time gives for that frame (Expected), where the local time comes from, and when an alarm is raised. The compare needs a flow whose RTP timestamps are on the PTP time scale and a PTP grandmaster on the input. When it cannot run, the panel says why in plain words, for example "Not compared: no PTP grandmaster is seen on the input of this flow." The timecode is a local time of day, so without a local offset the probe compares it with UTC and says "No local offset configured: shown against UTC"; the offset is then shown but never alarmed.
The checks are set in the Timecode and ANC card of Settings > Checks (section 31.4). Four check types are used; Timecode off PTP time is off until it is switched on, the others are on:
| Check type | Incident | Settings |
|---|---|---|
anc_tc_jump | Timecode jump | enabled |
anc_tc_stall | Timecode stall | frames: repeats before the incident, 1 to 250 (default 2) |
anc_tc_missing | Timecode missing | missing_ms: time without the timecode, 40 to 60000 ms (default 1000) |
anc_tc_ptp_offset | Timecode offset from PTP time | max_frames 0 to 10000 (default 1), hold_s 1 to 3600 (default 5), utc_offset_min -720 to 840, dst |
For the PTP offset, utc_offset_min is local time minus UTC in minutes (60 for UTC+1) and dst adds one hour for daylight saving. The incident is raised when the offset stays over max_frames for hold_s seconds and ends after 5 seconds within the limit. The ST 2022-6 page shows the same panel under VANC Ancillary.
SCTE-104 panel. When the flow carries SCTE-104 messages (ANC DID 0x41, SDID 0x07, SMPTE ST 2010: the splice and segmentation cues an automation system sends to the downstream SCTE-35 inserter), the SCTE-104 panel shows them:
- Counters per operation type, for example
splice_request_data 12,time_signal_request_data 3,insert_segmentation_descriptor_request_data 3. A duplicate (a message the encoder sends again and marks DUPLICATE_MSG) is not counted twice. - Messages, duplicates and not listed: the probe lists at most 20 messages a second per flow; the others are counted only.
- Decode errors with the last reason, for example
data_length 3, splice_request_data needs 14: messages that could not be read and operations whose data does not fit their table. Incomplete counts messages over several ANC packets that lost a packet; non-conforming counts packets of a mapping version the probe does not know. - The last 50 messages, newest first: the time, the RTP timestamp, the message number and one line such as
splice start normal, event 42, pre-roll 4000 ms, break 30.0 s with auto return, program 1. Click a message to see every field of it.
A message split over several ANC packets is joined before it is shown. Each listed message of a monitored flow is also a line in the log (28.1), an alarm event (31.19) and kept in the history database for 7 days (GET /api/v1/flows/{id}/scte104 in the REST API); a flow that is not monitored, and any flow in the first seconds after its input starts (the run-in), shows its messages here only. The ST 2022-6 page shows the same panel under VANC Ancillary.
16.3 Ancillary Data Types (DID/SDID Table)
The Ancillary Data Types panel lists every distinct DID/SDID (Data Identifier / Secondary Data Identifier) combination observed in the stream:
| Column | Description |
|---|---|
| DID/SDID | The identifier pair in hexadecimal notation (e.g. 0x61/0x01). |
| Description | The human-readable name of the ancillary data type (e.g. "EIA-708 Closed Captions"). |
| Specification | The relevant standard reference (e.g. "SMPTE ST 334-1"). |
| Count | The number of ANC sub-packets of this type received. |
This table provides a complete inventory of all metadata types carried in the ancillary stream. Common DID/SDID types include:
| DID/SDID | Description |
|---|---|
| 0x60/0x60 | SMPTE 12M-2 Timecode |
| 0x61/0x01 | EIA-708 Closed Captions (CEA-708) |
| 0x61/0x02 | EIA-608 Closed Captions (CEA-608) |
| 0x41/0x05 | Active Format Description (AFD) |
| 0x41/0x07 | Payload Identifier (SMPTE ST 352) |
| 0x41/0x04 | SCTE 104 Messages |
16.4 Closed Captions Panel
When closed caption data is detected, the Closed Captions panel displays:
- Caption statistics - Characters per second, words per minute, and total character count.
- Live caption viewer - A scrollable display showing decoded caption text in real-time, with:
- Timecode timestamps for each caption line
- Color-coded text spans (supporting CEA-608 color attributes)
- A blinking cursor on the current line
The caption viewer supports both EIA-608 (line 21) and CEA-708 (digital television) caption formats.
The Caption Services panel next to it lists the caption services the flow has carried, each as Present, Quiet or Stopped: CEA-608 field 1 (CC1 and CC2), CEA-608 field 2 (CC3 and CC4), CEA-708 and OP-47 subtitles (OP-47 on the ST 2110-40 and ST 2022-6 pages only; a video PID of an MPTS has the first three). XDS data on field 2 is not caption content, so a stopped CC1 is found while XDS goes on. A service that was present for 10 seconds is expected; when it then has no captions for the delay (30 seconds by default) the incident Captions stopped is raised (25.5) and the panel shows Stopped with the time of the last captions. Admin and NOC users see Stop expecting on an expected service, for a program that has no captions: the service is not expected again until it has been present for 10 seconds. The panel is on the ST 2110-40 and ST 2022-6 flow pages, and on the MPTS tab for the video PIDs of a transport stream.
16.5 Active Format Description (AFD) Panel
When AFD data is present, the AFD panel shows:
- Visual diagram - A representation of the active picture area relative to the full frame, showing pillarbox or letterbox regions.
- AFD Code - The AFD code value in hexadecimal.
- Description - A human-readable description of the active format.
- Aspect Ratio - The signaled aspect ratio (e.g. 16:9, 4:3).
- Bar Data - Top, bottom, left, and right bar values in pixels.
16.6 Packet Statistics Panel
| Metric | Description |
|---|---|
| RTP Packets | Total RTP packets received for this flow. |
| ANC Sub-packets | Total ancillary data sub-packets extracted from the RTP stream. |
| Sequence Gaps | RTP sequence discontinuities. Should be zero. |
| Duplicates | Duplicate RTP sequence numbers. |
16.7 Inter-Arrival Time Panel
| Metric | Description |
|---|---|
| Average | Mean inter-arrival time between consecutive packets. |
| Max | Longest gap between packets. |
| Jitter | Variation in inter-arrival time (in nanoseconds). |
17 ST 2022-6
The ST 2022-6 page provides real-time monitoring of SMPTE ST 2022-6 SDI-over-IP (HBRMT) streams. ST 2022-6 encapsulates a complete SDI signal - video, embedded audio, and ancillary data - into a single IP flow using the High Bitrate Media Transport (HBRMT) protocol.
The page follows the Source page layout pattern.
This page requires the 2022-6 license option.
17.1 Flow Picker
Choose the flow with the flow picker in the toolbar. Click it to open the list: search by name or address, filter, group and sort the flows, and pick one. Every row shows a health dot, the flow name and its bitrate; the counts of active and inactive flows are in the picker. The chosen flow is in the page address (?flow=<id>) and the browser remembers it, so a link opens the same flow again. The arrows next to the picker step to the previous or next flow. For ST 2022-6 flows the rows also show the video thumbnail (when available), the format (for example 1080i50) and an SDI rate badge.
The SDI rate badge in the flow picker is color-coded:
| Badge | SDI Rate | Color |
|---|---|---|
| SD-SDI | 270 Mbps | Gray |
| HD-SDI | 1.485 Gbps | Blue |
| 3G-SDI | 2.97 Gbps | Green |
| 12G-SDI | 11.88 Gbps | Orange |
17.2 SDI Format Panel
The SDI Format panel shows the detected video format:
| Field | Description |
|---|---|
| SDI Rate | The SDI interface rate badge (SD, HD, 3G, or 12G). |
| Format | The detected video format string (e.g. 1080i50, 720p59.94). |
| Resolution | The active picture resolution (e.g. 1920x1080). |
| Frame Rate | The detected frame rate. |
| Scan Type | Interlaced or Progressive. |
| Source / Destination | The source and destination IP addresses of the flow. |
17.3 HBRMT Header Panel
The HBRMT Header panel displays the transport-layer metadata from the ST 2022-6 encapsulation:
| Field | Description |
|---|---|
| VSID | Video Source Identifier. |
| Reference Clock | Whether the sender is locked to a reference clock. |
| Scrambling | Whether the SDI payload is scrambled. |
| FEC | Whether Forward Error Correction is enabled. |
| Clock Frequency | The sample clock frequency. |
| MAP | Media Access Point value. |
| Frame Code | The video format code from the HBRMT header. |
| Frame Rate Code | The frame rate code. |
| Sample Code | The sample structure code. |
17.4 Video Thumbnail
A live decoded thumbnail of the video content, updated approximately once per second. The thumbnail includes a format and resolution overlay.
17.5 Embedded Audio Panel
The Embedded Audio panel shows the status of SDI embedded audio:
| Field | Description |
|---|---|
| Status | Active or None - whether embedded audio is present. |
| Audio Groups | Which of the 4 SDI audio groups (1-4) are active, shown as a visual grid. |
| Total Channels | The total number of active audio channels. |
| Sample Rate | The audio sample rate (typically 48 kHz). |
| Bit Depth | The audio bit depth (typically 24-bit). |
17.6 VANC Ancillary Panel
The VANC Ancillary panel shows the status of metadata carried in the Vertical Ancillary data space of the SDI signal:
| Data Type | Standard | Status |
|---|---|---|
| Timecode | SMPTE 12M-2 | Present / Not detected |
| AFD | ST 2016-3 | Present / Not detected |
| Captions | CEA-708 | Present / Not detected |
| EDH | RP 165 | Present / Not detected |
| SCTE 104 | ST 2010 | Present / Not detected |
| Payload ID | ST 352 | Present / Not detected |
A collapsible DID/SDID Survey section shows the full list of observed DID/SDID pairs with their labels and packet counts.
17.7 Statistics Panel
The Statistics panel shows overall flow health and performance:
- Health badge - Healthy, Degraded, Error, or Inactive.
- Reset counters button - Resets all cumulative counters to zero.
| KPI | Description |
|---|---|
| Frames | Total video frames received. |
| Packets | Total RTP packets received. |
| Packets/Frame | Average packets per video frame. |
| FR Gaps | Frame rate gaps (missing frames). Should be zero. Shown in red when non-zero. |
| Measured FPS | The actual measured frame rate. |
| Bitrate | The current data rate. |
The panel also includes:
- Bitrate sparkline - A 5-minute rolling chart showing bitrate over time.
- Frame rate sparkline - A 5-minute rolling chart showing measured frame rate.
- Incidents list - Any active warnings or errors for this flow.
18 MPTS
The MPTS (Multiple Program Transport Stream) page provides comprehensive monitoring of MPEG Transport Streams. MPTS is the standard container format for digital television, carrying multiple programs (each with video, audio, and data) in a single multiplexed stream.
This page requires the MPTS license option. Audio playback also needs the Audio option, and pictures and video playback also need the Video option.
18.1 Stream Overview
The MPTS page displays each detected transport stream as a card showing:
- The total bitrate of the multiplex
- The number of programs detected
- The transport stream ID
- Packet size (188, 192, or 204 bytes)
- The number of TS packets per Ethernet frame
Streams picker. The button Streams 3 of 5 above the cards chooses which streams the page shows. Click it to open the list: the streams are grouped by input, each row shows the number of programs, the bitrate, the P1, P2 and P3 lamps of ETR 290 and a status (Error, Was in error, OK, Inactive). Type in the search box to find a stream by name, address or input, tick or untick a stream with a click or the Space key, and use Select shown and Clear in a group heading to choose or drop a whole group. The chosen streams also show as chips beside the button (above four streams only the count). When you open the page without a saved choice, all streams are chosen and a stream that appears later is added (the footer of the list says so). With a choice in the page address or saved by the browser, exactly those streams stay chosen and new streams are not added; a stream that only appears because you changed the filter bar is not new. The choice is in the page address (/mpts?flows=13,14) and the browser remembers it. A stream that is not monitored cannot be chosen: its card always shows.
A card with Waiting for PAT... receives TS packets but has not seen a PAT yet. A card with No TS data: the RTP packets are empty (60 bytes, 10/s) receives packets that carry no TS packets at all: for example a DVB-to-IP gateway keeps an untuned channel alive with empty RTP packets. No PAT will come until the sender tunes the channel and sends TS data. The text names the packet size and the packet rate of the flow; after the first TS packet arrives it gives way to the normal view. The same text replaces Waiting for PAT... in the flow details and on the MPTS tab. ETR 290 does not report errors for such a stream: all its indicators stay Starting until TS packets arrive.
18.2 Program Grid
Each program within the transport stream is shown as a tile, sorted by name within each transport stream. A tile contains:
| Element | Description |
|---|---|
| Video thumbnail | A live decoded thumbnail of the program's video content (updated periodically). |
| Codec badge | The video codec: H.264, H.265, or MPEG-2 (top-left corner). |
| Resolution badge | The native resolution with category label: SD, HD, FHD, or 4K. |
| Audio chip | Click to toggle audio playback. Shows the channel count (e.g. "2ch", "6ch"). |
| CC badge | Appears when continuity counter errors are detected for this program. |
| PCR badge | Appears when the PCR (Program Clock Reference) interval exceeds 100 ms. |
| Lock badge | Appears when the program is scrambled (encrypted). Scrambled programs cannot be played. |
| Radio icon | Shown for radio services (audio only). |
| Data icon | Shown for services without video and audio. |
| Expand button | Opens the full-screen live video for this program. |
A program is detected as scrambled from the CA descriptor in its PMT and also from the TS scrambling bits of its elementary streams, because some multiplexers signal no CA descriptor. A scrambled program shows a padlock and a SCRAMBLED test picture.
To record one program, use the Record button on the stream bar of the transport stream or on the row of the program in the Services list. The New recording dialog opens with that program and a name from the service (29.5.1, 29.5.2).
18.2.1 Resolution Categories
Programs are categorized by resolution:
| Category | Resolution |
|---|---|
| SD | Up to 720x576 |
| HD | Up to 1280x720 |
| FHD | 1920x1080 |
| 4K | 3840x2160 and above |
Resolution filter chips at the top of the page allow showing only programs of a specific category.
18.2.2 Audio Playback
Click a program's thumbnail or audio chip to start audio playback. When audio is playing:
- The thumbnail shows a green border and a green speaker icon.
- Click again to stop playback.
Scrambled programs (indicated by a lock icon) cannot be played.
In the full-screen video (Expand button) the sound plays in step with the picture (lip sync): each piece of audio is played when the picture with the same PTS is on screen. Audio that arrives early waits for its picture (up to 5 s); audio that arrives after its picture holds the picture back briefly instead (up to 1 s). After a source change, a replay loop or a jump in the stream the sound fades out for a moment and continues in step. A program without audio PTS plays its sound with a short fixed delay (100 ms, raised automatically up to 0.5 s when the audio arrives unevenly).
18.3 Transport Stream Analysis
IP-Probe monitors the following transport stream health indicators:
| Indicator | Description |
|---|---|
| Continuity Counters | The 4-bit continuity counter in each TS packet must increment by 1 for each PID. Any discontinuity indicates lost packets. |
| PCR Accuracy | The Program Clock Reference must arrive at regular intervals (typically every 40 ms, maximum 100 ms per DVB specifications). |
| PID Table | A complete list of all PIDs (Packet Identifiers) in the transport stream, showing their type (video, audio, PMT, PAT, etc.) and individual bitrate. |
18.4 Packet Size
IP-Probe automatically detects the transport stream packet size:
| Size | Description |
|---|---|
| 188 bytes | Standard MPEG-TS packet. Used in IP streaming, file storage, and HLS. |
| 192 bytes | 188-byte packet with a 4-byte timing/sync prefix. Used in some broadcast systems and professional equipment. |
| 204 bytes | 188-byte packet with 16 bytes of Reed-Solomon forward error correction (RS(204,188)). Used in DVB-T, DVB-S, and DVB-C transmission. |
Any change in the detected packet size or the number of TS packets per Ethernet frame is flagged as an event.
18.5 ETR 290 (TR 101 290)
The ETR 290 page checks every MPTS stream against the measurement guidelines of ETSI TR 101 290 v1.4.1 (often called "ETR 290"). It shows the 24 indicators of clause 5.2 per stream, reports every error as an episode with a severity (a RAISE when it starts, repeats while it goes on, a CLEAR when it ends; see 18.5.4), and keeps a count, the first and last occurrence and a history of the last hour (per minute) and the last 24 hours (per half hour).
Open it from the menu under Sources > MPTS > ETR 290, from the ETR 290 lamps on an MPTS card, or from the info bar of the MPTS pages.
ETR 290 is part of the MPTS license option. Without it, the menu entry is grayed and the probe does not run the checks.
The three priorities. TR 101 290 groups the indicators by how much they matter:
| Priority | Title in TR 101 290 | Indicators |
|---|---|---|
| Priority 1 | Necessary for decodability (basic monitoring) | 1.1, 1.2, 1.3a, 1.4, 1.5a, 1.6 |
| Priority 2 | Recommended for continuous or periodic monitoring | 2.1, 2.2, 2.3a, 2.3b, 2.4, 2.5, 2.6 |
| Priority 3 | Application dependent monitoring | 3.1a to 3.8 (11 indicators) |
Next to the three priorities, the DVB subtitle checks form a fourth group, Level 4 (DVB subtitles, EN 300 743), as level 4 in the DVBControl products. It has its own tile, lamp, filter and 24-hour strip and its own on / off switch in the profiles, and its errors do not count for Priority 3 (see "DVB subtitle checks" below). A stream without DVB subtitles shows Level 4 as not evaluated.
States. Every indicator has one of these states, shown as a color, an icon and a word:
| State | Look | Meaning |
|---|---|---|
| Error | red, exclamation mark | At least one episode of the indicator is open: the error goes on, or it ended less than the resolve hysteresis ago (2 s by default). |
| Was in error | amber outline, clock | The indicator had an episode that ended within the hold time (10 minutes by default). |
| OK | green check | Checked, no error. |
| Not present | gray dash | An optional table is absent, the stream is scrambled, or not evaluated. |
| Off | outline, block sign | Switched off by the settings or the profile. |
| Starting | blue hourglass | The stream (re)started; waiting for the PAT, the PMTs or the SI. |
The color of an Error or Was in error follows the severity of the indicator's episode (critical, major, minor, warning or info, see 18.5.4); the icon and the word keep the state.
While 1.1 TS_sync_loss is set, all other indicators are not evaluated, as the standard says. A packet with the transport_error_indicator set and a section with a bad CRC are not checked any further.
Overview. The overview lists every MPTS stream the filter bar shows, worst first: in error now, was in error, OK, not checked. The tiles at the top count the streams and filter the table when clicked. Table shows a lamp and count per priority, the checks in error as chips, the last error, the profile and a 24-hour strip, with a compact "Errors at a glance" matrix above it. Matrix shows one cell per indicator per stream. The Table / Matrix choice is saved in your own preferences on the probe. When you come from a source on the Endpoints page, the overview shows only the streams of that source and a chip From: <name> (the page address has ?src= and the source address); the x on the chip shows all streams again, and the quick filters work on top of it. Above the table the panel Recent episodes lists the episodes of all streams with their input and TS ID (18.5.6).
Stream view. Click a stream to open its view: the priority tiles, an IP transport tile (RTP loss and MDI; not part of TR 101 290), a compact matrix of the 24 indicators, and the check table grouped by priority. A group with nothing in error is collapsed; click its header (or press Enter or Space) to open it. A group you collapse yourself stays collapsed when a new error comes in, but its badge turns red; Reset view in the page menu clears your choices. Click a check to open its details: the setting, the count, a chart per minute, a row per item (a table, a program or a PID) with its own limit, last and worst value (18.5.5), and the episodes of the stream (18.5.4).
Limits: P and S. A limit marked P comes from the profile assigned to the stream; S means it was changed for this stream only. A limit without a mark is the TR 101 290 standard value.
Profiles and stream changes. The limits live in Settings > ETR 290 (Admin only). A named profile (for example Global, Contribution, DTT) holds the limits; every stream uses the default profile unless another one is assigned, and a stream can add its own changes on top ("Profile DTT + 1 change"). A stream can also be set to Not checked.
Tolerance. Under Settings > ETR 290 > Behavior the Tolerance field (default 5 %, 0 to 50 %, Admin only) lets every maximum interval be exceeded by that much before an error is raised: with 5 %, a limit of 10 s alarms above 10.5 s. The check table writes such limits with the tolerance, for example "10 s (+5 %)". Minimum intervals, PCR accuracy and counts are not affected. The level 2 threshold (18.5.4) is calculated from the widened limit.
Table intervals on the stream's own clock. The repetition checks of the tables (1.3a, 1.5a, 3.1a to 3.8, 3.2) measure the time between sections on the PCR time of the stream, not on their arrival at the probe: network jitter or a datagram held back on the way does not make a table late or too fast. A stream without a usable PCR is measured on arrival; the Tables page of the stream then says Timed on arrival (no PCR).
PIDs and what they are. Wherever the stream view names a PID (the last error of a check, the PIDs involved and the latest events) it also says what it is, for example "PID 256 - PES HEVC video", "PMT (program 28006)" or "BAT". PIDs are shown as decimal numbers (256) or as hexadecimal (0x0100), as set in Settings > System > Display > Show PIDs as; the choice is saved in your own preferences and applies to every page that shows a PID. Log lines always show both, for example "PID 256 (0x0100)".
Resets. Right-click a cell or a check row (or use the context-menu key, or the "..." button) to reset one counter, all counters of the stream, or, on the overview, one check on all streams. Admin and NOC users can reset; every reset is written to the audit log.
18.5.1 Priority 1 checks
1.1 TS_sync_loss
What it checks. Whether the receiver keeps the transport stream in sync: every TS packet starts with the sync byte 0x47.
When it is set. Two or more consecutive packets have a bad sync byte. Sync is regained after five consecutive good sync bytes. Over IP the check is done per datagram; a payload that is not a multiple of 188 or 204 bytes counts as lost sync.
Typical causes. A source that sends something other than a transport stream on this address, a wrong packet size setting at the encoder, a broken gateway or a damaged datagram.
What to do. Check that the source sends 188-byte (or 204-byte) packets and that only one source sends to this address and port. Compare with the packet size shown on the MPTS pages.
Settings. Good sync bytes to acquire: default 5 (2 .. 20). Bad sync bytes for loss: default 2 (1 .. 20).
Related. While 1.1 is set, every other indicator is not evaluated. 1.2 Sync_byte_error counts single bad sync bytes.
1.2 Sync_byte_error
What it checks. The sync byte of every TS packet.
When it is set. A sync byte is not 0x47.
Typical causes. Data corruption on the way, a device that changes the stream, or a source that mixes packet sizes.
What to do. Look for errors on the network path and for a second source on the same address. Frequent sync byte errors lead to 1.1 TS_sync_loss.
Settings. No limit to set; the check can be switched off per profile or per stream.
Related. 1.1 TS_sync_loss.
1.3a PAT_error_2
What it checks. That the PAT (Program Association Table) arrives often enough and is valid.
When it is set. No section with table_id 0x00 arrives on PID 0x0000 within the limit, a section on PID 0x0000 has another table_id, or a packet on PID 0x0000 is scrambled.
Typical causes. A multiplexer that sends the PAT too rarely, lost IP packets, or a multiplexer that puts other data on PID 0.
What to do. Check the PAT interval set in the multiplexer. If RTP loss is shown at the same time, look at the network path first.
Settings. Max. interval: default 0.5 s (0.1 .. 10 s).
Related. While the PAT is missing, 1.5a, 1.6 and 3.4a are not evaluated (they would only repeat this error).
1.4 Continuity_count_error
What it checks. Every TS packet carries a 4-bit continuity counter per PID that goes up by one for each packet with payload. The probe follows it on every PID.
When it is set.
- packets of a PID arrive in the wrong order;
- a packet occurs more than twice (one repeat is allowed);
- a packet is lost.
Not counted: the null PID 0x1FFF, a jump where the discontinuity_indicator is set, and packets without payload.
Typical causes.
- Lost IP packets: one lost datagram is 7 TS packets, so several PIDs get an error at the same moment. Compare with RTP lost in the IP transport tile.
- A remultiplexer or splicer that does not keep the counters.
- Switching between two sources (main / backup) without the discontinuity flag.
What to do.
- RTP loss at the same time: look at the network path (switch drops, link load, the multicast route).
- No RTP loss: look at the source or the multiplexer named in the event detail (PID, service).
- Check the errors per PID in the details panel.
Settings. No limit to set; the check can be switched off per profile or per stream.
Related. 1.1 TS_sync_loss suppresses this check; packets with 2.1 Transport_error are not checked further.
1.5a PMT_error_2
What it checks. That the PMT (Program Map Table) of every program in the PAT arrives often enough.
When it is set. No section with table_id 0x02 arrives on a PMT PID of the PAT within the limit, or a packet on a PMT PID is scrambled. PIDs of the network information are not included.
Typical causes. A multiplexer that sends the PMTs too rarely (TS 101 154 recommends 100 ms, but many networks send them less often), lost IP packets, or a program that was removed without a PAT update.
What to do. Check the PMT interval of the multiplexer. If the network sends PMTs every second on purpose, raise the limit in the profile (for example 1 s) instead of switching the check off.
Settings. Max. interval: default 0.5 s (0.1 .. 10 s).
Related. 1.3a PAT_error_2 (while the PAT is missing, this check is not evaluated).
1.6 PID_error
What it checks. That every elementary stream listed in a PMT actually occurs.
When it is set. A video or audio PID listed in a PMT has no packet within the limit. Data, subtitles and audio with an ISO 639 audio type above 0 (for example audio description) are excluded from the video and audio limit.
Typical causes. An encoder that stopped, a service whose audio was switched off without a PMT update, or a multiplexer that drops a PID.
What to do. Find the PID and the service in the details panel and check the encoder of that service.
Settings. Video and audio PIDs: default 5 s (1 .. 60 s). Other PIDs: Not checked by default; when checked, the Other PIDs limit applies (default 10 s, 1 .. 300 s).
Related. 1.5a PMT_error_2, 3.4a Unreferenced_PID.
18.5.2 Priority 2 checks
2.1 Transport_error
What it checks. The transport_error_indicator bit in the header of every TS packet.
When it is set. A packet has the transport_error_indicator set: a device before the probe (a demodulator, a gateway) found an error it could not correct.
Typical causes. Reception problems at a satellite, cable or terrestrial receiver that feeds the IP stream.
What to do. Check the receiver or the gateway that makes the IP stream. The count is a running total that you can reset.
Settings. No limit to set.
Related. A packet with this bit set is not checked further.
2.2 CRC_error
What it checks. The CRC of every section of the PAT, CAT, PMT, NIT, EIT, BAT, SDT and TOT.
When it is set. The CRC_32 of a section does not match its content.
Typical causes. Corruption on the way, or equipment that changes a table without computing a new CRC.
What to do. Look at the table and PID named in the event. If the error comes with RTP loss, look at the network.
Settings. No limit to set.
Related. A table with a bad CRC is not checked further.
2.3a PCR_repetition_error
What it checks. The time between two PCRs (Program Clock References) of a program.
When it is set. Two consecutive PCRs of a program are further apart than the limit.
Typical causes. An encoder or multiplexer that sends too few PCRs, or lost IP packets that carried a PCR.
What to do. Check the PCR interval of the encoder or the multiplexer for that program.
Settings. Max. PCR interval: default 100 ms (10 .. 1000 ms).
Related. 2.3b PCR_discontinuity_indicator_error, 2.4 PCR_accuracy_error.
2.3b PCR_discontinuity_indicator_error
What it checks. The step between two consecutive PCRs of a program.
When it is set. The difference between two PCRs is negative or larger than the limit, and the discontinuity_indicator is not set.
Typical causes. A switch between sources or a splice without the discontinuity flag, or a multiplexer that rewrites the PCR badly.
What to do. Check what happened at the time of the event (a source switch, a splice) and the device that did it.
Settings. Max. PCR step: default 100 ms (10 .. 1000 ms).
Related. 2.3a PCR_repetition_error.
2.4 PCR_accuracy_error
What it checks. Whether each PCR matches its position in the stream (PCR_AC, TR 101 290 clause 5.3.2.6). The check uses the byte position of the PCR, not the arrival time, so network jitter does not count.
When it is set. A PCR is off its byte position by more than the limit. The check applies to a constant bitrate stream only; a variable bitrate stream shows "not present".
Typical causes. A multiplexer that does not correct the PCR when it moves packets, or a remultiplexer with a bad PCR restamping.
What to do. Check the PCR restamping of the last multiplexer before the probe.
Settings. Max. inaccuracy: default 500 ns (100 .. 10000 ns). Off by default: the check compares the PCR with the byte position and the probe's own clock, which is not locked to the PTP grandmaster yet, so a correct stream can be reported. The settings page says "Off by default: needs a probe clock locked to PTP" next to the check. Switch it on in Settings > ETR 290 (per profile or per stream) when you want it anyway; a setting you saved before is kept. The PCR table on the Packet Timing page and the MPTS Timing tab still shows the measured accuracy.
Related. 2.3a PCR_repetition_error.
2.5 PTS_error
What it checks. The time between two PTSs (Presentation Time Stamps) of an elementary stream.
When it is set. Two PTSs of a PID are further apart than the limit. Still pictures are excluded, and a scrambled stream shows "not present" because its PES headers cannot be read.
Typical causes. An encoder that sends long gaps, a radio service with still pictures, or lost IP packets.
What to do. For a service that carries still pictures on purpose, switch the check off for that stream and give the reason.
Settings. Max. PTS interval: default 700 ms (100 .. 5000 ms).
Related. 1.6 PID_error.
2.6 CAT_error
What it checks. That a stream with scrambled packets has a CAT (Conditional Access Table), and that PID 0x0001 carries only the CAT.
When it is set. Packets are scrambled but no CAT is present, or a section on PID 0x0001 has a table_id other than 0x01.
Typical causes. A scrambler or multiplexer that does not insert the CAT, or a configuration that puts other data on PID 1.
What to do. Check the conditional access setup of the multiplexer.
Settings. No limit to set.
Related. 3.4a Unreferenced_PID (EMM PIDs are listed in the CAT).
18.5.3 Priority 3 checks
3.1a NIT_actual_error
What it checks. The NIT (Network Information Table) of the actual network on PID 0x0010.
When it is set. No NIT actual section arrives within the maximum interval, a section on PID 0x0010 has a table_id other than 0x40, 0x41 or 0x72, or two sections arrive closer than the minimum interval.
Typical causes. A multiplexer without NIT, a NIT sent too rarely, or SI equipment that sends bursts.
What to do. Check the SI generator or the multiplexer. Contribution feeds often carry no NIT: use a profile with Priority 3 off for them.
Settings. Max. interval: default 10 s (1 .. 60 s). Min. interval: default 25 ms (0 .. 1000 ms).
Related. 3.1b NIT_other_error, 3.2 SI_repetition_error.
3.1b NIT_other_error
What it checks. The NIT of other networks, once it is present in the stream.
When it is set. The gap between two NIT other sections with the same section number is longer than the limit (over all other networks). A single NIT other that repeats too slowly is reported by 3.2. The check starts only when NIT other appears; until then it shows "not present".
Typical causes. An SI generator that sends NIT other irregularly.
What to do. Check the SI generator.
Settings. Max. interval: default 10 s (1 .. 120 s).
Related. 3.1a NIT_actual_error.
3.2 SI_repetition_error
What it checks. The repetition of every SI table against the limits of ETSI TR 101 211 clause 4.4.
When it is set. A section repeats faster than the minimum interval (the next copy of the same section starts less than 25 ms after the end of the last one, ETSI EN 300 468 clause 5.1.4) or slower than the maximum interval of its table (from the start of one copy to the start of the next). Every section counts on its own, so one slow service or EPG segment is reported here even when the table as a whole is in time. The time of a section is the PCR time of the packet that carries its first byte (its arrival when the stream has no usable PCR), so network jitter does not count.
Typical causes. SI equipment with wrong cycle times, or a network type with other intervals than the profile expects.
What to do. Choose the network type that fits the stream (satellite / cable or terrestrial) in the profile, and set single table limits where needed.
Settings. The network type, the maximum interval per table and one minimum interval for all tables (default 25 ms, 0 .. 1000 ms), in the profile (3.2 SI repetition: table limits). The severity and the level 2 threshold are set per check in the profile table (18.5.4). The details of the check show one row per table with its own limit, and a click on a row shows the events of that table (18.5.5). Satellite / cable maxima (ETSI TR 101 211 clause 4.4.1): NIT 10 s, BAT 10 s, SDT actual 2 s, SDT other 10 s, EIT p/f actual 2 s, EIT p/f other 10 s, EIT schedule of the first 8 days 10 s and of the later days 30 s (actual and other), TDT / TOT 30 s. Terrestrial (clause 4.4.2) allows longer intervals for EIT p/f other (20 s) and the EIT schedule: the first full day actual 10 s and other 60 s, the later days actual 30 s and other 300 s.
Related. 3.1a, 3.5a, 3.6a, 3.8.
3.4a Unreferenced_PID
What it checks. That every PID in the stream is known: listed in a PMT or the CAT, or a PID with a fixed meaning.
When it is set. A PID that is not a PMT PID, not in the range 0x0000 .. 0x001F and not a user-defined private PID is not referenced within the limit.
Typical causes. A multiplexer that forwards a PID of a removed service, or a PMT that misses a component.
What to do. Find the PID in the event detail and check which device inserts it.
Settings. Grace time: default 0.5 s (0.1 .. 10 s).
Related. 1.6 PID_error (the other direction: listed but missing).
3.5a SDT_actual_error
What it checks. The SDT (Service Description Table) of the actual transport stream on PID 0x0011.
When it is set. No SDT actual section arrives within the maximum interval, a section on PID 0x0011 has a table_id other than 0x42, 0x46, 0x4A or 0x72, or two sections arrive closer than the minimum interval.
Typical causes. An SI generator that sends the SDT too rarely, or a stream without service names.
What to do. Check the SI generator. Without an SDT, receivers show no service names.
Settings. Max. interval: default 2 s (0.5 .. 60 s). Min. interval: default 25 ms (0 .. 1000 ms).
Related. 3.5b SDT_other_error, 3.2 SI_repetition_error.
3.5b SDT_other_error
What it checks. The SDT of other transport streams, once it is present.
When it is set. The gap between two SDT other sections with the same section number is longer than the limit (over all other transport streams). A single SDT other that repeats too slowly is reported by 3.2.
Typical causes. An SI generator that sends SDT other irregularly.
What to do. Check the SI generator.
Settings. Max. interval: default 10 s (1 .. 120 s).
Related. 3.5a SDT_actual_error.
3.6a EIT_actual_error
What it checks. The EIT present / following (Event Information Table) of the actual transport stream on PID 0x0012.
When it is set. EIT present (section 0) or EIT following (section 1) of the actual transport stream is missing for the limit (over all services), a section on PID 0x0012 has a table_id outside 0x4E .. 0x6F and 0x72, or the same section arrives again closer than the minimum interval. A single service whose EIT repeats too slowly is reported by 3.2.
Typical causes. An EPG system that stopped, or a multiplexer that drops the EIT.
What to do. Check the EPG system and the multiplexer.
Settings. Max. interval: default 2 s (0.5 .. 60 s). Min. interval: default 25 ms (0 .. 1000 ms).
Related. 3.6b EIT_other_error, 3.6c EIT_PF_error.
3.6b EIT_other_error
What it checks. The EIT present / following of other transport streams, once present.
When it is set. The gap between two EIT present (or two EIT following) sections of other transport streams is longer than the limit (over all services). A single service whose EIT repeats too slowly is reported by 3.2.
Typical causes. An EPG system that sends EIT other irregularly.
What to do. Check the EPG system.
Settings. Max. interval: default 10 s (1 .. 120 s).
Related. 3.6a EIT_actual_error.
3.6c EIT_PF_error
What it checks. That EIT present and EIT following come as a pair.
When it is set. Only one of the two (present or following) exists for a service: one stays away for twice the 3.6a / 3.6b limit while the other keeps coming (at least three times). A section that is only late is reported by 3.2.
Typical causes. An EPG system with a gap in the schedule.
What to do. Check the EPG data of the service named in the event.
Settings. No limit to set.
Related. 3.6a EIT_actual_error.
3.7 RST_error
What it checks. The RST (Running Status Table) on PID 0x0013, when present.
When it is set. A section on PID 0x0013 has a table_id other than 0x71 or 0x72, or two RST sections arrive closer than the minimum interval.
Typical causes. Wrong data on PID 0x0013, or bursts from the SI generator.
What to do. Check the SI generator.
Settings. Min. interval: default 25 ms (0 .. 1000 ms).
Related. 3.2 SI_repetition_error.
3.8 TDT_error
What it checks. The TDT (Time and Date Table) on PID 0x0014.
When it is set. No TDT arrives within the maximum interval, a section on PID 0x0014 has a table_id other than 0x70, 0x72 or 0x73, or two sections arrive closer than the minimum interval.
Typical causes. An SI generator without a time source, or a TDT sent too rarely.
What to do. Check the SI generator and its time source. Receivers use the TDT for their clock and the EPG.
Settings. Max. interval: default 30 s (5 .. 120 s). Min. interval: default 25 ms (0 .. 1000 ms).
Related. 3.2 SI_repetition_error.
5.3.4.1 Transport_Stream_ID_check
What it checks. That the stream carries the transport stream you expect: the transport_stream_id in the PAT is compared with the expected TS ID set for the stream (TR 101 290 clause 5.3.4.1). It is listed under Priority 3.
When it is set. The TS ID in the PAT differs from the expected TS ID. Without an expected TS ID the check shows "not present".
Typical causes. A wrong multiplex on this address (a routing or multicast mix-up), or a multiplexer configured with another TS ID.
What to do. Check which source sends to this address and port, and the TS ID set in the multiplexer.
Settings. The expected TS ID is set per stream in the stream dialog (Settings > ETR 290 > Streams), in hexadecimal (for example 0x0403).
Related. 1.3a PAT_error_2.
Table_version_conflict
What it checks. That each sub-table of the stream (for example the EIT present/following of one service) comes from one source. It is not a TR 101 290 indicator; it is listed under Priority 3 and switched with the other Priority 3 checks.
When it is set. A sub-table keeps changing back and forth between the same two versions: it goes back to the version (and number of sections) it had just before, three times within 60 seconds. Every further change back counts as an error. The item of the episode is the sub-table, for example "EIT p/f actual, service 6390 (NAIZ irratia)" or "EIT p/f other, TS 1007, service 100"; the detail names the two versions, for example "versions v1 (2 sections) and v2 (1 section)". The episode stays open for 60 seconds after the last change back, so a table sent only every few seconds (an SDT or EIT of another transport stream, the EIT schedule) keeps one episode; then it is cleared.
Typical causes. Two sources send the same table for one service: two EIT or SI inserters, or two multiplexer inputs that both carry the EIT of a service, so the receivers see a new version with almost every section. A receiver then keeps reloading the table, and the EPG of that service may flicker or stay empty.
What to look for. On the Tables page (18.6.1) the sub-table shows Conflicting versions with the two versions and their number of sections. Check which inserters and multiplexer inputs carry the EIT (or the other table) of that service, and remove the second source or give it another service or table.
Settings. On by default, severity warning (minor at level 2), with a level 2 in errors per second that is off by default; set per profile in Settings > ETR 290 like the other checks.
Related. The Tables page (18.6.1).
Level 4: DVB subtitle checks (EN 300 743)
Five checks of DVB subtitles (ETSI EN 300 743 V1.6.1) in their own group, Level 4, after the three priorities. TR 101 290 itself has no subtitle checks; the names and the group are the ones of the DVBControl products, so both report alike. A profile switches the whole group on or off with its own switch; the Priority 3 switch does not cover it. They run on every PID that a PMT signals with a subtitling descriptor of a DVB subtitle type, all the time (also while the Subtitles page is closed). The item of an episode is the subtitle PID, for example "PID 131 (0x0083) DVB subtitles". Without a DVB subtitle stream in the PMTs the checks show "not present"; a PID whose subtitling descriptor lists only teletext subtitles does not count. They are on by default, have the severity warning (minor at level 2) and a level 2 in errors per second that is off by default; the Subtitles page (25.3) shows each display set with its faults.
| Check | When it is set | Value of an event |
|---|---|---|
| Subtitle_arrival_error | A display set is complete more than 100 ms after its PTS (EN 300 743 clause 5.1.2). A set that is late by no more than 100 ms is not an error. | how late, in ms |
| Subtitle_PTS_step_error | The PTS of a display set is less than one video frame (40 ms) after the PTS of the set before (clause 8.3). | the step, in ms |
| Subtitle_buffer_error | A display set is larger than the coded data buffer of the decoder: 24 KB, or 100 KB once the PID sent a display definition segment (clause 5.0). | the size, in bytes |
| Subtitle_page_id_error | A segment is for a page that the subtitling descriptor of the PMT does not list for the PID (clause 8.0). | the page_id |
| Subtitle_segment_error | The PES or a segment breaks the syntax: no PES header, a wrong data_identifier or subtitle_stream_id, a byte other than the sync byte where a segment belongs, a segment cut off or longer than the PES, or a progressive pixel block that cannot be read (clauses 6.2 and 7.2). | the byte offset in the PES data field |
Typical causes. A subtitle inserter or multiplexer that sends its display sets too late or too close together, an encoder that makes display sets too large for SD decoders, a wrong page id in the PMT after a configuration change, or a damaged subtitle PID.
What to do. Open the Subtitles page (25.3) on the PID: the report shows which display sets have the fault, with the numbers and the rule of EN 300 743. A wrong page id is usually fixed in the PMT (the subtitling descriptor), the other faults at the subtitle inserter or the multiplexer.
Settings. On or off, the severity and the level 2 per check in the profile (Settings > ETR 290); the 100 ms late margin is fixed.
Related. The incident Subtitles stopped (31.4.4.12).
Not measured: 3.3, 3.9 and 3.10
3.3 Buffer_error, 3.9 Empty_buffer_error and 3.10 Data_delay_error need a model of the decoder buffers (the T-STD model of ISO/IEC 13818-1). They are not measured yet and are listed as "not measured".
18.5.4 How ETR 290 Errors Are Reported
The probe does not report every single error. It follows an error as an episode: one episode for each stream, check and item, from the first error until the error is gone. An episode is announced once (RAISE), repeated now and then while it goes on, and closed once (CLEAR). A table that is late in every cycle for an hour is therefore one episode with a count, not thousands of lines. The same messages go to the log, the incident history, the ETR 290 pages and the alarm export (31.19).
What an item is. The item says what the check is about:
| Item | Checks | Example |
|---|---|---|
| The stream | 1.1, 1.2 | the whole stream |
| A table | 1.3a, 3.1a, 3.1b, 3.2, 3.5a, 3.5b, 3.6a, 3.6b, 3.6c, 3.7, 3.8 | "EIT schedule actual (first 8 days)" |
| A PID | 1.4, 1.6, 2.1, 2.3a, 2.3b, 2.4, 2.5, 2.6, 3.4a, the subtitle checks | "PID 256 - PES HEVC video" |
| A program | 1.5a | "Das Erste HD (program 28006)" |
| A PID and a table | 2.2 | PID 17 and table_id 0x42 |
A program is named by its service name from the SDT, followed by its program number ("Das Erste HD (program 28006)"); a name longer than 48 characters is cut, and on the pages the full name is in the tooltip. A program the SDT does not name shows as "PMT (program 28006)". A service renamed in the SDT is shown under its new name at once, also for an episode that is still open; a history row keeps the name it had when it was written, and the log and alarm lines keep "PMT (program 28006)" because the alarm names the service itself.
When a check runs out of room for separate items (a stream with more than 48 items in error at once), the rest is summed in one item other; the stream view says how many errors went there.
The life of an episode.
- Start. The first error of a stream, check and item opens the episode. With Raise after at 0 (the default) it is announced at once; with a higher value the error must last that many seconds first, and a shorter episode is only recorded in the history (marked "short"), without a log line or alarm.
- RAISE. The first message names the stream (name, address and TS ID), the check, the item, the severity and the first value, for example
RAISE ETR 290 3.2 SI_repetition_error [EIT schedule actual (first 8 days)] (warning) => PID 18 (0x0012) interval 12.4 s, limit 10 s (+5 %). A RAISE can also start a recording that is set to start on an ETR 290 event. - Repeats. While new errors keep coming, the probe repeats the message at growing intervals, set by the Error transmit times. A repeat says how many errors came since the previous message and the worst value of the whole episode:
+30 more (total 36), worst 13.8 s. There is no repeat while no new error comes, and never more often than the list allows, so a long fault is a handful of messages. The running count and worst value that the pages and the history show are updated at most every 5 seconds. - CLEAR. The episode ends when the error is gone for the Resolve hysteresis (2 s by default). For a check that measures an interval (3.2 and the other repetition checks) the table must also have been on time once more, and the error must be gone for at least the check's own limit when that is longer than the hysteresis (for example 10 s for the EIT schedule of the first 8 days, 5 s for a missing video PID). So a table that is late in every cycle does not clear and raise again in every cycle. For such a check the worst value is the longest interval measured, including the one that ended an absence (a section that arrives after 30 s gives 30 s, not the 21 s at which level 2 was reached). The CLEAR gives the duration, the total count and the worst value:
CLEAR ... duration: 18 min 23 s count: 1104 worst: 14.2 s. The duration runs to the last error; the hysteresis is not part of it. A CLEAR line is stamped with the end of the episode, so it appears in the log a few seconds after that time.
How an episode can end. The history and the alarm targets name the reason:
| Reason | Meaning |
|---|---|
| cleared | The error was gone for the resolve hysteresis (and, for an interval check, the table was on time and the error was gone for at least the check's limit). |
| level_change | The error went to the other level (below); a new episode of the new level follows at once. |
| reset | An Admin or NOC user reset the counters of the check (18.5, Resets). |
| settings | The check was switched off, or the stream is no longer checked, in the settings. |
| restart | The analysis of the stream started over (a gap in the data, or the probe stopped). |
| monitoring_off | The monitoring of the flow was switched off (31.11). |
| quiet | The stream went away (no more traffic, or its analysis was released) while the episode was open. |
The timeline of one episode. An example with the default settings: stream "Mux 3" (input 1, 239.1.1.1:5000, TS ID 1019). From 06:37:06 its multiplexer sends the EIT schedule sections late: about one section a second is over the 10 s limit of the table (10.5 s with the 5 % tolerance). Check 3.2 has the severity warning (the default of Priority 3). The fault goes on until the multiplexer is repaired at 06:55:29.
| Time | Message | What it says |
|---|---|---|
| 06:37:06 | RAISE | first value: interval 12.4 s, limit 10 s (+5 %) |
| 06:37:11 | REPEAT (5 s later) | +5 more (total 6), worst 12.9 s |
| 06:37:41 | REPEAT (30 s later) | +30 more (total 36), worst 13.8 s |
| 06:38:41 | REPEAT (60 s later) | +60 more (total 96), worst 14.2 s |
| 06:43:41 | REPEAT (300 s later) | +300 more (total 396), worst 14.2 s |
| 06:53:41 | REPEAT (600 s later) | +600 more (total 996), worst 14.2 s |
| 06:55:29 | last late section | nothing is sent; the probe waits for an on-time cycle and then the limit of the table (10 s) |
| 06:55:29 | CLEAR (sent about 10 s later) | duration 18 min 23 s, count 1104, worst 14.2 s |
The next repeat would have come 1800 s after 06:53:41 (then 3600 s, and from then on every 3600 s). Note that each wait is counted from the previous message, not from the start: 5 s after the RAISE, then 30 s after that repeat, and so on. The first entry of the list (0) is the RAISE itself.
Where the operator sees this:
- Logs page. Three kinds of lines: the RAISE above, the REPEAT lines (
=> REPEAT ETR 290 3.2 ... (warning) => +30 more (total 36), worst 13.8 s) and the CLEAR line. RAISE and REPEAT are at the log level of the severity (Error for major and critical, Warning for the others, Status for info); a CLEAR is a Status line. A CLEAR that ended because of a level change, a reset, a settings change, a switched-off monitoring or a stop of the probe says so at the end of the line, for example(level change). - ETR 290 stream view. From 06:37:06 the indicator 3.2 is in Error, colored by its severity (amber for a warning). In the 3.2 details (18.5.5) the row "EIT schedule actual (first 8 days)" shows the limit "10 s (10.5 s)", the last and the worst value and the error count growing, with a chip Warning, level 1 while the episode is open. In the Episodes list a row appears at 06:37:06 with the item, the severity, the count and the worst value against the limit, marked open with a duration that runs; after the CLEAR it shows the final duration, and the reason as a tooltip.
- ETR 290 overview. The panel Recent episodes (18.5.6) badge shows 1 open from 06:37:06; its row names the stream, input, TS ID, check, item, severity, duration and count.
- Incidents page and ETR 290 history. One episode row for the whole fault, not one per cycle. At most 20 episodes per stream, check and hour have their own row; further ones are summed in a row "+N more".
- Alarm targets. The RAISE is a raise, each REPEAT is an update of the same alarm (same alarm id; count, "+N more" and worst value), and the CLEAR is a clear with the reason. Syslog, the SNMP trap
ipprobeAlarmUpdate, the Alert Center, webhooks and Zabbix get these as they happen. Email collects alarms for 30 seconds into one mail: an update that arrives while the mail of its raise still waits is folded into that mail ("+N more since this alarm was raised", the worst value with its unit and the last detail); later updates do not make a mail of their own. In this example the mail goes out at 06:37:36 with the RAISE and the repeats of the first 30 seconds, and the CLEAR follows in its own mail in the same thread.
Two levels. Most checks have a second, higher threshold. The first level is the limit of the check (with the tolerance, for the maximum intervals); the second level is a percentage of that limit, or for a count check a number of errors per second. In the example, level 2 of check 3.2 is 200 %, that is 21 s (200 % of 10.5 s). If a section is 22 s late at 06:44:10, the warning episode ends with a CLEAR (level change), and a RAISE of a level 2 episode with the level 2 severity (minor by default) follows at once for the same item. When no value is over level 2 for the resolve hysteresis, the episode steps back: CLEAR (level change) of level 2 and a RAISE of level 1 again if the error goes on. For an interval check the step back also waits for at least the check's own limit and for one interval of the item within level 2 again, so a table that is sent every 30 s against a 10 s limit stays one level 2 episode instead of stepping back each time its sections arrive. An item that is fine again goes from level 2 straight to CLEAR, without the step back. In the lists, a level 2 episode shows "Minor, level 2".
| Setting in the profile | Meaning | Default |
|---|---|---|
| Severity | The severity of the level 1 episode of the check. | P1 major, P2 minor, P3 warning |
| Level 2 | The threshold of level 2. Percent of the limit for the checks 1.3a, 1.5a, 1.6, 2.3a, 2.3b, 2.4, 2.5, 3.1a, 3.1b, 3.2, 3.5a, 3.5b, 3.6a, 3.6b, 3.6c and 3.8 (101 to 1000 %); errors per second for 1.2, 1.4, 2.1, 2.2, 2.6 and the subtitle checks (1 to 100000). Empty = no level 2. | 200 % for the percent checks; off for the count checks |
| Level 2 severity | The severity of the level 2 episode. A check without a level 2 (1.1, 3.4a, 3.7, the TS ID check and the checks that are not measured) shows a dash. | P1 critical, P2 major, P3 minor |
The severities. There are five, from the least to the most serious: info, warning, minor, major and critical. They set the color on the pages, the level of the log line, the severity of the incident and the severity of the alarm sent to the targets (so a target with a minimum severity of major gets the Priority 1 errors and no warnings). By default Priority 1 errors are major (critical at level 2), Priority 2 minor (major at level 2) and Priority 3 and Level 4 warning (minor at level 2).
| Severity | Color on the ETR 290 pages |
|---|---|
| critical | red, filled |
| major | red |
| minor | amber |
| warning | amber |
| info | blue |
The stream view has a legend for these colors. The icon of the state (error or was in error) does not change with the severity. Episodes that were recorded before the severity was kept (from an older version) show a dash as the severity in the history.
The tolerance. The tolerance (default 5 %, Settings > ETR 290 > Behavior) widens the maximum intervals: with 5 %, a limit of 10 s is an error above 10.5 s, and the level 2 threshold is calculated from the widened limit. It does not widen minimum intervals, PCR accuracy or counts, it does not change the standard limit that the pages show (they show both, "10 s (10.5 s)"), and it does not change the severity: a value inside the tolerance is simply not an error.
Settings and where they are. All in Settings > ETR 290 (Admin only); a change applies to the running streams within a second.
| Setting | Where | Meaning | Default |
|---|---|---|---|
| Raise after | Behavior | How long an error must last before it is announced (0 = at the first error). | 0 s |
| Resolve hysteresis | Behavior | How long without the error before the episode is cleared; interval checks also need one on-time interval and at least their own limit. Whole seconds, 1 to 600. | 2 s |
| Error transmit times | Behavior | The waits between the messages of a continuing episode, in seconds, rising, starting with 0 (up to 16 entries, each up to 86400). After the last entry the last wait repeats. | 0; 5; 30; 60; 300; 600; 1800; 3600 |
| Show "was in error" for | Behavior | How long an indicator shows "was in error" after the last episode ended (0 = until the counters are reset). | 10 min |
| Ignore after a stream starts | Behavior | No check counts this long after a stream (re)starts. | 0.5 s |
| Tolerance | Behavior | See above. | 5 % |
| Also report old indicators | Behavior | Also show the old combined indicators 1.3, 1.5, 2.3, 3.1, 3.5, 3.6. | off |
| On, Setting, Severity, Level 2, Level 2 severity | Profile table (per check) and the stream dialog | The check, its limit and the severity and level 2 above. The priority row has Severity for all and Level 2 severity for all to set a whole priority at once (level 2 only on the checks that have one). A changed value has a colored border and a button back to the default. | see above and the check texts |
| Send repeats | Settings > Alarm export, per target | On: the target also gets the repeats (updates). Off: it gets the RAISE and the CLEAR only. | on |
A value that is not valid (for example transmit times that do not rise, or a hysteresis that is not a whole number of seconds) is marked at the field and the page does not save until it is corrected. In the stream dialog, P marks a value that comes from the profile (not the standard value) and S a value set for this stream only, with a button to take it back.
A source that is slightly out of spec. The limits stay as the standard says. When one stream is always just over a limit (for example a PTS interval of 750 ms against 700 ms, or an EIT present / following that repeats every 2.3 s instead of 2 s) and you accept it, silence only that stream: open Settings > ETR 290 > Streams, add the stream (or open its row), and in the stream dialog set the check to Off, or to Custom with a looser limit (for example 2.5 Max. PTS interval 1000 ms). The other streams keep the standard limits. The stream dialog marks such a value with S. The timing checks of Settings > Checks (31.4.4.13) have the same per-flow override for the packet timing.
Alarm export. Every target has the switch Send repeats (on by default, 31.19). Syslog, SNMP, the Alert Center, webhooks and Zabbix then get the repeats as they happen; Email never sends a repeat as a mail of its own but folds it into the mail of its raise while that mail waits, as described above; the daily digest counts repeats. A target with the switch off sees one raise and one clear per episode. The hourly cap works for the alarms as for the history: after 20 episodes of one stream and check in an hour, the next ones are summed in one "+N more" alarm.
18.5.5 The 3.2 Details per Table, and the Item Click
TR 101 211 gives every SI table its own repetition limit, and several tables share one PID: PID 0x0012 carries EIT present / following (2 s on the actual stream), the EIT schedule of the first 8 days (10 s) and the later days (30 s). A PID row with "last 11 s" can therefore not be judged: it is late for one table and fine for another. For this reason the details of 3.2 SI_repetition_error have one row per table:
| Column | Meaning |
|---|---|
| Table | NIT actual / other, SDT actual / other, BAT, EIT p/f actual / other, EIT schedule actual / other (first 8 days and later), TDT, TOT. A row appears when the table has been measured (up to 13 rows). |
| PID | The PID the table was last seen on. |
| Limit | The limit of this table, with the limit including the tolerance in brackets, for example "10 s (10.5 s)". The limits come from the network type of the profile (satellite / cable or terrestrial). |
| Last | The last interval measured for the table. |
| Worst | The largest interval since the counters were reset. It is marked red only when it is over the limit of its own table. |
| Errors | The number of errors of the table. |
While the table has an open episode the row also has a chip with its severity and level. The same view exists for other checks that have several items: 2.2 per PID and table, 1.5a per program ("Per program"), 1.6 per PID with the limit for video and audio PIDs or for other PIDs, and the PID checks "Per PID". The title of the view names the kind ("Per table", "Per program", "Per PID").
Click a row (or press Enter on its name) to open the last events of that item, up to 16, newest first: time, value against the limit, PID, section number and level. Below them the Episodes of this item are listed. The list refreshes every 2 seconds while it is open; press Esc, the close button or click the row again to close it. The Episodes list under the check table has the episodes of the whole stream: open ones first, then the newest closed ones (time, item, severity, duration, count, and the worst value against the limit), and says when older episodes are no longer listed.
18.5.6 Recent Episodes on the Overview
Below the tiles the ETR 290 overview has the panel Recent episodes. Closed, it shows a badge with the number of open episodes of all streams, colored by the worst severity of them. Open it to see the episodes of all streams, newest first, with the columns time, stream (click it to open the stream with the check selected), input, TS ID, check, item, severity, duration and count. Open only hides the closed ones. This answers "which stream, which input and which table or PID" without opening each stream. The panel is for everyone who may see the page, the Demo role included; it needs the MPTS option like the rest of the page.
18.5.7 Defaults at a Glance
| What | Default |
|---|---|
| Raise after | 0 s |
| Resolve hysteresis | 2 s |
| Error transmit times | 0; 5; 30; 60; 300; 600; 1800; 3600 s |
| Tolerance | 5 % |
| Severity level 1 / level 2, Priority 1 | major / critical |
| Severity level 1 / level 2, Priority 2 | minor / major |
| Severity level 1 / level 2, Priority 3 | warning / minor |
| Severity level 1 / level 2, Level 4 | warning / minor |
| Level 2 threshold | 200 % of the limit (percent checks); off (count checks) |
| Show "was in error" for | 10 min |
| Episodes with their own row | 20 per stream, check and hour (history and alarms) |
| Send repeats | on, per alarm target |
| 3.2 table limits | TR 101 211 4.4.1 (satellite / cable) or 4.4.2 (terrestrial), see 3.2 |
18.6 Cards, Highlights and Detail View
Besides the overview in section 18.1, the MPTS page (menu Sources > MPTS > Overview) has the following details.
Stream cards. A card shows the stream name, bitrate, null percentage, the network name and one picture per service. A service shows a CC badge when it has continuity counter errors and a PCR badge when its PCR timing is out of limits. A red line at the bottom of the card summarizes the same ("CC errors", "PCR violation"). Click the open icon of a card to see the MPTS pages of that stream, on the page you used last. The small links PIDs | Services | Tables under the stream name open that page for that stream directly.
ETR 290 lamps. Next to the bitrate each card shows three lamps, P1, P2 and P3, for the three priorities of TR 101 290 (green: no error, red: in error now, outlined: was in error recently, gray: not evaluated). The same lamps are in the info bar of the MPTS pages. Click the lamps to open the ETR 290 stream view of that stream (Sources > ETR 290, section 18.5), where each indicator is listed with its count and history.
Highlights. When ETR 290 is checking the stream, the CC and PCR highlights follow it:
- CC is shown while check 1.4 (Continuity_count_error) is in error or was in error within the hold time, also when another Priority 1 check is in error. A service gets the badge when it has continuity counter errors of its own. Counters that stay above zero after the hold time no longer light the badge.
- PCR is shown while check 2.3a (PCR repetition) or 2.3b (PCR discontinuity) is in error or was in error within the hold time. On a card the badge marks the service that carries the PCR PID of the latest PCR error; when that PID is not known, the PCR error shows only in the red line at the bottom of the card. In the detail view the PCR value of a service is marked when it is above the PCR limit of the stream, or when the service carries that PCR PID. The limit shown includes the tolerance set in Settings > ETR 290 (for example 100 ms + 5 % = 105 ms).
- When ETR 290 is off, the stream is not checked, or a check is switched off for the stream, that highlight uses the raw counters and a fixed 100 ms PCR limit.
MPTS pages. The stream has three pages under Sources > MPTS: PIDs (/mpts/pids), Services (/mpts/services) and Tables (/mpts/tables); the page title is "MPTS PIDs", "MPTS Services" or "MPTS Tables". There is no tab bar: the side menu items PIDs, Services and Tables switch the page, and so do the keys 1, 2 and 3. Every page starts with the same info bar, so the stream stays in view when you change the page.
Info bar. One row with, from left to right, the back arrow, the stream picker and the facts of the stream: TSID, Network (when the NIT names it), Bitrate (in kbit/s, Mbit/s or Gbit/s, whichever fits the value), Null (in percent: amber above 10 %, red above 30 %), the number of Services and PIDs, and the ETR 290 lamps (click them to open the ETR 290 stream view). At the right are the DEC / HEX switch for the PID numbers, which shows the PIDs of the pages in decimal or hexadecimal (table ids are always hexadecimal), and the Record button. Below the row an alarm strip appears only while the stream has an alarm: CC errors detected (CC errors ETR 290 1.4 when the stream is checked by ETR 290) with the PIDs that have errors and the Reset CC link, and the PCR alarm (PCR interval > 100 ms, or PCR error (ETR 290) with the limit of the stream). Reset CC clears the continuity counters of the stream; it does not clear the ETR 290 state of check 1.4, which ends after its hold time or with a reset on the ETR 290 page. When ETR 290 no longer reports a CC error but the counters are still above zero, Reset CC is a button in the row instead. On a phone the row wraps: the picker first, the facts below it.
Switching stream. The picker in the bar is the same as the one of the MPTS page for a single stream: its button shows the chosen stream with its address and program count, the arrows step to the previous or next stream, and the list has the groups by input, the search, the ETR 290 lamps and the bitrate of every MPTS stream. Choosing another stream opens it on the same page: the Tables page keeps its group filter, its Tables / Sub-tables choice and its filter text, and starts again with the rows of the new stream. The page and the stream are in the page address (/mpts/tables?flowId=14), so the browser's Back button returns to the page or stream before and the address can be sent as a link. Press [ or ] (when no text field has the focus) to go to the previous or next stream in the order of the list.
Keys and the page used last. Press 1, 2 or 3 to open PIDs, Services or Tables; like [ and ] the keys do nothing while a text field, a list or a dialog has the focus, or when Ctrl, Alt or Meta is held. The side menu items open the stream you looked at last on their page; a link that names only a stream (from the Audio, Video or AV Monitor pages, the IPs page or the Packet Timing page) opens the page you used last. The browser remembers both.
Services and Timing tabs. Each stream card has a tab bar. Services shows the programs of the stream (the picture of each service). Timing shows the timing of the stream, the same views as on the Packet Timing page (chapter 21): the key values (packet rate with the TS packets per datagram, mean IAT, IAT peak-to-peak, standard deviation, 99.9th percentile, peak and lost packets), the IAT trend and the IAT histogram with the choice of the time window, then the PCR timing table per PCR PID (interval, arrival jitter, overall jitter, frequency offset, drift and accuracy, with the PCR jitter histogram of a PID on a row click) and the MDI panel (delay factor, media loss rate and the minute trend). Open in Packet Timing opens the stream there; the Packet Timing page links back to the open Timing tab. The open tab is in the page address (?timing=<id>), so a link or a reload keeps it. The stream's histogram is read from the probe only while the Timing tab is open. The values are judged against the timing checks of Settings > Checks (31.4.4.13). On a phone the key values are in two columns and the PCR table is a list of cards.
PIDs page. Every PID of the stream: its type, a description, the details, the bitrate (kbit/s over the last second, with a bar) and the continuity counter errors. The description of a PSI/SI PID names the tables it carries ("SDT actual, SDT other, BAT"); the ECM and EMM PIDs, the PIDs of the services with their stream type and language, PIDs that no PMT lists (Unreferenced) and the null PID are listed too. A PCR mark shows the PID that carries the PCR of a program, and a lock shows a scrambled PID. Details gives the picture size, frame rate and scan of a video stream ("1920x1080, 25 fps, Interlaced (TFF)") and the channels and sample rate of an audio stream ("5.1, 48 kHz"). When the page is wide enough, Min, Avg and Max show the lowest, average and highest one-second bitrate of the PID since its first packet (every PID, PSI/SI and other section PIDs too), and Packets the number of packets of the PID (a PID that a PMT lists but that never arrives shows 0 in red). On a wide page two more columns appear when the stream needs them. Crypto O / E / Sum (when a PID is scrambled) shows, in seconds, how long the last odd key period and the last even key period of a scrambled PID lasted and their sum, the crypto period; a dash stands for a period that has not completed yet (the first one after the PID turned scrambled is not counted, as its start was not seen). PCR TS Bitrate (on the PIDs that carry a PCR) shows the bitrate of the whole transport stream calculated from that PID's PCRs, over about one second; the REF mark shows the PCR PID that the PCR time of the table timing is locked to. The chips filter by video, audio, PSI/SI and other.
Services page. One line per service of the stream: its picture (the one the stream card on the overview shows, updated the same way; a placeholder says Scrambled, Radio, Data or Empty), the name with the provider and the SID, the type (SD TV, HD TV, UHD TV, Radio, Data, from the SDT or else from the components) with a lock when the service is scrambled, a short summary of the components ("H.264 1080i25", "MP2 deu", "Teletext", "Data x2"), the PCR interval, the bitrate with a bar (relative to the largest service of the stream) and the status.
- Status. OK when there is nothing to report. A red tag and a red edge on the line for an error: continuity counter errors ("2 CC errors"), a PCR interval over the limit of the stream (PCR interval) or a PCR accuracy over the ETR 290 limit (PCR jitter). An amber tag for a warning: a component that the PMT lists but that has not sent a single packet since the probe started (PID not received, shown once the page has seen it silent for 5 seconds; a PID that stops after it was received is reported by ETR 290 check 1.6 instead; SCTE-35 and private section components, which may stay silent for a long time, are never counted), video or audio without a PCR (PCR PID 8191, No PCR; a data service without a PCR is normal), a service the SDT calls not running (Not running), and scrambled packets without a CA descriptor in the PMT (Scrambled, no CA). A service whose PMT lists no components shows the gray tag No components. Not running shows only when the SDT lists the service as not running, starting soon or pausing; a service without SDT information is not marked. When a service has more than one issue, "+1" after the tag names the others when you point at it. The CC and PCR rules follow ETR 290 the same way as the info bar (see Highlights above).
- Opening a line. Every line starts closed. Click a line (or press Enter or Space on its arrow) to open it; several lines can be open at once, and Expand all and Collapse all in the toolbar open or close every line shown. The page remembers which lines you opened for each stream in the browser, so they are open again on your next visit. An opened line shows each issue in a sentence ("1 CC error on the video PID 2220, counted since the last Reset CC.") with a link to the page that tells more (Show in ETR 290, Show in PIDs, Show the PMT in Tables), the facts of the service (service type, SID, PCR PID, PCR interval and jitter, share of the transport stream, running state, CA), Open in the AV Monitor for a service with video or audio, and the component table: component, PID, codec (with profile and level), format (picture size, scan and frame rate, chroma and bit depth; channels and sample rate), language, stream type, bitrate, CC errors and whether the component is scrambled. The JPEG XS video descriptor and the anc_data_descriptor of a component are shown under it. On a wide window the larger picture of the service is shown next to the table. On a phone a line takes two rows and the components are small cards.
- Toolbar. The search finds a service by its name or provider (part of the text, any case), its SID or one of its PIDs (the whole number, in decimal, with 0x in hex, or as hex digits while the PIDs are shown in hex). All, TV, Radio and Data filter by type, Problems shows the services with an error or a warning, and Scrambled the scrambled services; each shows its count. When nothing matches, the page says so and Clear search and filters brings every service back. The search and the filters stay while you switch pages. The Sort menu sorts by Problems first (the default: errors, then warnings, then the rest), name, SID, type, PCR interval, bitrate, CC errors or the program order; the button names the sort and its direction (Problems last when the Status column is sorted the other way, an arrow for the other columns).
- Stream stopped. When no packets arrive for a few seconds, a banner says how long ("No packets for 1m 14s.") and the list shows the last known services and bitrates, dimmed. When the stream is gone (no longer received or removed), the pages say so and link to the MPTS overview; while the PAT and PMTs have not arrived yet they say Waiting for the service list.
Column headings. On the PIDs, Services and Tables pages the heading row stays at the top of the window while the list scrolls.
Sorting. On all three pages a click on a column heading sorts by that column, a second click reverses the order and a third click goes back to the natural order; the arrow in the heading and the screen reader state show the direction. Numbers sort by value, text without regard to case, and empty values come last in both directions. Columns of numbers start with the largest value, text and PID columns with the smallest. Sorting the State column of the Tables page from the largest puts Late first, then Too fast, Not received and Incomplete. On the Tables page the PIDs sort by the column and the tables of a PID and the sub-tables of a table follow the same column where it applies to them; on the Services page the column headings Service, Type, PCR int., Bitrate and Status sort the services, and sorting by bitrate or CC errors also orders the components of an opened service. On a narrow window the Tables page shows a Sort menu in the toolbar instead of column headings; its Default item goes back to the natural order. When the PIDs page is too narrow to show the column it is sorted by (Crypto and PCR TS Bitrate below 1080 pixels, Min, Avg, Max and Packets below 900, Details below 640), the list keeps the natural order until the column is shown again. Each page remembers its sort while you switch pages and in the browser. The printable report and the CSV and JSON exports of the Tables page follow the sort. When the page body is narrower than a list needs (a narrow window with the side menu and the summary open), the list scrolls sideways in a box of its own and the Tables page shows cards.
18.6.1 Tables page
The Tables page lists every PSI/SI table the transport stream carries, grouped per PID as a tree: PID > table > sub-table. It shows actual and other tables separately (NIT, SDT, EIT), and also the CAT, BAT, TDT, TOT, the EIT schedule, the PMTs of a shared PMT PID, private tables (AIT, SCTE 35, DSM-CC) and the ECM and EMM sections. The page reads the tables from the probe every 2 seconds while the page is open; nothing is read on another page. When the probe does not answer, the page keeps the last rows with a message above them and tries again less often (up to every 30 seconds) until it answers again.
A sub-table is one instance of a table: a service of the EIT or a transport stream of the SDT other, a network of the NIT, a bouquet of the BAT, a program of the PMT. The EIT schedule is shown in the classes of ETSI TS 101 211 (actual or other, first 8 days or later days), because the limits differ per class; the table ids and services are its sub-tables.
Toolbar.
- All / PSI / SI / EIT / Other shows one group of tables. PSI: PAT, CAT, PMT, TSDT. SI: NIT, SDT, BAT, TDT, TOT and the other SI tables. EIT: event information. Other: AIT, SCTE 35, DSM-CC, ECM / EMM and private tables.
- Tables / Sub-tables: Tables shows the tables closed (open one with its arrow); Sub-tables opens every table. On a narrow window the page always shows the tables.
- The filter field finds text in the PID (decimal or hex), the table name or id (in hex with 0x, or in decimal) and the names of the open sub-tables (a service name, for example). A table stays when one of its sub-tables matches.
- Intervals since tells when Last, Min, Avg and Max started counting: the start of the analysis, or the last reset.
- Export (see below) and Reset (Admin and NOC). Reset asks for a confirmation and starts Last, Min, Avg, Max and the counters of every table of the stream again; the tables, their versions and the cycles in progress stay. Reset clears the measurements of this page only. Users who may reset ETR 290 also see a check box Also reset ETR 290 for this stream in the confirmation (off by default): ticked, the reset also clears the ETR 290 counts, times and history of the stream, which ends its open episodes with a CLEAR (reset) to the alarm targets; the reset is written to the audit log. The result of the reset is shown above the table.
The summary line counts the PIDs with tables, the tables, the sub-tables, the sections per second and the CRC errors, names the tables that are late or too fast, counts the PMTs not received and the NIT, SDT or EIT present/following tables that are due but have not arrived. Timed on arrival (no PCR) there means that the stream carries no usable PCR, so its intervals are measured on arrival at the probe and include network jitter (see "How the interval is measured").
Columns.
| Column | Meaning |
|---|---|
| Table | The PID with the names of its tables, the table name, or the sub-table in words ("Service 5001 BBC News (TS 1, ONID 2)", "TS 1002, ONID 1", "Network 1", "Bouquet 25"). A PID with one table shows it on the PID line. A long name is cut with "..."; its tooltip shows the whole name. |
| ID | The table_id, for an EIT schedule class the range (0x50-0x51). Always hexadecimal, as the standards write it; the filter also finds it by its decimal value. |
| Version | The version_number. On a table whose sub-tables carry different versions it says per service (EIT, PMT: each service or program has its own version) or per sub-table (for example the SDT other, one per transport stream); the versions are on the sub-table rows. The tooltip gives the number of version changes with their rate per hour, for example "45 (3/h)" (on a table since the reset or the start of the analysis, on a sub-table since its first section or the reset; shown once a minute is covered), and the sections with current_next_indicator 0. |
| Sub-tables | How many sub-tables the table has. |
| Sections | Sections received since the reset; on a sub-table the sections of its version received out of the number the table announces ("2 / 2"). |
| Repetition interval Last / Min / Avg / Max | The complete cycle of a sub-table: from the start of its first section to the start of the same section's next copy, with every section of the version received in between. A table shows its worst sub-table. Values below 1 s are in ms, above in s with two decimals; a value past the limit is red. |
| Limit | Min / Max, for example "25 ms / 2.10 s": the shortest allowed gap from the end of a section to the start of its next copy, and the longest allowed cycle including the tolerance of the ETR 290 profile. When the table has no room for both, only the maximum is shown. The tooltip gives both and whether ETR 290 checks the table now. |
| State | See below. A late or too fast table has a link to check 3.2 on the ETR 290 page. |
| CRC Err | Sections with a wrong CRC. |
| kbps / CC Err | On the PID line: the bitrate of the PID (over the last 10 s, so a table sent every few seconds does not show 0) and its continuity counter errors. The tooltip of the bitrate lists the share of each table. |
States. The state uses the limits of the stream's ETR 290 profile (TS 101 211 clause 4.4, with the tolerance from Settings > ETR 290). It is worked out also when the ETR 290 checks of the stream are switched off; switching the checks off only stops the episodes and the alarms.
| State | Meaning |
|---|---|
| OK | Within the limits. |
| Late | The last cycle, the cycle in progress or an incomplete version took longer than the maximum with the tolerance, or the table has gone for that long. |
| Too fast | A section came again sooner than the minimum gap after the end of its last copy. |
| Incomplete n of m | The table announces m sections (last_section_number) and only n of them have arrived. Once this takes longer than the limit, the state is Late. |
| No limit | The profile has no limit for this table (for example the CAT, the AIT or SCTE 35). |
| Conflicting versions | A warning on a sub-table, and on its table and PID: the sub-table alternates between two versions, for example "Conflicting versions: v1 (2 sections) and v2 (1 section) alternate". See below. |
| Waiting | No complete cycle measured yet. |
| Not received | A PMT that the PAT lists but that has not arrived yet. It has a row of its own, "PMT (Program 5001 BBC News)", on its PMT PID. Once the PMT limit has passed since the PAT listed the program, the state is Late. When the PMT arrives, its table takes the place of this row. The same state and kind of row is shown for the NIT actual when the PAT lists a network PID, for the SDT actual, and for the EIT present/following actual when the SDT actual marks at least one service as having it; each goes when its table arrives. |
How the interval is measured. The time of a section is the time of its first byte in the stream itself: the PCR time of the packet that carries it, calculated from the PCRs of the stream's reference PCR PID (the PCR_PID of the first program; another PID with regular PCRs when it stops), so network jitter and delays between the sender and the probe are not part of the intervals. A section is timed once the PCR after it has arrived, a fraction of a second after the section itself. When the stream has no usable PCR (none at all, or the reference PCR stops for a moment), the arrival time of the datagram at the probe is used instead, and the page says Timed on arrival (no PCR): in the summary line when the whole stream is timed that way, on a sub-table row when only its last section was. The repetition interval of a sub-table is its whole cycle: from the start of its first section to the start of the next copy of that section, once every section of the version has arrived in between. A cycle that is not complete when the first section comes again goes on until it is, and a version change does not stretch it. Sections with a wrong CRC and sections with current_next_indicator 0 are only counted. After lost packets or a silence of the stream the next interval is not measured, because it would measure the outage. The minimum gap is measured from the end of a section to the start of its next copy. The state follows the last cycle: Max can be past the limit (red) while the state is OK again, because the last cycle was within it. The 3.2 details of ETR 290 (18.5.5) measure each section on its own, start to start of the same section_number, and show the worst value since their reset, so a table can be listed there and still be OK here; a table that is Late here for its whole cycle, such as the EIT schedule, is also over its limit there.
Conflicting versions. When two sources send the same sub-table with different versions, for example two EIT inserters or two multiplexer inputs that both carry the EIT present/following of one service, the sub-table changes version with almost every section. Once it has gone back to the version it had just before three times within 60 seconds, the page shows Conflicting versions with the two versions and their number of sections on the sub-table row, and on its table and PID rows (with the number of sub-tables when there are more). The tooltip names the sub-table, says since when, and how many changes were not counted. These changes back and forth are not counted as version changes, so the version change count and its rate per hour stay meaningful; a change to a third version still counts. The warning goes after 60 seconds without such a change. While it lasts, a version with more sections can show as incomplete, because the other source's version replaces it before all its sections arrive. ETR 290 reports the same as the check Table_version_conflict (18.5.3). Look for the second source of that service's table in the inserters and the multiplexer inputs.
ECM and EMM sections are counted per table_id without sub-tables: the interval is the gap between any two sections. Sub-tables that are not seen for three times their limit (at least 60 s) leave the list.
Narrow window. Below 600 pixels each PID is a card with one line per table (name, id, version, state, "last, max / limit"). Tap a table to see its sub-tables in a panel at the bottom of the screen; the panel closes when you switch to another page or the window gets wider.
Export. The Export button offers:
- CSV: the rows and columns shown, in the order shown, with the active group and text filter; with Sub-tables the sub-table rows too. The first column says whether a row is a PID, a table or a sub-table.
- JSON: every table with all its sub-tables and all values, for scripts, in the order shown.
- Printable report: opens a new browser window with the stream header (name, TS ID, input, destination, time, intervals since) and the tables as the page shows them; print it with the browser (Ctrl+P). When the browser blocks the new window, the report is saved as an HTML file instead.
The ETR 290 detail of check 3.2 has a link to this page.
19 SRT
The SRT (Secure Reliable Transport) page monitors SRT connections detected on the network. SRT is an open-source protocol that provides low-latency, reliable transport of video and audio over unpredictable networks such as the public internet.
This page requires the Streaming license option.
IP-Probe performs passive monitoring of SRT traffic by analyzing the SRT control and data packets captured on the wire. No SRT endpoint configuration is required - IP-Probe detects SRT connections automatically.
History
When you choose a past range, the stream view shows what the probe stored instead of the live checks. Nothing is stored per error: the probe keeps one episode per check and error period, and per minute the number of errors and the number of errored seconds (seconds with at least one error) of every check that had errors.
Choose the range above the checks: Live (the current view), Last 24 h, 7 days, 30 days or Custom (a start and an end time in your local time). The address of the page carries the range, so you can send a link to it. The page then shows:
- Errors over time: one strip per check that had errors, with the number of errors and errored seconds of the range. The shade is the share of errored seconds. The resolution is 1 minute for the last 24 hours, 10 minutes up to 7 days and 1 hour for older data; the page names the resolutions in use.
- Episodes, newest first: check, PID with its type, start, end, duration, count and the worst value against the limit in force. An episode that is still going on is marked open; an episode of a check that was masked by a TR 101 290 dependency (for example while TS_sync_loss is set) is marked suppressed; a row +N more sums the further episodes of that check in that hour (at most 20 episodes per check and hour have their own row). Load more reads the next page.
- Check selects one check for both parts. Export saves the selection as CSV or JSON, either the episodes or the counts per minute (at most 100000 rows; the page says when the file was cut).
The ETR 290 overview has a History link per stream and a Past streams section for streams that have stored history but are not on the wire now. If the history database cannot be read, the page says so in one message; monitoring continues. The history needs the MPTS option.
19.1 Connection Overview
The SRT page displays each detected SRT connection with:
| Field | Description |
|---|---|
| Connection name | The user-assigned alias or auto-generated label. |
| State | The current connection state (Connected, Connecting, Broken, etc.). |
| Source / Destination | The source and destination IP addresses and ports. |
| RTT | The round-trip time measured by SRT's built-in handshake mechanism. |
| Bandwidth | The estimated available bandwidth reported by SRT. |
| Bitrate | The current data rate of the SRT stream. |
19.2 SRT Statistics
For each SRT connection, the following statistics are displayed:
| Category | Metrics |
|---|---|
| Packets | Sent, received, lost, retransmitted, dropped. |
| Bytes | Total bytes sent and received. |
| Loss | Packet loss rate (percentage), lost packet count. |
| Retransmissions | Number of retransmitted packets and retransmission rate. |
| Latency | Configured latency buffer, measured RTT. |
| Jitter | Arrival jitter as reported in SRT control messages. |
19.3 SRT Flow Detail
Clicking an SRT connection navigates to the flow detail page, which shows the full protocol analysis including:
- SRT handshake parameters (version, encryption, congestion control)
- Real-time bitrate and packet rate charts
- Loss and retransmission event timeline
- The underlying RTP/MPEG-TS payload analysis (when applicable)
20 RIST
The RIST (Reliable Internet Stream Transport) page monitors RIST connections detected on the network. RIST is an interoperable protocol for reliable video transport over lossy networks, standardized by the VSF (Video Services Forum).
This page requires the Streaming license option.
IP-Probe performs passive monitoring of RIST traffic by capturing and analyzing RIST data and control packets.
20.1 Connection Overview
The RIST page displays each detected RIST connection with:
| Field | Description |
|---|---|
| Connection name | The user-assigned alias or auto-generated label. |
| Profile | The RIST profile in use (Simple, Main, or Advanced). |
| State | The current connection state. |
| Source / Destination | The source and destination IP addresses and ports. |
| Bitrate | The current data rate. |
20.2 RIST Statistics
For each RIST connection:
| Category | Metrics |
|---|---|
| Packets | Total sent, received, lost, recovered. |
| Loss | Packet loss rate before and after error recovery. |
| Recovery | ARQ (Automatic Repeat reQuest) requests sent, packets recovered. |
| RTT | Round-trip time as measured by RIST. |
| Quality | Overall link quality indicator. |
20.3 RIST Flow Detail
Clicking a RIST connection navigates to the flow detail page with full protocol analysis, including the underlying RTP payload.
21 Packet Timing
The Packet Timing page shows how evenly the packets of each flow arrive: the inter-arrival time (IAT), the time between two packets of the same flow. It works for every flow on the network, whatever its protocol. The ST 2110 and MPTS pages have their own timing analysis for their formats; this page puts the timing of all flows side by side, so you see at a glance which flow is the worst.
The values come from the probe once per second. The timestamps are taken by the network card when it has hardware timestamps; with software timestamps all timing values are less exact. The measurements are defined in 21.4.
| Value | Meaning |
|---|---|
| Mean IAT | The average time between two packets in the last second. |
| IAT p-p | IAT peak-to-peak: the longest minus the shortest time between two packets in the last second. The main figure for packet timing. |
| Min IAT, Max IAT | The shortest and the longest time between two packets in the last second. Very short times mean bursts; long ones mean gaps, congestion or buffering on the path. |
| Packet rate | Packets per second; the nominal IAT is one second divided by it. |
| Std deviation, Percentiles | How much the IAT varies and the time within which 99.9 % of the packets arrived (see 21.4). |
| Gaps | RTP sequence gaps since the start; for a transport stream the continuity counter errors. |
The page has three views, chosen with All flows, Flow and Compare in the toolbar. The view is in the page address (/timing?view=flow), so a link or a reload opens the same view.
Pause holds the values and charts on screen while you read them; the probe keeps measuring, and Resume shows the current values again. Reset clears the history the page keeps (the sparklines, the trend, the histogram of the window and the peaks); the probe's own counters keep running.
21.1 All flows
The page opens on a table of every flow that passes the filter bar, worst first: flows with an error come first, then warnings, then the rest by name. Flows that are not monitored are listed dimmed at the end.
| Column | Shows |
|---|---|
| Status | The status of the flow's checks (the colored dot; hover for the word). |
| Flow, Type, Address | The flow's name, its type and its destination address. |
| Bitrate | The bitrate of the flow. |
| Mean IAT, IAT p-p | The values of the last second. The IAT p-p is colored when it is past the limit of the flow type. |
| Last 60 s | A small chart of the IAT p-p over the last minute. |
| Timing verdict | The timing verdict of the flow type, see below. |
| Gaps | RTP sequence gaps, or continuity counter errors for a transport stream. |
Click a column heading to sort by it; click it again to reverse the order. The search box finds flows by name, alias, address, type or input. Problems shows only the flows with a warning or an error. The table is worst first by default; the verdict column is colored by the same limits that raise the timing incidents.
Click a row (or the flow's name) to open that flow in the Flow view. Tick the check boxes of up to four flows to put them in the Compare view.
The timing verdict depends on the flow type:
- ST 2110-20 and ST 2110-22 video: the ST 2110-21 sender type, Narrow, Narrow linear or Wide (all three are compliant and shown green; Wide needs the larger receive buffer), or Not compliant (shown red). While the sender type is not known yet (the first moments of a flow) the verdict reads Measuring. Video has no IAT limit by default: a gapped sender pauses during the vertical blanking, so its IAT p-p is large by design.
- ST 2110-30 and ST 2110-31 audio: the packet time class and the packet time (for example "Class C, 125 µs"). By default the IAT p-p is a warning when it is more than half the packet time and an error when it is more than the packet time.
- Transport streams: the MDI as "DF:MLR", judged against the limits of the MDI checks.
- Other types: No limit; the values are shown without a color. When a check sets an IAT p-p limit for the flow, the verdict reads IAT limit and the IAT p-p is judged against it.
The limits come from the timing checks in Settings > Checks (31.4.4.13): the IAT peak-to-peak, the RTP jitter, the MDI delay factor and media loss, and the PCR frequency offset and drift. A flow without an override uses the all-flows values; the probe raises an incident when a flow goes over a limit.
On a phone the table becomes a list of cards with the name, the type, the timing verdict, the IAT p-p and its sparkline.
21.2 Flow
The Flow view shows one flow. Choose it with the flow picker in the toolbar: click it to search, filter, group, sort and pin the flows (the same picker as on the other analysis pages, described in 9.6; on this page each row also shows the IAT p-p), or use the arrows next to it to step to the previous or next flow. The chosen flow is in the page address (/timing?view=flow&flow=12) and the browser remembers it. The button with the arrow opens the flow's own page (for example the ST 2110-20 page for a video flow).
The Record button of the flow view records the chosen flow (29.5.1).
The time window, 60 s, 10 min, 1 h or 24 h, is chosen in the toolbar. The page keeps the last 60 s of every listed flow while it is open, so a flow you choose shows its last minute at once, and it gets the last 10 min of the flow from the probe, also right after a reload. The 1 h and 24 h windows come from the probe's timing history: one value per second for the last hour and one per minute for the last 24 hours, kept for up to 512 flows. They appear once the probe has answered. In the 24 h window each point is a minute: its shortest and longest inter-arrival time, the mean, and the highest percentile of its seconds. When the probe holds less than the window (a flow seen for 83 s) the texts say "last 83 s" instead of the full window. Reset starts the page's windows from now; the probe keeps its history.
The key values of the flow:
| Value | Meaning |
|---|---|
| Packet rate | Packets per second, with the nominal IAT (one second divided by the packet rate). |
| Mean IAT | The average time between two packets in the last second. |
| IAT peak-to-peak | The longest minus the shortest time between two packets in the last second, with both values below it. When the flow has a limit (by default ST 2110-30 and -31 audio: half the packet time, an error above the packet time), a tag says OK or High; the tooltip names the limits. |
| Std deviation | The standard deviation of the IAT over the time window, calculated from the IAT histogram. |
| 99.9th percentile | 99.9 % of the packets of the window arrived within this time after the packet before; the median (p50) and the 99th percentile below it. From the IAT histogram. |
| Peak IAT (window) | The longest time between two packets in the time window (peak hold), with the time it was seen. |
| Lost packets | RTP sequence gaps in the time window ("Sequence gaps"); continuity counter errors ("CC errors") for a transport stream. |
On a phone only Mean IAT, IAT peak-to-peak, the 99.9th percentile and Lost packets are shown, in two columns.
The figures from the histogram are as exact as its buckets: 1 ns below 1 µs, 10 ns from 1 to 10 µs, 100 ns from 10 to 100 µs, and so on. They never lie outside the shortest and longest IAT the probe measured.
Inter-arrival time per second is a trend over the time window, the newest second on the right:
- The shaded min-max band runs from the shortest to the longest IAT of each second.
- The mean line is the average IAT of each second, the p99.9 line the 99.9th percentile of each second.
- A dashed warning line marks the limit where the flow has one: the nominal IAT plus the allowed peak-to-peak. For ST 2110-30 and -31 audio with the default limits it lies at the packet time plus half of it: a second whose band reaches above it usually has an IAT peak-to-peak above the limit. The line is one-sided (it does not show packets that come too early); the full check is the OK or High tag on the IAT peak-to-peak key value.
- A second in which the flow sent no packets leaves a gap. Hover over the chart (or touch it) to read the values of a second.
The IAT histogram shows how many packets arrived after which time since the packet before, on a logarithmic time axis:
- Window sums the time window; Last second shows only the last second, with the peak hold as an outline: the highest count of each bar in any second of the window.
- Linear and Log switch the count axis; Log shows the rare long or short gaps next to the many normal ones.
- Dashed lines mark the nominal IAT, the p99.9 and the warning limit.
- Hover over a bar to read its time range and its packets.
A well-paced flow has a narrow peak at the nominal IAT. A long tail to the right means occasional congestion or queuing delays; a second peak means two traffic patterns, for example a routing change or load balancing. An ST 2110-20 video sender that pauses during the vertical blanking (a gapped sender) shows a second group of long gaps by design.
Under the charts, panels show the timing that matters for the flow's type. Most panels have a status word in their header: the ST 2110-21 and audio panels show the same word as the Timing verdict column of All flows, the media clock panel its state and the PCR panel the state of the PCR accuracy check. On a phone the panels are closed and show a one-line summary; tap a panel to open it.
ST 2110-21 sender timing (ST 2110-20 video):
- CINST max is the most packets the receiver buffer held at once; VRX peak is the fullest the virtual receive buffer got. Each is a bar with a mark for the limit of each sender type: Narrow (or Narrow linear for a linear sender) and Wide. The bar scale is logarithmic, so a Narrow limit of 8 and a Wide limit of 720 are both readable.
- The header says Narrow, Narrow linear or Wide for a compliant sender. A Wide sender is compliant (it needs a receiver with the larger buffer), so it is shown green. Not compliant means the sender is beyond the Wide limits; Measuring means the sender type is not known yet.
- Sender profile says whether the sender leaves the vertical blanking empty (gapped) or spreads its packets over the whole frame period (linear).
- CINST and VRX over time is a chart of the highest CINST and VRX of each second, the last 60 seconds, with the limits as dashed lines; a Wide limit far above the values is named under the chart instead of squeezing the chart. Last second gives the two values of the newest second. The chart collects the seconds while the panel is open.
- RTP jitter is the RFC 3550 interarrival jitter of the flow (see 21.4), colored when it is over the limit of the RTP jitter check.
- VRX underrun appears only when packets found the buffer empty (the sender was late), with the lowest level.
- TRO mean is the time from the start of each frame or field to its first packet, with its minimum, maximum and the default TRO of the format. It needs a PTP-locked receive time.
- Timestamps says whether the probe used NIC hardware timestamps; software timestamps make all timing values less exact.
The probe does not measure the ST 2110-21 sender timing of ST 2110-22 (JPEG XS) flows: the panel and the verdict say Not measured. For a raster that is not in the ST 2110-21 tables the Narrow limits do not apply: the panel shows the Wide limits, keeps the bars gray and says so.
RTP timing (an RTP flow that is no ST 2110 type, and a RIST flow that carries RTP): the RTP interarrival jitter of the last second and its highest value, colored by the limits of the RTP jitter check. A flow without RTP (SRT, plain UDP) has no such value, and a flow whose RTP clock rate is not known cannot be measured.
Audio packet timing (ST 2110-30 and -31): the RTP jitter, the packet time and its class, the mean packet interval (PIT) against the nominal one, the PIT peak-to-peak with its minimum and maximum, the packet rate (5 s average) against the nominal rate, and the RTP timestamp step (minimum, nominal, maximum). A step that is not constant is marked: the sender changes the number of samples per packet.
Ancillary data packets (ST 2110-40): the RTP packets per frame (or per field for interlaced video) and the RTP timestamp step with the frame rate it gives, both from the newest packets, and the ANC packets per RTP packet.
Media clock (every ST 2110 flow): the offset of the RTP timestamps to PTP (or to the probe's receive clock when the input has no PTP), the drift in ppm, the steps of the offset since the start, the grandmaster and the timestamp source. The header says Locked, Drifting, Measuring or Stale (no new estimate for 3 s). Open PTP page opens the PTP page.
PCR timing (MPTS and SPTS transport streams): one row per PCR PID with the programs that use it. The columns are:
| Column | Meaning |
|---|---|
| Interval | The time between two PCRs against the ETR 290 limit of the stream (2.3a, with the tolerance of the ETR 290 settings; without the ETR 290 data the interval is not judged). |
| Arrival jitter | How far the last PCR arrived from the time its own value predicts. |
| Overall jitter | The PCR against its arrival time (TR 101 290 PCR_OJ), peak to peak over the last second (21.4). |
| Freq. offset | How far the PCR clock runs from 27 MHz (PCR_FO), in ppm and in Hz; red above the limit of the PCR frequency offset check (810 Hz by default; the check that raises an incident is off until you enable it on a probe whose clock is locked to PTP, the colors are always shown). |
| Drift | How fast the frequency offset changes (PCR_DR), in ppm per second and in Hz per second; amber above the limit of the PCR drift check (0.075 Hz/s by default). |
| Accuracy | The PCR accuracy of ETR 290 2.4: the last value and the worst. |
The line above the table gives the worst accuracy of the stream against its limit, and the header the state of the 2.4 check (for example Accuracy OK). The accuracy is measured for constant bitrate streams only; for a variable bitrate stream the panel says so. The offset and the drift need some seconds of PCRs before they appear (10 s and 20 s), and a value disappears 2 seconds after the last PCR. Click a row (or press Enter or Space on it) to open the PCR jitter histogram of the last minute and the overall jitter trend of that PID; the trend is kept in the browser while you look at it. Open MPTS Detail opens the MPTS page of the stream that you used last (PIDs, Services or Tables).
MDI (transport streams): the Media Delivery Index of RFC 4445, below the PCR timing panel. The bar shows the delay factor of the last second in ms with the warning and error limits of the MDI delay factor check as marks. The values next to it are the media loss rate (lost TS packets in the last second), the largest delay factor since the start and, for a variable bitrate stream, a note that the delay factor is shown but not judged. The chart shows the last minute: the delay factor as a line going up and the lost packets as red bars going down. The header says the state of the stream's MDI against the limits; before the stream has sent for a second the panel says that MDI is measured once it has.
For a PTP flow, a panel links to the PTP page, where its clocks are shown.
Heat map over time is a section that is closed by default; click its title to open it (the page remembers the choice). It shows the IAT histogram of the last second at the top and, below it, the last 120 seconds as a heat map, so a change in the distribution over time stands out.
21.3 Compare
The Compare view puts up to four flows side by side. Choose them with Compare in the toolbar (the multiple mode of the flow picker, 9.6) or tick them in All flows; each flow gets a color and a marker shape, so the flows are told apart without the color.
- Compared value chooses what the trend chart draws per second: IAT p-p, Mean IAT, Std dev, p99.9 or Max IAT, one line per flow over the time window.
- IAT histograms draw the histogram of each flow as one line (the share of the flow's packets per bar), so a flow with a wider spread or a second peak stands out.
- The table lists, per flow, the mean IAT, the IAT p-p, the standard deviation, the 99.9th percentile, the Peak p-p (the largest IAT p-p of the time window), the minimum and maximum IAT, the packets per second, the lost packets of the window and the timing verdict. An ST 2110-20 flow also shows its highest CINST and VRX of the last second, and an RTP flow its RTP jitter. A flow over a limit shows a warning icon.
The time window and Pause work as in the Flow view. A flow that has stopped sending shows dashes.
21.4 Measurements
Every value of the page, how it is measured and what it tells you. The probe measures once per second; limits come from the timing checks (31.4.4.13).
| Value | Definition |
|---|---|
| IAT | The time between two successive packets of one flow, from the arrival timestamps of the network card (or of the software when the card has none). |
| IAT peak-to-peak (p-p) | The longest IAT of the second minus the shortest. A paced sender has a small value; bursts and pauses make it large. |
| Standard deviation | The spread of the IAT around its mean. The probe calculates it exactly for each second; over a time window it is read from the histogram. |
| Percentiles (p50, p95, p99, p99.9) | The IAT within which that share of the packets arrived: with p99.9 only one packet in a thousand took longer. Read from the histogram, so a value is the upper end of its bar: never below the true value and at most one bar width above it (the bars are 1 ns wide below 1 us and always under 1 % wide above it, up to 1 s). Never outside the measured shortest and longest IAT. |
| Window and history | The window is 60 s, 10 min, 1 h or 24 h. The probe keeps one value per second for the last hour and one per minute for the last 24 hours for up to 512 flows. A minute holds its shortest and longest IAT, the mean and the highest percentile of its seconds. |
| RTP jitter (RFC 3550) | The smoothed change of the transit time between packets, as a receiver reports it: for each packet the difference between the arrival spacing and the RTP timestamp spacing, averaged with a weight of 1/16. In us; the highest value of the second is judged. For video it is measured on the first packet of each frame (all packets of a frame share one timestamp), so a video flow gets one value per frame. Needs RTP with a known clock rate. |
| CINST max, VRX peak | The highest packet count the ST 2110-21 network compatibility model (CINST) and the virtual receive buffer (VRX) reached in the last second, against the limits of the sender types. |
| MDI delay factor (DF) | RFC 4445. The probe follows a virtual buffer that fills with the arriving bytes and empties at the stream's own rate in that second. DF is the largest spread of its level divided by the rate, in ms: the receive buffer a decoder needs to ride out the arrival pattern. The limits default to a warning above 150 ms and an error above 300 ms. The DF of a stream whose rate changes by more than 2 % from one second to the next (variable bitrate) is marked VBR for 10 s, shown but not judged. |
| MDI media loss rate (MLR) | RFC 4445. The TS packets lost per second: the RTP sequence gaps times the TS packets per datagram, or the packets that jumps of the continuity counters skipped, whichever is larger. Over plain UDP (no RTP) only the continuity counters tell, so the value is a lower bound: it cannot see a loss of a multiple of 16 packets of a PID, or of packets of a PID that no longer sends. Any loss is an error by default. |
| PCR arrival jitter | The arrival of the last PCR against the time that its value and the previous PCR predict. |
| PCR overall jitter (PCR_OJ) | TR 101 290. The probe fits a straight line through the PCR time against the arrival time of the PCRs of the last 30 s, taking its slope from the least delayed PCR of each second so that datagram packing and queueing do not tilt it. The overall jitter is the distance of each PCR from that line (positive: late); the page shows the peak to peak of the last second and a histogram of the last minute (0 to 50 ms). |
| PCR frequency offset (PCR_FO) | TR 101 290. The slope of that line: how fast the PCR clock runs compared with the probe's clock, in ppm (1 ppm is 27 Hz at 27 MHz) and in Hz. The standard allows 810 Hz (30 ppm). It needs 10 s of PCRs. The reference is the probe's own clock, so the probe's oscillator error is part of the value. |
| PCR drift (PCR_DR) | TR 101 290. The change of the frequency offset per second over the last 60 s, in ppm/s and Hz/s. The limit of the broadcast profile is 0.075 Hz/s. It needs 20 s of PCRs. |
| PCR accuracy (PCR_AC) | ETR 290 2.4. The error of the PCR against the constant bitrate of the stream, the largest of the current and the previous second, per PID. Not measured for a variable bitrate stream. |
A measurement restarts after a break of the PCR (discontinuity indicator, a step of more than a second, a PCR more than 100 ms from its arrival, or no PCR for a second).
22 Video
The Video page provides a professional video workbench with broadcast-standard analysis tools for real-time video monitoring. It supports both uncompressed ST 2110-20 video and compressed MPTS video (H.264, H.265, MPEG-2).
This page requires the Video license option.
22.1 Selecting a Service
Pick the service with the Service field at the top of the page (or press Ctrl+K). It is the same picker as on the AV Monitor: services are grouped per input and transport stream, ST 2110-20 senders have their own group, and you can search by name, provider, TS id or address. Radio and data services are listed dimmed with "no video"; they cannot be picked here. The arrows next to the field step to the previous or next service.
The selected service is shared with the Audio page and the AV Monitor: switching pages keeps it. It is also in the page address (/video?svc=...), so a link or a reload opens the same service, and the browser remembers it. With nothing selected the page shows your recent services. A service reached by a link that has no video says so and offers Open in Audio. AV Monitor in the toolbar opens the same service there with sound.
Switching services does not restart the stream for others. A scrambled service shows the SCRAMBLED picture.
The Record button in the toolbar records the service you watch (29.5.1).
22.2 Layout and Overlays
| Control | Description |
|---|---|
| Focus | The picture with three scopes under it: Waveform, Vectorscope and a third scope you choose from its title (Histogram, Diamond or CIE). |
| Grid | Picture and Waveform side by side, Vectorscope, Diamond and CIE under them. |
| Safe areas | Action safe (90 %) and title safe (80 %) frames on the picture; each one can be switched in the more menu. |
| False color | Shows the luma in bands: purple below 5 %, blue to 10 %, green at mid gray (40-45 %), pink at skin tones (55-60 %), yellow at 90-95 % and red above 95 %. For an ST 2110-20 flow and an MPTS program with MPEG-2, H.264 or HEVC video the button becomes a choice of Luma and Gamut drawn from the values the probe has (see 24.3). |
| Inspect, Freeze | ST 2110-20 flows (the 10-bit values) and MPTS programs with MPEG-2, H.264 or HEVC video (the decoded values at the stream's bit depth): the values of a pixel, with the Pixel panel above the properties (see 24.3 and 24.3.1). |
The picture has a name strip with a tally (green: clean; amber or red: errors such as CC errors or RTP gaps, or no packets) and the clock. On a laptop screen the properties move under the scopes; on a phone one scope is shown at a time (tabs) and the property sections fold to a one-line summary. The layout and the scope choices are remembered in this browser.
22.2.1 Video Properties
The Video properties panel lists what the probe knows about the selected video, each value with a small tag saying where it comes from (PMT, SPS, VUI, SEI, RTP, SDP, measured, calc):
- MPTS services: codec, profile and level, resolution, frame rate (with fields per second for interlaced video), scan and field order, chroma and bit depth; color primaries, transfer, matrix and range with an SDR, HDR PQ or HDR HLG tag; sample and display aspect; bitrate now, average, peak and minimum with a 60 s graph and a CBR or VBR tag; PID, stream type, program and CC errors.
- ST 2110-20 senders: essence, size, rate, scan (including PsF), sampling, depth and pgroup from the RTP headers; packing, sender type, packets per frame, payload, TRO and Cinst / VRX; media clock, grandmaster, RTP payload type, flow address and sequence gaps. Colorimetry, TCS and range come only from an SDP and say "needs SDP"; they are never guessed from the picture size.
- JPEG XS (MPTS and ST 2110-22): profile, level and sublevel from the picture header (tag PIH); a JPEG XS section with whether the probe can decode the stream (and why not, for example RGB JPEG XS), the declared bitrate next to the measured one, the compression ratio, the time code, slice height and components; for an ST 2110-22 flow also the packetization, packets per frame and the three JPEG XS error counters. An MPTS stream adds the JPEG XS video descriptor and the anc_data_descriptor of the PMT (the latter as hex bytes and as text); a buffer model other than 2 is marked as a warning. JPEG XS is all intra, so there is no GOP section. The size the probe decodes at appears as Decoded at while the service plays.
Coding and GOP. For an MPTS service with MPEG-2, H.264 or HEVC video the probe also reads how the picture is coded, in the words MediaInfo users know. These values appear in a Coding section and are added to the GOP section:
- Coding: the codec ID (the PMT stream type), slices per frame or field, tiles, wavefront and DPB size (HEVC), the store method of field pictures, the quantization or scaling matrix (default, flat or custom), the bitrate mode that the stream signals (constant or variable) with its nominal bitrate, and the bitrate mode the probe measures (constant or variable, with the spread over the last seconds). Facts that other sections already show (reference frames, scan and field order, color range, mastering display and light levels) are not repeated here; an HDR format row appears only when it says more than the Color tag, for example HDR10.
- GOP: the GOP type (open or closed), whether B pictures are used, and the number of frames or fields analyzed. The GOP structure itself (M and N) is the Structure row above them.
A row appears only when the value is known. The values build up over the first GOPs: right after you select a service the Coding section says collecting, and Measured bitrate shows Measuring (n of 5 s) during its first 5 seconds. A stream whose pictures do not line up with the PES packets can need about 20 seconds before the rows appear. The probe collects these details only for a service you look at (the Video page or the AV Monitor); they stop when you select another service or leave the page.
A value the probe cannot know yet says why, for example "after start" (it is read once the service plays) or "not read yet". The decoder values appear for the service that plays. The copy button puts the panel on the clipboard as text, for a ticket.
22.3 Video Player
The video player shows a live decoded view of the selected video source.
The live video is encoded to H.264 by a hardware encoder when the machine has one (the video encoder of the graphics processor, for example through VAAPI on Linux), otherwise in software (OpenH264). On Linux the service user needs the render group for VAAPI. This also applies to the full-screen video and the AV Monitor.
Live MPTS video is paced by its timestamps for smooth playback, so it trails the live signal by up to about 0.7 s (statistical multiplexes deliver frames unevenly).
Several people can watch live video at the same time, on this page, the AV Monitor and the MPTS player. People who watch the same source (the same ST 2110 flow, or the same MPTS program) share one picture, and one of them closing it does not stop it for the others. Different sources are shown at the same time up to the Live Pictures at Once setting (Settings > System, admin only, default 2). When that many different sources are already shown, the player says which ones; close one of them or watch one of those sources. The player then waits and shows the picture as soon as a slot is free (when a viewer closes one of the pictures); Retry now asks again at once. Waiting stops when you pick another source, leave the page or the tab is hidden, and starts again when the tab is visible. Stream mode in the AV Monitor shows one source at a time for all viewers. The service picker of the AV Monitor and the Video page tags a service that is being watched with Live: N viewers (Live: 1 viewer for one); the number includes you when you watch it yourself, and the probe never shows who the viewers are.
22.3.1 Video Overlay
Press I on the keyboard to cycle through overlay modes:
- Hidden - No overlay, full video view.
- Minimal - A single bar showing resolution, frame rate, codec, and bitrate.
- Full - Two information cards:
- SOURCE - Original stream information (type, resolution, frame rate, color space, sampling, bit depth, bitrate).
- STREAMING - Encoding output (resolution, frame rate, bitrate, compression, frames encoded/dropped).
22.3.2 Playback Behavior
- The player automatically adjusts to the stream's native frame rate.
- A 2-frame playout buffer absorbs network jitter.
- If the player falls more than 160 ms behind, it automatically resyncs.
- When the browser tab is hidden, playback pauses to save bandwidth and CPU.
22.4 Waveform Monitor
Displays the luminance and chrominance distribution across each column of the video frame, similar to a broadcast waveform monitor.
Display modes:
| Mode | Description |
|---|---|
| Y | Luma only (green phosphor). |
| CbCr | Chroma only (Cb in blue, Cr in red). |
| YCbCr | All three channels overlaid. |
| RGB | Red, green, and blue channels (converted using BT.709 matrix). |
Scale modes:
| Scale | Description |
|---|---|
| IRE | Broadcast standard scale (-7 to 109 IRE). Black at 0 IRE, white at 100 IRE. |
| Value | Raw digital values (0-255). Studio swing boundaries shown at 16 (black) and 235 (white). |
The waveform uses phosphor-style rendering with logarithmic intensity mapping. Bright spots indicate high pixel density at that level.
22.5 Vectorscope
Displays the color distribution of the video frame as a 2D plot of Cb (horizontal) vs Cr (vertical) values.
Reference elements:
- Color targets - Six corner brackets at the standard 75% color bar positions: Red, Green, Blue, Cyan, Magenta, Yellow.
- Skin tone line - Diagonal line at 123 degrees indicating the expected direction for natural skin tones.
- Reference circles - Concentric circles at 25%, 50%, 75%, and 100% of maximum chroma.
- Crosshairs - Center point represents neutral gray (no color).
A well-balanced image shows most data concentrated near the center. Color bars produce six distinct clusters at the target positions.
22.6 Histogram
Displays the distribution of luminance values across the entire frame:
- 256 bins showing how many pixels have each brightness level.
- Logarithmic height scaling to make small peaks visible alongside dominant values.
- Reference lines at studio swing boundaries (16 = black, 128 = mid-gray, 235 = white).
- RGB mode available when source data supports it.
A well-exposed image shows a distribution that uses the full range without clipping at either end.
23 Audio
The Audio page provides a universal audio workbench for monitoring and analyzing audio from any source type (ST 2110-30, MPTS, SRT, RIST).
This page requires the Audio license option.
23.1 Selecting a Service and Audio
Pick the service with the Service field at the top of the page (or press Ctrl+K). It is the same picker as on the AV Monitor: services are grouped per input and transport stream, ST 2110 audio flows and senders have their own group, and you can search by name, provider, TS id or address. The arrows next to the field step to the previous or next service. Scrambled services are listed dimmed and cannot be picked (a lock marks them); a data service says it carries no audio.
The Record button in the toolbar records the audio of the service; for an ST 2110 sender it records only its audio flow (29.5.1).
The selected service is shared with the Video page and the AV Monitor: switching pages keeps it. It is also in the page address (/audio?svc=...), so a link or a reload opens the same service, and the browser remembers it. With nothing selected the page shows your recent services.
The second field depends on the service:
- MPTS service: the Audio track menu lists the tracks of the program (language, codec, channel layout, sample rate, bitrate, PID; tags for the default track and audio description). No audio stops the sound and the meters. SMPTE 302M pairs, Dolby E programs and AC-4 are described in section 23.7.
- ST 2110-30 / AES67 audio flow: the Listen field lists the named channel groups of the flow (for example PGM 1-2, 5.1) and its pairs; picking one sets the channel switches under the meters, which also accept any other choice of channels. For a flow with a 5.1 group the menu also has the 5.1 channel order (LRC or LCR, see 23.6.1).
Playback starts automatically when a service is picked.
23.2 Playback Controls
| Control | Description |
|---|---|
| Play/Stop | Toggle audio playback for the selected source. |
| Volume | Adjust playback volume (0-100%). |
| Dim | Reduce volume by 10 dB for quick level comparison. |
| Mute | Mute audio output completely. |
Audio playback stops when you go to another page.
23.2.1 Audio Transfer: AAC or PCM
The probe sends the audio you listen to AAC compressed when your browser can decode it: about 200 kbit/s per stereo pair instead of about 1.5 Mbit/s as 16-bit PCM, which matters on a VPN or a slow remote link. A small tag next to the controls shows the format and the measured rate, for example AAC 198 kbit/s or PCM 1470 kbit/s; hover over it for the details.
The tag. AAC or PCM, followed by the rate measured over the last seconds. When the audio is PCM although AAC was possible, hover over the tag: the tooltip names the reason (see below).
Start delay. AAC adds about 0.1 s before the first sound on an ST 2110-30 stream (PCM starts after about 0.02 s). On an MPTS service you do not notice a difference: starting the audio decoder takes about 0.6 s either way and the encoder fits inside that time.
Bit rate. 192 kbit/s per stereo pair (the LFE channel counts as half a channel), so a stereo stream sends about 200 kbit/s including overhead and a 5.1 stream about 530 kbit/s. The administrator can change it with the command line option
--audio-aac-kbps(32 to 320).
- Lip sync is the same in both formats. Every piece of AAC audio carries the time stamp of its first sample on the same clock as PCM, and the browser plays it at that time. The compressed audio starts a little later (about 0.1 s), but it stays in step with the picture, because the delay is already in the time stamps.
- Meters. The level meters and the loudness meter are measured on the probe before encoding, so they read the same in both formats. The goniometer, the phase display and the browser's own level bars show the decoded signal, which can differ from PCM by a fraction of a dB on peaks.
- PCM instead of AAC. The audio comes as 16-bit PCM, and plays on without a break, when the browser cannot decode AAC (some Firefox versions, or a page opened over plain HTTP), when decoding fails in the browser, for a stream AAC cannot carry (more than 8 channels, so a 16-channel stream always comes as PCM, or an unusual sample rate), or when the probe already runs its maximum of 16 audio encoders or their processor core is busy. The tooltip of the tag names the reason.
- Scrambled (encrypted) audio. A service whose audio is scrambled cannot be decoded by the probe, so it is refused with the message "audio is scrambled (encrypted)" in both formats; it is also logged as a warning. Services that have no sound at all in the player are often of this kind.
23.3 Level Meters
Per-channel RMS and peak levels in dBFS, updated in real-time. Each meter bar shows:
- Bar - The sample peak in dBFS (digital peak program meter, IEC 60268-18): instant attack, release 20 dB in 1.7 s.
- Peak hold marker - Holds 2 s, then falls at the same rate.
- Alignment tick - At -18 dBFS (EBU R 68).
- Colors - Green below -18 dBFS, yellow up to -9 dBFS, red above.
- Readout - The held peak. RMS is shown in the tooltip.
The meter shows exactly the number of channels the stream carries (maximum 64). Channel numbers are shown up to 32 channels, dB values up to 16 channels.
23.3.1 Per-Channel Solo
When playback is active, the meters provide interactive channel selection:
- Click a channel number under the meters (ST 2110-30 / AES67 flows) to switch that channel on or off for the speakers. Any combination of channels can be heard; more than two are downmixed to stereo. The Listen field shows what is selected, and Listen to channels 1 and 2 in the more menu returns to the default.
- Click any meter bar of an MPTS service to switch to the stereo pair containing that channel.
- Use the pair navigation arrows to step through pairs.
- Click a channel within the current pair to solo it - only that channel will be audible.
- Double-click any channel to immediately switch to its pair and solo it.
When solo is active, the soloed channel's meter glows amber, all other channels dim, and an amber "SOLO" badge appears.
23.3.2 Channel Groups
Channel groups define the logical structure of an audio flow's channels. A single flow can carry up to 64 channels, and groups organize them into meaningful units.
Supported layouts:
| Layout | Channels | Speaker Labels |
|---|---|---|
| Mono | 1 | M |
| Stereo | 2 | L, R |
| 5.1 | 6 | L, R, C, LFE, Ls, Rs |
| 7.1 | 8 | L, R, C, LFE, Ls, Rs, Lrs, Rrs |
| 7.1.4 | 12 | L, R, C, LFE, Ls, Rs, Lrs, Rrs, TFL, TFR, TRL, TRR |
23.4 Goniometer
The goniometer displays stereo phase correlation as a Lissajous (X-Y) plot. It requires active playback.
- A vertical line indicates mono-compatible content.
- A diagonal or circular pattern indicates stereo content with spatial information.
- A horizontal line indicates out-of-phase content (potential phase cancellation).
23.5 Loudness Meter
EBU R128 integrated loudness measurement displaying:
| Measurement | Description |
|---|---|
| M (Momentary) | 400 ms sliding window loudness in LUFS. |
| S (Short-term) | 3-second sliding window loudness in LUFS. |
| I (Integrated) | Program-integrated loudness since the last reset. |
| LRA | Loudness Range - the difference between soft and loud passages in LU. |
23.5.1 Loudness Radar
The Loudness Radar is a polar display showing loudness history over time. A sweeping clock hand rotates around the display, leaving traces behind it:
- Momentary trace (M, blue) - 400 ms momentary loudness as a radial plot.
- Short-term trace (S, orange) - 3-second loudness as an overlay line.
- Integrated ring (I, green dashed) - Current integrated loudness as a static ring.
- Reference rings - Concentric circles at key levels (-41, -23, -14, -9 LUFS). The -23 LUFS ring (EBU R128 target) is highlighted.
Controls:
- Time window - 1, 2, 5, 15, 30, or 60 minutes per rotation.
- Scale mode - LUFS, +9 (EBU +9), or +18 (EBU +18).
23.5.2 Loudness checks on the Audio page
For an ST 2110-30, ST 2110-31 or AES67 track, and for the audio of an MPTS program whose loudness is measured (Settings > Checks, Loudness standard, Measured), the meters and the loudness history draw the target band and the Too loud / Too quiet limits of the loudness standard in force for the flow (31.4.4.29), instead of a fixed -23 LUFS. The value line says how far the short-term loudness is from the target, for example +7.4 LU, too loud 32 s. The history offers 10 min and 1 h.
When a loudness alarm of the track is open, a banner names it ("Too loud for 1 min 38 s") with a link to the incident. The Loudness checks panel lists every loudness check of the track with its state: open (since when), still N s to raise (pending: over the limit, not an alarm yet), OK, N of M in the window (true peak and momentary peaks), off, or why it does not apply. Edit opens Settings > Checks. When the measurement is off for the flow the panel says so, with a link to the overrides; for an MPTS program that is not measured it says how to switch it on. An incident of an MPTS program names the program, its service and the audio PID ("program 13001 ORF 2: PID 257 deu AC-3 5.1: short-term ...").
On the Incidents page a loudness incident has a Trace button: the detail shows the highest (or lowest) short-term loudness, the integrated loudness over the episode and the true peak maximum, and draws short-term, momentary and integrated loudness with the target band, the limit and the raise from 60 s before the start to the end. A Silence incident draws the highest channel peak against the silence level.
23.6 Room Meter (Surround Spatial Display)
The Room Meter provides a top-down spatial visualization for surround sound content (5.1, 7.1, or 7.1.4):
- Speaker dots - Positioned in standard ITU locations. Each dot grows and brightens proportionally to channel level.
- RMS Polygon - Connects all non-LFE speakers; vertex distance from center is proportional to RMS level. A symmetric polygon indicates a balanced mix.
- Dominance Indicator (DMI) - A crosshair showing the amplitude-weighted center of gravity. A centered DMI indicates balanced content.
- Center difference - The level difference between the center channel and L/R average.
- Phase correlation bars - Per-pair correlation indicators (L-R, Ls-Rs, Lrs-Rrs).
23.6.1 Surround vectorscope
For a 5.1 or 7.1 group (ST 2110-30, AES67) and for a decoded 5.1 or 7.1 track of an MPTS program, the Vectorscope panel shows how the speakers work together. It has a row of its own under the meters on the Audio page (surround view), and it replaces the phase scope in the Phase panel of the AV Monitor when you pick Vectorscope in the title of that panel (the choice is remembered in this browser; a stereo track keeps the phase scope and has no switch). The probe computes the vectorscope only while a page shows it.
- The scope. One point per speaker, in the direction of the speaker, at its level: the rings are 0, -10 ... -50 dBFS from the outside in. The points are joined by one edge per neighboring pair. An edge bows outward while the two speakers are correlated and inward while they are out of phase. On every edge a white dot shows where a listener hears the pair between the two speakers, and the bar around it how wide that image is. The LFE is the dotted ring in the middle, with its level.
- The pairs. Next to the scope every pair has a bar from -1 to +1, its value and a word: in phase above +0.5, wide from +0.5 to -0.3, out of phase below -0.3. Value and word are always written, the colors (green, amber, red) only repeat them. A pair that is too quiet for phase to mean anything shows - and no dot. A badge in the panel title names a pair that is out of phase, for example R-Rs out of phase; a pair whose speakers carry the same signal with one of them inverted is the usual cause.
- A flow without a 5.1 or 7.1 group in Settings > Audio channels has no vectorscope (the panel says so).
5.1 channel order. Sources of 6 channels do not always use the same order. In the Listen menu of the Audio page and in the Audio track menu of the AV Monitor, a block 5.1 channel order offers LRC (L R C LFE Ls Rs, the SMPTE order and the default) and LCR (L C R Ls Rs LFE, the film order). Your choice is remembered in this browser and applies to everything you see and hear of the audio: the vectorscope, the channel names under the meters and the stereo downmix you listen to. The block is shown for ST 2110 audio with a 5.1 group only; an MPTS track uses the order of its decoder.
23.7 Dolby Decoding
The probe decodes AC-3 (Dolby Digital), E-AC-3 (Dolby Digital Plus) and Dolby E carried in SMPTE 302M, and shows AC-4. You see this in the Audio track menu of the AV Monitor and the Audio page, in the meters and in the Audio properties panel.
AC-3 and E-AC-3. The tracks play, meter and measure loudness at the level they are encoded with: the dynamic range control words of the stream are not applied (the properties say DRC off for metering). The Dialnorm row puts the dialogue level the stream announces next to the measured integrated loudness ("-23 dB - matches I -23.1", or the difference in LU with a warning when they are more than 2 LU apart). For E-AC-3 with Dolby Atmos the properties show Atmos (JOC): core decoded: you hear and meter the channel core, the objects are not rendered. When a service switches its layout (for example from 5.1 to 2.0 between programs), the meters and the channel view follow the new channel count and the loudness measurement starts again.
SMPTE 302M. A 302M PID carries up to four AES3 pairs, each with PCM or SMPTE 337 data such as Dolby E. Once the probe has looked at the pairs, the Audio track menu lists the PID itself (PID 0x0101 - all pairs, every channel on the meters) and under it every pair as a track of its own, named after what it carries, for example PID 0x0101 pair 1 - PCM or PID 0x0101 pair 2 - Dolby E 5.1+2. The second line gives the sample width, the Dolby E frame rate, the programs and the channels of the pair. The SMPTE 302M section of the Audio properties shows the pair, what it carries, the sample width (and the width of the Dolby E words when it is narrower), the Dolby E program configuration and frame rate. A pair with other SMPTE 338 data is listed dimmed with NO DECODER.
Dolby E programs. A Dolby E pair can carry several programs, for example 5.1+2 (a 5.1 program and a stereo program). The meters show every decoded channel, grouped and labeled per program (P1 5.1, P2 2.0). What you hear, the loudness and the AAC audio are one program: program 1 when the track starts. To switch, open the Audio track menu and pick the program under Dolby E program (for a whole PID the list is called Program heard and also offers its PCM pairs); the field then names the program heard (for example P2 2.0). On the AV Monitor the program plays in its own layout (a 5.1 program as 5.1); on the Audio page picking a program switches its channels on under the meters.
Up to 16 channels. The meters and the channel view show up to 16 channels at full size, for example a 302M PID with two Dolby E pairs of 8 channels. Listening, the AAC audio and the loudness use the selected program or pair, at most 8 channels.
ST 2110-30 (AES67) audio. A pair of an ST 2110-30 flow that carries SMPTE 337 data (Dolby E, AC-3 or E-AC-3) instead of PCM is recognized by the probe. The flow shows the pair with its content (for example Dolby E 5.1) instead of full-scale noise on the meters. When you listen to such a flow on the Audio page or the AV Monitor, the probe decodes the pair for you: the meters show a second set of levels for the decoded channels (the badge Decoded n ch), and for Dolby E with several programs the program picker works as described above. The decoder runs only while someone listens.
ST 2110-31 (AES3) audio. ST 2110-31 flows carry the whole AES3 signal, so the probe also reads the channel status of every pair. They are listed as ST 2110-31 Audio and work like ST 2110-30 on the meters, in listening and for Dolby E, AC-3 and E-AC-3 pairs. A pair whose channel status says non-audio is treated as data: it has no levels and is not played as noise, also when it carries no Dolby E the probe can read (it is then shown as data). Dolby E without the non-audio flag is still recognized from its content. See section 15.9 for the AES3 checks.
Decoder limit. The probe decodes at most four audio tracks, pairs or data pairs at the same time. A fifth request is refused (Refused with the reason too many decoders) until a listener stops.
AC-4. The probe has no AC-4 decoder. AC-4 tracks are listed with their language and channel mode and the tag NO DECODER, but cannot be picked (the tooltip says No decoder for AC-4). A service picks another track to start with; a service with only AC-4 starts without sound.
24 AV Monitor
The AV Monitor page provides combined audio and video monitoring in a single view, designed for operators who need to verify audio-visual synchronization and overall program quality.
This page requires the Video or the Audio license option and shows only the part that is included.
The page displays a video preview alongside audio level meters and timing information, allowing operators to:
- Verify lip-sync between video and audio.
- Monitor audio levels while watching video content.
- Detect frozen video or silent audio conditions.
- Confirm that the correct audio is associated with the correct video.
The page plays one service at a time: an MPTS program, an ST 2110-20 video sender or the probe's test stream. The toolbar at the top holds everything you need to choose and play it.
The Record button in the toolbar records the service you watch: for an MPTS program that program (29.5.2), for an ST 2110 sender its video flow and the audio flow that belongs to it (29.5.1).
Service. The Service field shows the service that plays: a status dot (green: playing, red: no packets, amber: scrambled), its name, and below it the provider, the TS (with its tsid), the input and the video format. Click it, or press Ctrl+K anywhere on the page, to open the service list:
- Type to search on name, provider, TS id, program number, address or alias; the list filters as you type.
- The tabs All, TV, Radio, ST 2110, Scrambled and Data show how many services each has (a tab with none is hidden).
- An ST 2110 row shows the brand of the sender after its name ("Nevion - 10.10.4.11"): the manufacturer of the source MAC address (IEEE OUI). It is left out when the MAC belongs to a router.
- Services are grouped per input and TS (the header shows the tsid, the number of services and the rate), with the ST 2110-20 senders and the test stream in their own groups. Click a group header to fold it; a folded group names its first services.
- Up / Down move the highlight, Enter plays the highlighted service and Esc closes the list.
- Scrambled and data services are listed dimmed and cannot be played; hover over one to see why. Radio services play their audio with the meters.
- Each MPTS service shows how many audio tracks its PMT lists (speaker with a count, for example 4); a TV service without audio says no audio. ST 2110 senders have no count: any audio flow can be played with them. Their line says which audio they start with (audio: CAM 1 Audio, with (paired) for a manual pair), or how many audio flows match them (2 matched, none matched).
- The filter bar also applies here: the footer says how many services it hides, and show all lists them anyway.
The previous and next buttons in the field step to the service before or after the current one, skipping the ones that cannot play.
Picking a service starts its picture and its default audio track together. The service stays selected when you reload the page: it is kept in the address (/av-monitor?svc=..., so you can bookmark or share the link) and in the browser. If the service is not in the stream list (any more), the page says so and starts it by itself as soon as it appears; the services you played last are offered below.
Audio track. Next to the Service field, the Audio track menu lists only the audio of the chosen service:
- MPTS: the tracks from the program's PMT, the default track first, each with its language, codec, channel layout, sample rate, bitrate and PID, and tags for the language, AD (audio description) and DEFAULT. Audio of another service is not offered: it runs on another clock and can never be in sync with this picture.
- ST 2110-20: the ST 2110-30 / AES67 audio flows on the probe, in two groups (see Matched audio below).
- Always No audio: picture only, the meters stop.
Choosing a track switches only the sound; the picture keeps playing. If the probe cannot play a track (for example when four tracks are already being decoded), the field shows Refused with the reason. SMPTE 302M pairs, Dolby E programs and AC-4 tracks (listed, not playable) are described in section 23.7.
Matched audio (ST 2110). An ST 2110 video and its audio are separate flows, so the page works out which audio flows belong with the video. The menu shows them first under Matched with video, the most likely on top, each with tags that say why:
- PAIRED: a manual pair saved in the probe configuration (see below). Always first.
- SAME SENDER: sent by the same device (the same source IP address, or the same source MAC address) as the video. A MAC address with many senders behind it (a router, on a routed network) is not used. A device that sends several videos, such as a gateway or a playout server, does not match all its audio this way: there the tag is shown, but the flow is only listed under the other flows unless its name also matches.
- ALIAS: the alias of the audio flow starts with the alias of the video, for example CAM 1 and CAM 1 Audio 9-16 (words such as Video and Audio at the end are left out; CAM 1 does not match CAM 12, CAM A does not match CAM B Audio).
- NEARBY: the destination address is next to the video's (for example 239.10.1.11 and 239.10.1.12). This only decides between flows that match for one of the reasons above.
The other audio flows follow under Other audio flows; any of them can still be played with the video. Picking a video starts its manual pair; without a pair, the best match when there is one clear winner; otherwise No audio, with the candidates in the menu. When you choose an audio flow yourself, the video starts with that one next time on this browser (a manual pair still comes first).
CLOCK DIFFERS on a flow means the probe measured that it does not run on the same media clock (PTP grandmaster) as the video: its timestamps drift away from the video's, or both are on the PTP time line and more than 50 ms apart. Such audio can be played, but never in lip sync with this picture; hover over the tag for the details. The field shows clock differs too. No tag is shown while the probe cannot tell yet (a flow that just started, or two flows on different inputs without PTP).
Pairing. Admin and NOC users can save which audio belongs with a video when the rules cannot tell. The menu footer names the flow it acts on: the highlighted one, which is the one that plays when the menu opens. So pick an audio flow, listen, open the menu again and click Pair video with audio (with the keyboard: highlight a flow, press Tab to reach the footer, then Enter). On the paired flow (or on No audio) the footer offers Remove the pair. A video has at most one pair; pairing it again replaces the old pair. Pairs are saved in the probe configuration, so every user and every open page sees them at once. Other users see which flow is paired, without the action.
Controls. Raw / Stream chooses how the picture is sent: Raw decodes it in the browser (the default, shortest delay), Stream gets H.264 from the probe (less bandwidth on a remote link). Both have lip sync (see below). The tag before them shows how the sound is sent, AAC (the default) or PCM, with its measured rate (see 23.2.1). The speaker button mutes and unmutes, the slider sets the volume. Test stream plays the probe's generated pattern and 1 kHz tone (the arrow next to it chooses the frame rate); press it again to go back to the service you had. The more menu (three dots) can also list the services outside the filter bar, show the Summary panel (hidden on this page; the s key toggles it too) and open the MPTS or flow detail of the service.
On a narrower window the toolbar takes two rows; on a phone the fields take the full width and the test stream moves into the more menu.
The picture. Below the toolbar the picture is shown as on a monitor:
- The strip at the bottom of the picture (UMD) shows the tally light, the service name, the picture format and codec, the audio track and the time of day. When something is wrong it also says what (for example CC errors or silent 4 s).
- The top left shows how the picture is sent (RAW or STREAM), how long a frame waits in the browser before it is shown (in ms; the delay of the probe and the network is not included) and the lip sync offset (AV +N ms, see below).
- The frame around the picture and the tally light give the state at a glance: green while the service plays clean; amber on a warning (scrambled, continuity (CC) errors or RTP sequence gaps in the last 10 seconds, a PCR interval over 100 ms, a check warning on the flow, or the audio silent for 3 seconds or more on every channel); red when the flow sends no packets or a check reports an error. Hover over the tally light to see the reason.
- Level bars on the picture (more menu) adds thin level bars on the right edge of the picture. They are on by default on a phone, where they take the place of the level meter panel.
Audio tools. Next to the picture (under it on a narrower window) are three instrument panels. They keep their dark background in the light theme too.
- Audio levels: one bar per channel in dBFS with the usual program meter timing (instant rise, 20 dB fall in 1.7 s). The number above each bar is the highest peak of the last 2 seconds (red from -1 dBFS); the dashed mark is -9 dBFS. Bars turn yellow from -20 dBFS and red from -9 dBFS. CLIP appears in a bar that reached full scale, SILENT in a channel below -60 dBFS for 3 seconds. Channels are grouped and named by the audio channel configuration (for example PGM 1-2, INT 3-4), otherwise in pairs. Click a channel of another pair to listen to that pair; click a channel of the pair you hear to solo it. Under the bars, the correlation of the pair you hear: +1 is mono, around 0 a wide stereo image, below 0 (red) out of phase.
- Loudness (EBU mode): the momentary (M) and short-term (S) bars on the EBU +9 scale (0 LU = -23 LUFS) with the target band, the integrated loudness in large figures (green within -23 +/-1 LUFS, amber outside), the measuring time, loudness range (LRA), the highest peak, the highest M and S, and the short-term loudness of the last minute with the target band. The restart button starts the measurement again.
- Phase: the stereo image of the pair you hear. A mono signal is a vertical line, a wide stereo signal a cloud, a horizontal line means the channels are out of phase. For a 5.1 or 7.1 track the title has a Phase / Vectorscope switch; the vectorscope shows every speaker pair (see 23.6.1).
Status cells. Under the picture a row of measured values: video format and rate, audio format and rate, then for MPTS the CC errors of this service and the PCR jitter (how much later (+) or earlier (-) the last PCR arrived than its value says; red when PCRs come less often than every 100 ms), for ST 2110 the RTP sequence gaps of the video, the lost audio packets and the PTP domain and grandmaster seen on the input. Lip sync shows the AV offset. A value is colored only when it matters: green when it is good, amber when an error count is not zero, red while it still rises. The last cell opens the MPTS or flow detail.
Video coding cells (MPTS). After the video rate, an MPEG-2, H.264 or HEVC service shows up to four cells with how its picture is coded: GOP (for example "M=3, N=12, open"), Scan (progressive or interlaced, field order, MBAFF), Coding (CABAC, reference frames, slices) and Rate mode (constant or variable, with the nominal rate). A cell is left out when the stream does not say. Hover over a cell for the other compact details (GOP type, HDR format and so on); the full list is in the properties of the Video page. In the first seconds, and for up to about 20 seconds on a stream that is not aligned to its pictures, one Coding cell says collecting.
States. When nothing plays, the picture area says why:
- Pick a service to start: nothing chosen yet; the Choose service button opens the list, and the services you played last are offered below it.
- Starting: between the pick and the first picture (up to about 5 seconds in Raw mode, while the picture and sound are lined up).
- No signal since ...: the flow stopped sending packets. The last picture stays, dimmed, the frame turns red, and playback resumes by itself when the flow returns.
- Scrambled service (amber): the service cannot be decoded; the audio track menu is disabled and a button opens the MPTS detail.
- Radio service: no picture; the level meters, loudness and phase take its place.
- Data service: nothing to play; the PIDs of the PMT are listed with a link to the MPTS detail.
From the MPTS page. Each program tile on the MPTS page has an Open in AV Monitor button (monitor icon) that opens that program here.
Audio playback stops when you go to another page.
Lip sync. In Raw and Stream mode the audio plays in step with the picture when both share a clock: the video and audio of one MPTS program (PTS), or ST 2110-20 video with ST 2110-30 / AES67 audio (RTP timestamps derived from PTP, ST 2110-10). Each piece of audio is played when the picture with the same timestamp is on screen; audio that arrives early waits, audio that arrives late holds the picture back briefly. Audio from an unrelated source (no common clock) plays with a short fixed delay. In Stream mode the probe sends, with each piece of the H.264 stream, the timestamp of the source picture in it, so the audio follows the picture the browser's player shows, whatever it has buffered; the picture and sound start together there too. The video's own sound stays muted while an audio source is chosen, so the audio is heard once.
AV badge. While the audio follows the picture, the top left of the picture shows AV +N ms / AV -N ms: the remaining offset after alignment (+ = sound after the picture). Green below 20 ms, amber below 50 ms, red above. Hover over the badge to see how long the audio waits for its picture and whether the video is held back. No badge is shown when audio and video have no common clock. The badge measures the timing of the stream, not the content: a source whose sound is already out of step with its picture (wrong timestamps at the encoder) shows a small offset but still looks and sounds out of sync.
The AV Monitor is particularly useful in master control rooms and playout environments where operators need a quick "confidence check" of a program feed.
24.1 Multiviewer
The Multiviewer page (side menu, Tools) shows up to 16 pictures at once in a grid, like the multiviewer of a control room. Each tile shows the live thumbnail of a service (about one picture a second) with the alarm state of its flow, so one screen tells you which feed needs attention. Click a picture to open that service in the AV Monitor with sound and the full picture.
This page requires the Video license option. MPTS pictures also need the MPTS option and ST 2110 pictures the ST 2110 option.
Layout. The Layout buttons in the toolbar choose the grid: 2 x 2, 3 x 3, 4 x 4, 1 + 5 (one large picture and five small ones) or 1 + 7 (one large picture and seven small ones). The first tile is the large one. The grid is made as large as the window allows without scrolling. A smaller layout hides the last tiles but keeps them: a larger layout shows them again. On a phone the grid has two columns and the large picture goes across both.
Choosing the pictures. Click Add picture in an empty tile: the service list opens (the same list as on the AV Monitor and the Video page; radio and data services cannot be picked). The picked service goes into that tile and the next empty tile is selected, so you can fill the grid one pick after the other. The Service field in the toolbar shows the selected tile (it has a colored outline); its previous and next buttons change the picture of that tile. Click the name bar of a tile to select it. The menu of a tile (the three dots in its name bar) has Open in the AV Monitor, Choose another picture, Make it the large picture (1 + 5 and 1 + 7) and Remove from the grid. A service can be in one tile only: picking it for another tile moves it there. Empty every tile in the menu at the right of the toolbar clears the grid.
The layout and the pictures are saved with your user, so you find the grid as you left it in any browser. A saved service the probe does not see now is shown as Not on the probe now until it comes back.
Saved layouts. Besides your own grid, the probe keeps saved layouts for everyone: a name, the grid and the picture of every tile. The Saved layouts button in the toolbar lists them (the default one has a star and default); choosing one sets the grid and all its pictures at once, and it becomes your own grid. Its name stays on the button; an asterisk (*) after it means you changed the grid or the pictures since you loaded or saved it. The actions button next to it has:
- Save "name": stores the grid on screen in the selected layout.
- Save as...: stores it as a new layout under a name you type (1 to 64 characters; two layouts cannot have the same name, also not in another case).
- Rename... and Delete...: a deleted layout is gone for everyone; the page asks first, and the grid on your screen stays as it is.
- Set as default (or Clear the default): the layout the page opens with for a user who has no grid of their own yet.
- Copy link and Copy wall display link: see below.
- Export to a file: downloads the layout as a small file (
multiviewer-<name>.json), for example to take it to another probe. - Import from a file...: reads such a file. When a layout of that name exists, choose Replace it or Save as a new name, or Cancel.
Admin and NOC users save, rename, delete, import and set the default. The other users load the layouts, copy the links and export. At most 50 layouts are kept. They are part of the configuration backup (Settings > Backup), and every change is in the audit log.
Links and wall displays. Copy link copies the address of the page with the selected layout, for example https://probe:2300/multiviewer?layout=Studio%20A: whoever opens it sees that layout. Copy wall display link adds &fullscreen=1, which opens the layout straight in full screen. Browsers only allow full screen after a click, so when the browser refuses it the page fills the whole browser window with the grid and shows an Enter full screen button (and Not now); one click on it gives the real full screen. The button goes away by itself after 15 seconds, so an unattended screen shows the grid. If a link names a layout that does not exist, the page says so and opens as usual.
For a wall display that nobody operates, start the browser in kiosk mode with the wall display link, for example chromium --kiosk "https://probe:2300/multiviewer?layout=Wall%201&fullscreen=1" (Microsoft Edge: msedge --kiosk <link> --edge-kiosk-type=fullscreen). The browser then covers the whole screen, and the page fills it with the grid without a click. Log in once in that browser with a user that may see the page.
What a tile shows.
- Name bar (UMD): the alias of the flow, or its source and destination address; for an MPTS program the program name.
- Alarm border: the color of the worst open incident of the flow: red for Critical, amber for Major, blue for Minor, none when all is well. An Info incident and an incident that is still below trigger (see 27.1) draw no border. The name bar names the level too (Critical, Major, Minor; on a small tile the icon only). It follows the incidents of the Incidents page: an incident stays open until 10 seconds without a new event. Point at the badge to see why: the open incidents of the flow, the worst level first, each with its number of events (at most 8, then "+n more"). For an MPTS program the first line says On the transport stream with the name of the stream: the probe checks the transport stream, so every program of that stream shows the same alarm. Click the badge (in full screen: double click) to open the Incidents page with the open incidents of that flow.
- NO SIGNAL: no packets arrive for the flow (a No Data incident is open). The last picture stays dimmed behind it.
- BLACK, FROZEN, FROZEN + SILENT and the outlined still N s: the picture checks of an ST 2110-20 or ST 2022-6 flow (see 24.2), with how long the state lasts.
- NOT MONITORED: monitoring is off for the flow, so it raises no alarms.
- Timecode at the top right, when the probe reads one for the flow.
- Level meters at the right edge: the peak levels of the first two channels of the audio matched to the video (the audio the AV Monitor would play with it), green, amber from -9 dBFS, red at full scale. An MPTS program shows its levels while its audio is played somewhere.
Full screen. The Full screen button in the toolbar shows the grid alone on a black screen; press Esc to leave it. In full screen a single click on a picture does nothing, so a touch or a stray click does not change the wall; a double click leaves full screen and opens that service in the AV Monitor. The tile menus are hidden; the tooltip of an alarm badge still shows when you point at it, and a double click on the badge opens the incidents of that flow. Outside full screen a click on a picture opens the AV Monitor as before.
The probe works out the alarm state only for the flows on screen, and it decodes MPTS pictures only for the programs shown, so a large grid does not load it for flows nobody watches.
24.2 Black and Frozen Pictures
The picture checks (Settings > Checks, card Picture, see 31.4.4.22 to 31.4.4.28) judge every ST 2110-20 and ST 2022-6 flow once a second, also when no page shows its picture. Compressed video (the programs of an MPTS stream and ST 2110-22 JPEG XS flows) is judged on I-frames in turn (31.4.4.28): with many programs each one is judged every few seconds, and the pages say Judged on I-frames with how often. Their state is shown in four places.
Multiviewer tiles. On the normal grid a flag at the top left names the state with how long it lasts, and the top edge of the tile takes the severity color (red Critical, amber Major, blue Minor); the picture itself stays readable, as the last frame often tells why:
- BLACK 1:12: the picture is black (minutes and seconds since it went black).
- FROZEN 0:42: the picture does not move. A freeze on a named slate color says Frozen on name in the tooltip of the flag.
- FROZEN + SILENT 0:06: the picture is frozen and its paired audio is silent.
- still 6 s (outlined, no color): the picture has not moved for 4 seconds or more, the alarm comes at the raise time. It is a hint only: never an alarm, never counted or exported.
In full screen (and on a wall display) the state is shown as a big word in the middle of the tile (BLACK, FROZEN, FROZEN + SILENT) with its duration, and the picture is dimmed, so it can be read across the room.
AV Monitor. The flag is shown at the top of the picture, and the Picture status cell under it says Moving, Black or Frozen with the duration. Next to the audio tools, the Picture checks panel of an ST 2110-20 or ST 2022-6 video shows the black and freeze state, the moving samples of the last comparison (for example "3120 of 16200"), the mean luma of the judged area, when the picture was last judged, the settings in use and a trace of the moving samples and the mean luma of the last 2 minutes (refreshed every 2 seconds). Open the incident and Settings lead to the Incidents page and the Checks page.
MPTS Services. The Status column of a service shows its picture state as a tag: Frozen 0:42, Frozen and silent 0:42, Black 1:12 or Frozen on name in the color of the check's severity, the outlined hint still 6 s, or Freeze check off / Black check off when an override turns the check off for that program. A service with a picture alarm counts as a problem. The opened line says it in a sentence, with a link to the incidents, and how the program is judged ("I-frames every 1 s, judged about every 5 s, last judged 2 s ago": how often the stream sends an I-frame, and how often the probe judges one). A multiviewer tile of an MPTS program shows the flags of its own program.
Flow pages. The ST 2110-20, ST 2110-22 and ST 2022-6 pages have a Picture checks panel: the black and the frozen state with the measured values, where the settings come from (global, a group or this flow), Override for this flow (the Overrides tab of the Checks page for this flow) and three lanes of the last 24 hours: Frozen, Black and No signal.
The pictures of a flow are sent to a page only while that page shows the flow, so a page with many flows does not load the probe for pictures nobody looks at. The checks themselves and their marks do not depend on it.
24.3 Pixel Inspector and False Color
For an ST 2110-20 flow played in Raw mode, and for an MPTS program whose video is MPEG-2, H.264 or HEVC (in Raw and Stream mode), the toolbar of the AV Monitor and of the Video page has Inspect, Freeze and False color. They read the values of the picture on the probe, not the 8-bit picture of the browser: the true 10-bit values of an ST 2110-20 flow, the decoded values of an MPTS program at the stream's own bit depth (see "MPTS programs" below). For an ST 2110-20 flow in Stream mode the buttons are dimmed. For a scrambled program, or a program whose video the probe does not decode (JPEG XS, JPEG 2000, VVC and others), they are dimmed and their tooltip says why; radio programs do not show them. ST 2110-22 and ST 2022-6 flows and the test stream do not show them (the Video page keeps its own False color for them, see 22.2).
Inspect puts a crosshair on the picture and opens the Pixel panel in place of the audio tools (on the Video page above the properties; on a phone under the picture). Click or tap a pixel, or move the point with the arrow keys (one pixel, ten with Shift). A small readout next to the point and the panel show:
- the position in the source picture (x, y and the line; for interlaced pictures the field and the line in the field) and the RTP timestamp of the frame;
- Y', Cb, Cr and R', G', B' as 10-bit code, percent and mV;
- Legal (Y' 64-940 and Cb, Cr 64-960 in narrow range), In gamut (EBU R103 version 3.0: R', G', B' and Y' from -5 % to 105 %) and whether Cb, Cr are cosited with the pixel or interpolated (odd x in 4:2:2);
- for PQ and HLG flows also nits (the light on the display: PQ after ST 2084, HLG on a 1000 nits reference display) and stops relative to 203 nits, the HDR reference white of ITU-R BT.2408;
- the 5 x 5 neighbors in one component (Y', Cb, Cr, R', G' or B'), colored like the picture;
- with the window 3x3 or 5x5: the minimum, maximum and mean around the point.
Without Freeze the values follow the newest frame once a second. Freeze holds the frame the probe has when you press it: every value then comes from that frame and the badge FROZEN - RTP names it. The picture on the page stops too, at the frame the browser showed; with motion it can be a few hundred milliseconds away from the held frame, so read the values, not the picture. The probe releases a held frame when you unfreeze, close the viewer, after 10 minutes or when the picture stops; the panel then says so and the values follow the live picture again.
Pinned points. Press P (or Pin) to keep the point; up to four points (P1 to P4) are marked on the picture and listed with their Y', Cb and Cr. A fifth is refused until you unpin one. Copy values copies the point, the pins and the window statistics as tab-separated text (paste it into a spreadsheet).
Colorimetry. R', G', B', percent, nits and the gamut flag depend on the matrix (BT.709, BT.2020 or BT.601), the range (narrow or full) and the transfer (SDR, PQ or HLG). The probe reads no SDP for ST 2110-20 flows yet, so it guesses them from the picture size (BT.601 up to 576 lines, BT.709 above, narrow range, SDR) and the panel says so. Choose the right ones under Colorimetry; the choice applies to this viewer only (the values, the window and false color).
Inspector window. Inspector in the panel opens a larger window on the held frame (it freezes the frame when it was not frozen and releases it when you close the window):
- the picture zoomed to the pixel grid (16x, 32x) with the codes of one component on the cells (the button on the picture chooses the component), or the whole picture (Fit); a minimap shows where the zoom is;
- Point: click a pixel for its values and the window around it; Area: drag a rectangle for the minimum, maximum and mean of every component; Line: click a line for its statistics;
- the waveform of the line of the point (Y', Cb Cr, or an RGB parade), with the point marked; a click on the waveform moves the point along the line;
- arrows move the point (Shift: 10), P pins it, New frame holds the next frame, Save picture (PNG) saves the picture as the browser shows it.
False color colors the picture from the 10-bit values, twice a second, with a legend on the picture:
- Luma: bands of Y' in percent (on PQ and HLG flows in nits), from below black to above 100 %;
- Gamut (EBU R103): pixels outside the R103 limits in warm colors when above and cool colors when below, everything else in gray.
The probe works on these requests only while somebody uses them; at most ten requests a second per login (all viewers and windows of one login together). The legend names the matrix, range and transfer the false color was drawn with.
Hex. The value tables of the panel and the window, and Copy values, also give every code in hexadecimal (two digits at 8 bits, for example 0x71; three at 10 bits, for example 0x2A4).
24.3.1 MPTS Programs
For an MPTS program the inspector reads the picture the probe itself decodes for the live picture, before it is reduced to the 8-bit picture the browser gets:
- Values are the decoded codes at the stream's bit depth: 8 bits for most programs (Y' 16-235 in narrow range), 10 bits for HEVC Main 10 (64-940). The panel shows the bit depth and the chroma format, for example 8-bit and 4:2:0. In 4:2:0 one Cb, Cr pair serves two by two pixels; the probe interpolates it at the pixel from where the stream says the chroma samples lie (the chroma sample location, left when the stream does not say), and the panel names the pixels that share the pair, for example "Cb, Cr 4:2:0: x 1404-1405, field lines 170-171".
- Freeze holds the frame on the screen at that moment, whatever picture type it is (I, P or B), with no extra decode and no wait. The picture you see can be a few frames away from the held one. A row Held picture under the picture says which picture is held: its type and whether it is a field pair, the PTS and time code, the profile and level, the bit depth, chroma format and chroma location, and the colorimetry and range the stream signals ("assumed" where the stream does not signal them). A picture the decoder had to repair after lost packets is held as it is and tagged after packet loss: its values show the repair, not the source.
- After compression. The values are those of the decoded picture: blocking and ringing come from the encoder, not from the source, and a P or B picture is usually softer than an I picture. The panel says so.
- Interlaced programs are shown as the woven frame, as for ST 2110-20: the point names its field (1 or 2, in the order the fields are sent) and its line in that field, for example "field 2, line 170 of 540".
- Colorimetry comes from the stream (the H.264 / HEVC VUI or the MPEG-2 sequence display extension): matrix, range and transfer, PQ and HLG included. What the stream does not signal is assumed (BT.601 up to 576 lines, BT.709 above, narrow range, SDR) and marked so. Colorimetry in the panel still overrides it, for a stream that signals it wrongly.
- Inspector window: From the stream lists what the stream says (codec, profile and level, samples, primaries, transfer, matrix and range with their codes, mastering display and MaxCLL / MaxFALL of HDR programs) and which of it the inspector uses. New frame holds the frame then on screen.
- False color: in Raw mode the Luma bands of an SDR program are drawn by the browser on every frame of the live picture; like for ST 2110-20 that picture is the probe's H.264 copy of the program, so a pixel on the edge of a band can land in the next band. Gamut, the luma bands of PQ and HLG programs and Stream mode come from the probe, from its own decoded pictures, at most twice a second. The gamut check (31.4.4.25) does not judge MPTS programs.
Without a viewer (through the REST API) a freeze waits for the next I picture of the program, which the probe decodes at full size: 10 seconds or three GOP lengths, whichever is longer. A program that refreshes with intra slices instead of whole I pictures (periodic intra refresh) has none, and the request says so. A field coded H.264 program sends the first field of its I picture alone: the probe then holds that field only, at half the height, and says "one field decoded". Each login can hold three pictures at a time. See the REST API documentation.
24.4 Teletext
Tools > Teletext shows the teletext of a transport stream as a receiver does: the whole carousel of a teletext PID, page by page, live. It needs a PID that the PMT signals as teletext (or one that the probe found by its content when that scan is on) and the MPTS option. The teletext of an ST 2110-40 flow (OP-47 or ST 2031, see 24.4.7) opens with the MPTS or the 2110 option.
The Record button in the header records the service of the teletext page you view; the name of the recording holds the service and the page (29.5.1, 29.5.2).
24.4.1 Source and Header
One drop-down at the left lists every flow that carries teletext, grouped per flow with its name, destination address and kind, and one line per teletext PID: a status dot (amber when the PID has check faults), the service, the PID and a badge for each signaled subtitle page ("deu 888"). A click on a line opens that flow and that PID in one step. Type in the search field to filter on flow, address, service, PID (decimal or hex), language or page; the arrow keys and Enter work too. The closed drop-down shows the chosen line and the flow name.
Next to it are Subtitle overview (the accessibility overview and timing graphs of the Subtitles page), the search field for the text of all pages, and Rec, which opens the recording of the flow.
The address of the page holds the flow, PID, page, subpage and tab, so a link opens the same view. The Open viewer links on the MPTS PIDs and Services pages, Open in viewer on the Subtitles page and every search hit open this page on the right PID and page.
24.4.2 Tabs
| Tab | What it shows |
|---|---|
| Page | The page drawn as on a TV (zoom 1x to 4x or Fit, Reveal for concealed text, Flash, Box for subtitle and newsflash pages), with the remote under it. A pop-out button opens the page in its own window; Export saves the page or the whole service in a file (24.4.6). |
| Raw codes | The received code of every cell after parity (hex); a click on a cell explains it. |
| Grid | The carousel as eight magazine grids (tens by units), colored by page kind, with a mark for pages with subpages. |
| List | The same pages as a list with titles, repeat interval and last seen, with filters; a row opens the page. |
| Packets | Packet counts of the PID, in total and per second. |
| Details | Carousel cycle, service data (8/30) and the counts of the teletext checks. |
| History | The earlier transmissions of the page on a scrub bar, side by side with the changed cells marked, and the comparison of two PIDs. |
The cards Page and Service at the right show on every tab: the kind, title, subpages, repeat interval, header bits, character set and links of the page, and the store, carousel, initial page, signaled subtitle pages and bitrate of the service. The remote and the Keys card show on the Page and Raw codes tabs only.
24.4.3 Remote and Keys
The remote has the page number entry between - and +, Index, the four Fastext keys with the labels and targets the page itself sends, and the subpage strip (Auto follows the rotation, a click holds one subpage).
Navigation. The four colored keys follow the links of the page. A line under the keys says where they come from: Links from the page (Fastext, X/27/0), Links from TOP (the service's table of pages), No links sent: + / - step to the next page or This page sends no links. Index opens the index page the service announces, and the title of a page in the Page card comes from the TOP tables when the service sends them (otherwise it reads "no TOP title sent"). A three-digit page number in the text of a page opens that page when you click it.
| Key | What it does |
|---|---|
| 0 - 9 | Type a page number (three digits). |
| + / - | Next or previous page in the carousel. |
| Left / Right | Previous or next subpage (holds it). |
| Space | Back to Auto (follow the rotation). |
| Home | The index page. |
| R, G, Y, C | Fastext red, green, yellow, cyan. |
| ? | Reveal concealed text. |
24.4.4 Search
Type a word in the search field and press Enter. The results are grouped per PID and page with the hit marked on the page; Open in viewer shows the page (a hit in an earlier transmission opens the History tab at that transmission). With history also searches the transmissions the probe kept. Escape leaves the search.
24.4.5 History and Comparison
The probe keeps the earlier transmissions of every subtitle page, and a shorter history of the other pages, so you can go back in time without a recording. Open the History tab. A bar with a handle A (and B in the side by side view) shows the transmissions of the page, oldest at the left and Live at the right. Tall marks on the bar are new text, short marks are repeats of the same text that the probe folds into one, gray marks are cleared pages. Drag a handle, or press [ and ] to step A one transmission back or forward (, and . step B). Live puts the handle on the page as it is now. A page with several subpages has a Subpage list.
The view buttons choose what you see:
| Button | What it shows |
|---|---|
| One page | The transmission at handle A, drawn as the page was sent. |
| Side by side | Transmission A and B next to each other. Mark changes outlines the cells that differ, Ignore the clock leaves out the clock and header cells, Only changed rows hides the rows that are equal. |
| Compare PIDs | Every page and subpage of two teletext PIDs (choose A and B, for example the main and the backup feed), see below. |
Compare PIDs lists one line per page and subpage with the result (only in A, only in B, different or the same), the number of rows that differ and what differs. The four counters above the table (Only in A (pages), Only in B (pages), Different (pages), Same (pages)) are also filters. Show diff on a line opens that page side by side. Header and clock cells are not compared; when nothing differs the view says that the two PIDs carry the same pages.
If the probe keeps no transmission of a page, the tab says so: the page is not in the carousel (any more), or nothing of it has arrived since the history started. The size of the history follows the memory budget of Settings > Teletext (24.4.8).
24.4.6 Export
The Export menu of the Page tab saves what the probe has stored of the PID:
| Entry | What you get |
|---|---|
| This page as text | The page as plain text. |
| This page as PNG | The page drawn as on a TV. |
| This page as a teletext file | The page in a teletext format of editors and inserters: MRG TTI (.tti) with all subpages in one file, Softel EP1 (.ep1), FAB ETT (.ttp) or T42 packets (.t42). A page with several subpages gives a zip for EP1 and FAB ETT, one file per subpage. |
| Page transmissions report | Every transmission of the page with its rows, control bits and repeats, as the DVBControl analyzers write it. |
| Subtitles as EBU STL | Only for a subtitle page: the New and Clear updates in the EBU Tech 3264 file of subtitling systems. |
| All pages as text | Every stored page of the PID in one text file. |
| All pages as teletext files | A zip with one TTI, EP1, FAB ETT or T42 file per page. |
An export that needs the page store of the PID is refused with "The page store of this PID is paused by the memory budget" when the probe has paused the store of a PID that nobody looks at (24.4.8); open the PID in the viewer and try again. Subtitle files (SRT, WebVTT, EBU-TT-D, transcript, PNG) are on the Export menu of the Subtitles page (25.2).
24.4.7 Teletext in ST 2110-40
An ST 2110-40 flow that carries teletext as OP-47 (SDP or multipacket) or ST 2031 is a teletext source too. It has no PID: the source drop-down lists it as ST 2110-40 ANC with the forms it carries (for example "OP-47 SDP"), the service name is the flow name, and the page opens the same way. The tabs, search, history, exports and the Subtitles page work as for a transport stream. Because the data comes in video frames and not in PES packets, there is no PTS: the timing checks of EN 300 472 (late, retention) and the PES layout check do not apply, and the Timing tab of the Subtitles page and the timing columns of its exports say so. The teletext of an ST 2110-40 flow opens with the MPTS or the 2110 option; a teletext PID of a transport stream still needs MPTS.
The check Teletext lost (31.4.4.38) watches both kinds of source.
24.4.8 Settings and Memory
Settings > Teletext (under Monitoring) has the Memory budget (default 128 MB, 16 to 16384 MB) that the page stores and the history of all teletext sources share, with the memory in use and how many sources keep a store. When the budget is full, the sources that were viewed longest ago keep their checks and lose their store and history; a source you open gets its store back. The switch PIDs without a teletext descriptor (off by default) also tries private PES PIDs, so a plant that forgot the descriptor is found; it costs a little CPU per such PID. Only an Admin changes these settings. The checks (31.4.4.38 to 31.4.4.48) run on every teletext source whatever the budget.
A session can show 4 teletext viewers at once (16 on the probe), the pop-out windows included.
25 Subtitles
The Subtitles page shows the DVB subtitles (EN 300 743) and the teletext subtitle pages (EN 300 472) of MPTS streams, the teletext subtitles of ST 2110-40 flows and the closed captions (CEA-608 and CEA-708) of MPTS video and ST 2110-40 flows. It shows each subtitle page as the viewer would see it, with the timing of every display set, so operators can check that subtitles are present, placed inside the safe areas and on time.
This page requires the Ancillary license option. DVB subtitles and closed captions of MPTS streams also need the MPTS option; without it the page lists only the ST 2110-40 caption flows and says so in the source list.
25.1 Choosing a Source
One drop-down at the left of the toolbar lists every source. Type in its search field to filter on input, transport stream, address, service, PID (decimal or hex), language or type; the arrow keys and Enter pick a line, Escape closes the list.
- The list is grouped by input > transport stream, with the destination address of the stream.
- A DVB subtitle PID is one line: service, PID with its hex value, language, the subtitling type ("hard of hearing") and the tag DVB-Sub, with the display size at the right.
- Closed captions: the video PID (or the ST 2110-40 flow) is a heading line with the tag CC; the channels found in it are the indented lines under it (608 CC1-CC4 and T1-T4, 708 services, XDS). Only channels that are present are listed.
- A teletext subtitle page is one line: service, PID, the page number (for example 888), the language and the tag Teletext; the subtitling type tells a page for the hard of hearing. A teletext subtitle page of an ST 2110-40 flow shows the flow and the form it is sent in (OP-47 SDP, OP-47 multipacket or ST 2031) instead of the PID.
- A red tag with a count ("2 late") marks a PID with check faults.
The closed drop-down shows only service, PID, language and type, for example "ZDF - PID 131 - German - DVB-Sub"; the input and transport stream are in the info line under the picture.
The Record button in the toolbar records the service you view; the name of the recording holds the service and the page or language (29.5.1, 29.5.2).
25.2 Toolbar
| Control | What it does |
|---|---|
| Start / Stop | Stop freezes the page (picture, counters and report); Start follows the source again. |
| Transparency | Shows transparent pixels as a checkerboard; when off, the page is black like a TV screen. |
| Safe areas | Shows the action safe area (green, 93%) and the title safe area (red, 90%). |
| Regions | DVB subtitles: outlines each region with its id. |
| Windows | Closed captions: outlines the caption window with its id. |
| Report | Shows or hides the report list under the picture. |
| Export | Saves the source in a file; see "Export" below. |
Export. One menu for every kind of source. An entry the source cannot be saved in is grayed out; point at it to see why (for example "DVB subtitles are pictures; they carry no text").
| Entry | Teletext | DVB subtitles | Closed captions | What you get |
|---|---|---|---|---|
| Report as CSV | yes | yes | yes | The report list (see the sections below). |
| Text transcript | yes | - | yes | One line per subtitle with its date and time (UTC). |
| SubRip (SRT) | yes | - | yes | Numbered subtitles with their times and colors, for players and editors. |
| WebVTT | yes | - | yes | Web subtitles with their times, place on the screen and colors. |
| EBU STL | yes | - | yes | The EBU Tech 3264 file of subtitling and playout systems; a teletext page keeps its rows and colors. |
| EBU-TT-D (TTML) | yes | - | yes | The EBU Tech 3380 XML file used for streaming. |
| Picture of the shown subtitle | yes | yes | yes | The subtitle on screen as a PNG picture, transparent around the text. |
| Every display set as PNG | - | yes | - | A zip with one picture per display set the probe keeps and an index.csv with their times and places. |
| Page transmissions report | yes | - | - | Every transmission of the page with its rows, control bits and repeats. |
The subtitle files come from what the probe keeps of the source, not only from what the page received since you opened it: the history of a teletext subtitle page, the last 200 display sets of a DVB subtitle PID (while a Subtitles page shows it) and the last 200 lines of a caption channel. The times in SubRip, WebVTT, EBU-TT-D and EBU STL start at the first subtitle (EBU STL at 10:00:00:00).
25.3 DVB Subtitles
The subtitle page is drawn at its display size: the size of the display definition segment, or 720 x 576 when the stream sends none. An HD page (1920 x 1080) is shown at half size. The regions are drawn from their pixel data and color table (CLUT), including semi-transparent colors.
The info line under the picture shows the input, transport stream and service, the display size, the page with its version and time-out, the number of regions with their bit depth and CLUT, the PTS, and how long the display set is (or was) on screen. "Waiting for an acquisition point" means the probe lost data and waits for the next complete page.
The picture follows the stream's clock, like a set-top box: each display set is shown at its PTS (the time stamp at which the receiver must display it, measured on the PCR of the program), not when it arrives. Broadcasters often send the set that clears the screen seconds before its PTS, while the text before it is still on screen; the page keeps showing the text until that moment. A display set that arrives after its PTS is shown at once. A page with a time-out is removed when its time-out ends (counted from its PTS) and no new display set came before, as a TV removes it. When the program has no usable PCR (none for a second, or just after a PCR discontinuity), the sets are shown as they arrive and the page says so. The report rows appear as the display sets arrive; clicking a row shows that display set at once, also one that is still waiting for its PTS (see 25.3.1).
The counters (the chips in the strip above the report, see 25.3.1) count since the page was opened: display sets, late display sets, the worst lateness, short PTS steps, the largest display set, buffer errors, page id errors, segment errors, version repeats, page time-outs passed, object data errors and object data warnings. A counter that is not zero is red for late display sets and for buffer, page id, segment and object data errors, and amber for short PTS steps, version repeats, time-outs and object data warnings. The checks run all the time, also while the page is closed. Late display sets, short PTS steps and buffer, page id and segment errors are also the five DVB subtitle checks of ETR 290 (18.5), with episodes, history and alarms. When the subtitles of a PID stop while the stream goes on, the incident Subtitles stopped is raised (see 31.4.4.12).
25.3.1 Display Set Report
The report lists the display sets, newest at the top, so a new set is visible without scrolling. When you have scrolled down in the list, it stays where it is while new rows arrive at the top; when it is at the top, it follows the newest row. The report has nine columns:
| Column | Meaning |
|---|---|
| Date/Time | When the first packet of the display set arrived (local time of the probe, year/month/day). |
| PCR Start | The PCR at the first packet of the display set. |
| PCR Stop | The PCR at the last packet of the display set. |
| PTS Time | The presentation time (PTS) of the display set. |
| Pre-roll | The PTS minus the PCR at the first packet: how early the display set started to arrive. A minus sign means the PTS was already passed; a dash means there is no PCR. |
| Late (ms) | How far the PTS had already passed when the display set was complete, in whole milliseconds (a dash when on time or within the 100 ms margin). |
| Check | One tag for each kind of fault found in the display set (below), with an "i" that explains it. |
| State | The page state: normal, acquisition, mode change or reserved. |
| Notes | A short text for the row. It takes the rest of the width; a long text ends in "..." (the whole text is in the tooltip and in the copy). |
The Notes start with the display set itself (clear screen, update or nothing), then lateness estimated when the arrival time was estimated from the bitrate of the stream, then the short form of each fault with its main number, the worst first, for example "Late 418 ms; Step 26.2 ms".
The Check tags. A display set with a fault shows one tag per kind of fault:
| Tag | Meaning |
|---|---|
| Late | The display set was complete more than 100 ms after its PTS. |
| Step | The PTS is less than one frame after the PTS of the previous display set. |
| Buffer | The display set is larger than the coded data buffer of the decoder. |
| Page id | A page id error in the display set. |
| Segment | A segment error in the display set. |
| Version | A page version was repeated with new content. |
| Time-out | The page was removed by its time-out before the next display set came. |
| Reference | A region, object or CLUT that the page uses is not defined or was not received, or a CLUT is defined only by an alternative CLUT segment (warning). |
| Object | The pixel data of an object: the field data blocks run past the end of the segment (error), more than one stuffing byte follows them (warning), or a progressive bitmap is sent for a service whose subtitling type in the PMT does not allow it (0x16 or 0x26 is needed; warning). |
A red tag is an error and an amber tag a warning; a row with an error is washed red and a row with a warning amber. Point at the tags or at the Notes of a row for the tooltip. For every fault it gives Error or Warning and a title, the explanation with the numbers of this display set (for example "Late by 592 ms (PTS 07:10:58.353, complete at 07:10:58.945; it must be complete by the PTS)"), and the rule of EN 300 743 V1.6.1 that the display set breaks, with its clause, for example "EN 300 743 clause 5.1: A display set must be complete in the decoder by its PTS".
A display set is late only when its last byte arrives more than 100 ms after its PTS: clean broadcasts are often a few tens of milliseconds late, which a viewer does not notice. The reference checks are made when a display set is complete and only after the probe has seen the start of the page epoch (a mode change), so a stream joined in the middle shows no false alarms. At most four notes of one kind are listed per display set; the rest is summarized as "N more object reference notes". A clear screen row is shown in a lighter color.
The strip above the report. From top to bottom it shows:
- The selected row. After you click a row, the strip says "Showing the selected row (date, PTS), not live" with Back to live. The picture and the set line show that display set until you click Back to live, click the row again or press Esc. When the probe no longer keeps the picture of the row (it keeps the last 200 display sets, at most 16 MB), the strip says so and shows the live page.
- The set line. The page, its version and time-out, then "clear screen" or the area, the bit depth, the regions and the CLUT, the size in bytes, the PTS, how long the set is shown, and the state.
- The counter chips. The counters of 25.3, for example "Late display sets 2" and "Worst late 592.0 ms".
Copy and CSV. The copy button at the top right of the report (or Ctrl+C in the list) copies the whole report as tab-separated text that you can paste into a spreadsheet. Export > Report as CSV in the toolbar saves the same as a file. Both hold the nine columns, then page, version, time-out, area, depth, regions, CLUT, size, shown time and the text of the checks, oldest row first.
25.4 Closed Captions
For a closed caption channel the picture shows the caption rows as a TV would: white on black by default, with the pen colors of the stream, placed on the caption grid inside the title safe area. Text in any language (for example Korean 708 services) is shown as sent.
The info line shows the service and its language, the caption window (rows, columns, roll-up or pop-on), the pen position, where the caption data comes from (for example H.264 SEI, A/53) and the XDS program information. The counters show the caption data packets, 608 parity errors, 708 sequence errors and whether the captions stopped.
The caption report has one row per change of the shown text: Date/Time, PTS (the timecode for ST 2110-40), Channel, Mode and Text. For ST 2110-40 flows the counters are the caption packets, characters, characters per second, words per minute and the idle time.
25.5 Absence Alarm
A DVB subtitle PID that sent subtitles and then stops while the stream goes on raises the incident Subtitles stopped (default delay 30 s, set per rule, 31.4.4.12). The five ETR 290 subtitle indicators are not part of the ETR 290 report yet.
A caption service of the picture (CEA-608, CEA-708, OP-47 subtitles in ST 2110-40, ST 2022-6, or the closed captions in the video of an MPTS) that was present for at least 10 seconds and then has no captions for the delay (default 30 s, set per rule, 31.4.4.20) raises the incident Captions stopped. The incident names the service ("CEA-708 stopped: no captions for 30 s"), stays open while a service is stopped and ends after 2 seconds of captions again. A flow that stops entirely is reported as No data, not as Captions stopped, and a service that was never seen does not raise the incident. Only caption content counts: a CEA-608 field that sends only null padding, such as a captioner who is quiet between programs, is the same as a stopped service. The two incidents do not overlap: Subtitles stopped is for DVB subtitle PIDs (EN 300 743) of an MPTS, Captions stopped for the caption services above. This helps detect:
- Encoder failures that stop generating captions.
- Routing changes that disconnect the ancillary data path.
- Program segments that should have captions but do not.
25.6 Teletext Subtitles, Timing and Overview
Choose a teletext subtitle page in the source list and the page shows the subtitle as a TV boxes it into the picture, with a report of every update the page received. The page is the same drawing as in the teletext viewer (24.4); Open in viewer opens the PID and the page there. Three tabs sit above the picture:
| Tab | What it shows |
|---|---|
| Updates | The picture and the report of the page. |
| Timing | The timing graphs of one subtitle page (below). |
| Overview | The accessibility overview of all services (below). |
Updates. One row per update: Date/Time, PTS, Pre-roll, Late (ms), State, Rows, Text, Check and Retention. The state is New (new text), Hidden (the page text is sent with Inhibit Display set, so a receiver does not show it), Repeat (the same text again, folded into the row with a count), Clear (the page is cleared) or Empty. The tags in the Check column are the faults of the update, for example Late; point at one to read the clause of the standard. A click on a row holds that update on the picture, Back to live follows the page again. The probe also stores the New and Clear updates of a subtitle page in its history database for 7 days, so the list can go back further than the time the page was open; the exports of the Subtitles page (25.2) use the same history.
Timing. Choose the Subtitle page and the range (10 min, Last 600 updates or 1 h). The first graph shows for every update how long after its PTS the last packet arrived: above zero is late, below zero is the time to spare (pre-roll), and the dashed line is the late margin. The second graph shows the retention of every update (the check Teletext subtitle late with its Retention option, off by default). The third shows the updates per minute of the last 24 hours, with the late minutes capped red. Shapes tell the classes as well as colors do: a circle is on time, a square late within the margin, a diamond late. Hover shows an update, a click on a late one opens it on the Updates tab, Table shows the same points as rows and CSV saves them.
Overview. One row per service for the accessibility of subtitles: its subtitle pages (teletext TTX and DVB subtitles DVB, HOH for the hard of hearing, with the language), whether text is on air now, the share of the last hour that had subtitles (coverage), a strip of the last 24 hours in 15 minute steps, the longest gap, a timing summary and the open faults. A service without any subtitle is marked No subtitles signaled. Choose the transport stream or All flows, Without subtitles only lists the services that need attention and Report (CSV) saves the table.
26 Compare
The Compare page allows side-by-side comparison of two or more flows.
This is useful for:
- Comparing a primary and backup feed to verify they carry identical content.
- Comparing input and output flows across a processing device to detect introduced errors.
- Comparing the same flow measured at different points in the network to identify where degradation occurs.
The page provides synchronized views of key metrics for the selected flows, making differences immediately visible.
27 Reports
The Reports page provides reporting and data export functionality for documentation, compliance, and historical analysis.
Reports are included in the Base license option.
Reports can be generated for:
- Flow health summaries: An overview of all monitored flows with their current status, bitrate, and error counts.
- Incident logs: A chronological record of all warnings and errors detected over a specified time period.
- Compliance reports: ST 2110-21 VRX compliance results for video flows.
Reports can be exported in structured formats suitable for analysis, auditing, or integration with external systems.
The Incident Summary tab shows:
- The number of flows in warning or error now, with Incident history (the Incidents page with the open episodes) and Export CSV (the active flows with their incident types and start times).
- Last 24 hours: the episodes and errors the probe stored in the last day, per incident type; click a type to see its episodes on the Incidents page.
- The active flows with their Incident types and Since (from the incident history, so it survives a reload); the history icon at the end of a line opens the History tab of that flow.
27.1 Incident History
The Incidents page (Analyze > Incidents) shows every incident the probe stored, for all users. The probe keeps an incident as an episode: one line per flow and incident type (for ETR 290 per check and table, PID or program, with its severity), from the first second with events to the last one. The episode closes after 10 seconds without an event, so a burst of errors is one line. Incidents are kept for the days set under Settings > System > History Database (default 90 days) and survive a restart of the probe and a reload of the page. An episode that ends because the monitoring of its flow was switched off (see 31.11) shows Ended: monitoring switched off.
The severity scale. Every incident has one of four levels: Critical (red), Major (amber), Minor (blue) and Info (gray). Critical, Major and Minor are alarms: they count as active incidents, mark the tile of a flow in the Multiviewer and are sent to other systems. Info is not an alarm: it is listed here and can be filtered, but it raises no mark on a tile and is not counted as an active alarm. Incidents stored before the levels existed are shown as Major (formerly error) and Minor (formerly warning). A check can be set to raise its alarm only after several episodes within a time window; the first episodes before that are listed with a gray below trigger mark and without a color, and they are not counted as an alarm.
The same episodes in the log. The probe log (the Logs page) shows an incident as two lines, not one line every second while it lasts: a RAISE line when the episode starts (flow id, input, protocol stack, the incident and the detail of its first event, at the error or warning level of the incident) and a CLEAR line when it ends (its duration, the number of events, the most in one second and the detail of the last event, at the status level). For example:
FlowId: 12 input:1 ETH:IPv4:UDP:RTP => RAISE RTP Sequence Counter Increment ERROR => last gap: 3
FlowId: 12 input:1 ETH:IPv4:UDP:RTP => CLEAR RTP Sequence Counter Increment ERROR duration: 4.5 s count: 7 peak: 3/s => last gap: 1
The CLEAR line is written 10 seconds after the last event and carries the time of that last event. A flow that flaps (errors that come back within 10 seconds) stays one incident. After 20 episodes of one flow and incident in an hour, one RAISE line says that the next ones are summed, and one CLEAR line at the end of the hour says how many there were. When more than 200 incidents start in one second, the rest are not logged one by one: one line a second says how many were left out (they are counted on the flows as usual). An incident that is still open when the monitoring of its flow is switched off or the probe stops gets its CLEAR line then, marked "(monitoring switched off)" or "(IPProbe stopped)".
No Data stays open until the flow is back. A flow that stops sending is one open No Data incident for as long as the silence lasts, also for hours: its CLEAR line comes when packets arrive again and says "flow resumed after" the silent time. If the flow goes silent again after that, it is a new incident.
Clear by hand. For a flow that is gone for good (a source taken out of service), an Admin or NOC user can end its open No Data incident: the episode shows Clear in the Detail column. Click it, then confirm with Clear incident in the same row (Cancel keeps it). The episode then shows Ended: cleared by hand with "cleared by" and your user name, the log gets a CLEAR line "(cleared by <user>)", and the clear is recorded in the audit log. Other users do not see the button.
Filters (kept in the page address, so a reload or a shared link shows the same selection):
- Time range: last hour, last 24 h (default), 7 days, 30 days or a custom from / to.
- Type (an ETR 290 check chosen from the summary appears as its own entry), Severity (Critical, Major, Minor or Info), Input, Open only, and a text search in the flow name, address and details.
- Clear filters resets them.
Summary. The heading counts the episodes of the range per level (Critical, Major, Minor, Info). Above the list, Per hour shows the episodes of each hour as a bar, colored by level from the bottom (red Critical, amber Major, blue Minor, gray Info); click a bar to show only that hour. By type and By flow list the types or flows with the most episodes; click a line to show only that type or flow.
Episodes. Newest first: the start, the duration (so far while open), the type, the flow (its alias at the time and its address; a link to the flow page while the flow is on the wire; the filter icon shows only this flow), the input, the severity (the level word with a colored dot), the number of events with the most in one second, and the first and last detail logged. Words mark special lines: open (still going on), interrupted (the probe stopped before the episode ended) and +N more (more than 20 episodes of that flow and type in one hour, summed into one line). Load more reads the next 100.
Live updates (on by default while the range ends now): the probe sends every episode that starts, grows or ends within about a second, so new episodes appear at the top and open ones move on without a reload. The event count of an open episode moves on every 10 seconds (the probe stores it every 10 s), and the summary follows at most every 10 seconds. Only episodes that match the filters are added; with Open only, an episode that ends leaves the list. As a check the page reads the list again once a minute, after the connection to the probe was lost and came back, and when the history database of the probe was replaced (a restore, Start fresh or Recover). Without a live connection it reads the newest episodes every 10 seconds instead. Switching Live updates off stops the updates; switching it on again reads the list at once.
Export saves the selection as CSV or JSON (at most 100000 rows; the message says when the file was cut).
On a phone every episode is shown as a small card.
History of one flow. The History tab of a flow's detail page shows that flow's stored incidents for the last 24 h, 7 days (default) or 30 days: the number of episodes per level (Critical, Major, Minor, Info) and of events, how many are open now, a line per incident type (click it to see these episodes on the Incidents page) and the newest 50 episodes. The tab follows the probe live: an episode of this flow that starts, grows or ends shows within about a second, the counts per type follow at most every 10 seconds. Open in Incidents opens the Incidents page filtered on this flow and range. A link to a flow page with ?tab=history in the address opens this tab directly.
27.1.1 Picture Evidence
A Black picture or Frozen picture episode has an Evidence button under its type (the type has a moon or a snowflake icon); it opens the evidence the probe kept with the episode below the row (see 31.4.4.27):
- Three pictures with their times: the last moving picture (for a black episode the last good picture, at most 5 seconds before the start), the first frozen picture (first black picture) and the latest picture, which is refreshed while the episode is open.
- The trace of the moving samples (blue) and the mean luma (purple) per second, from before the raise up to the clear (or now). The shaded part is the episode, the marker the moment the alarm was raised, and the dashed line the motion limit (the moving samples needed).
- Settings used: the values the check judged the flow with (motion threshold, samples, raise time, black level, skipped lines and margins).
The evidence is removed together with the episode. Without it the row says No evidence kept for this episode. Frozen and silent episodes have no evidence of their own: open the frozen picture episode of the same flow.
28 Logs
The Logs page (System in the sidebar) shows the probe's own messages as they come in: inputs that start and stop, configuration changes, recordings, license and update messages, warnings and errors. Faults on the flows themselves are listed under Incident History (27.1); the Logs page is where to look when the probe itself behaves unexpectedly, or when support asks for the log.
28.1 Reading the Log
Each line shows the time, the level and, when it has one, the part of the probe that wrote it in brackets (for example [WebServer]). The newest line is at the bottom and the view follows new lines. Scroll up to read older lines: the view stops following and a Jump to latest button appears at the bottom; click it to follow new lines again.
The levels are:
| Level | Meaning |
|---|---|
| Info | Normal operation. |
| Status | A state change, for example an input that comes up. |
| Warning | Something that needs attention but does not stop the probe. |
| Error | Something failed. |
| Debug | Detail for support; hidden unless you switch it on. |
28.2 Filtering
- Level buttons. The buttons at the left of the toolbar switch each level on and off; the number on a button is how many lines of that level the page holds.
- Search. The search box shows only the lines that contain the text; the number of matches appears next to it. The .* button switches the box to a regular expression; it turns red while the expression is not valid.
- Source filter. The filter button opens a row of chips with the parts of the probe that wrote lines. Log View chooses which of them the page shows; All shows every source.
The filters change only what this page shows; the probe keeps logging everything.
SCTE-104 messages. Every SCTE-104 message of an ST 2110-40 or ST 2022-6 flow is an Info line with the flow, its name, the RTP timestamp, the message number and what the message says, for example FlowId: 12 input:1 ETH:IPv4:UDP:RTP => SCTE-104 "Studio 1" RTP 3600000 message 4: splice start normal, event 42, pre-roll 4000 ms. A duplicate (a message the encoder sends again) is marked (duplicate). Only monitored flows write these lines, and not in the first seconds after their input starts. At most 20 lines a second are written per flow; the rest are summed in one line, 12 more SCTE-104 messages (over 20 a second, not listed). Search for => SCTE-104 to see only these lines.
28.3 Syslog File (Admin)
An Admin sees two more rows in the source filter. Syslog and Syslog levels choose which sources and levels the probe writes to its log file on disk. While that file is filtered, a small label in the toolbar says so and lists the levels that are written. Unlike the page filters, this setting applies to the probe itself, not only to your page.
28.4 Copy, Download and Clear
- Copy puts the lines that are shown on the clipboard.
- Download saves them as a text file, for example to send to support. For a complete package with the configuration and the system state, use Settings > System > Support (31.15).
- Clear empties the page. It does not delete anything on the probe.
29 Recordings
29.1 Overview
IP-Probe can record the packets of a stream to disk in the standard pcapng format. A recording holds every packet exactly as it appeared on the wire, including Ethernet headers, VLAN tags and, when the network card supports it, hardware arrival times. You can open the file in Wireshark, tshark or any other pcapng tool for offline inspection, fault finding or as evidence.
The normal way to record is one flow: you pick a flow, choose when to record and for how long, and press one button. Groups, inputs, several flows at once and all traffic are possible too, under Advanced in the same dialog (29.4).
There are three ways to record:
| Type | What it does |
|---|---|
| Manual | Starts now and stops after the time you choose, or when you press Stop. |
| Scheduled | Starts at a date and time you choose, once or on a repeat, and stops after the length you set. |
| Event | Waits for a fault on the flow (for example packet loss), then records. It also keeps the seconds before the fault. |
The probe runs up to 64 recordings at the same time (running recordings, and event recordings that are waiting). Recording runs on its own processor core, so monitoring is not slowed down.
Only an Admin or NOC login can make, stop, edit and delete recordings. Other users (Viewer, Management) can see the list and download files.
29.2 The Recordings Page
Open Recordings in the sidebar. From top to bottom the page shows:
- The disk banner. The recording disk with its used and free space as a bar, and next to it how much of the storage limit for recordings is used and how much is left. The bar turns amber when the free space gets close to the space the probe keeps free, and red when it is below it. An Admin sees a gear button for the storage settings (29.7).
- The toolbar. A search box (name, flow or address), the counter "N of 64 recording slots in use", and the New recording button. On a phone the button shows only the red dot.
- Filter chips. All, Running, Scheduled, Waiting, Finished, Problems and, for logins that can record, Mine. Each chip shows how many recordings it holds. Click a chip to show only those; click All to see everything.
- The table. One row per recording, with the columns Name (with a picture of a video flow, the flow and its address), State, When (when it started or will start), Duration, Size, Result and the actions at the end, and a box at the start of each row (only finished recordings can be ticked, 29.6). Click a column heading to sort by it; click it again to reverse the order (the arrow shows the direction). By default the table is sorted by state: running recordings first, then scheduled and waiting ones, then the others, newest first. When the table is narrow the When, Size and Result columns fold into the first column. Only on a phone is every recording a card instead, with a Sort by box.
If the page was opened from a flow (View after a recording was started there), a Flow: chip at the top shows that only the recordings of this flow are listed; click it to see all again.
29.2.1 States
| State | Meaning |
|---|---|
| Running | Recording now. The row shows a progress bar with the time recorded, the time left and the live data rate. A recording without a set length shows "Until stopped". |
| Queued | Ready to record, waiting for a free recording slot. |
| Scheduled | Waits for its start time. The row shows the start time. A repeating recording shows "Repeats daily" or the weekdays, the Next run and the result of the last run. |
| Waiting for event | An event recording that is armed. The row shows since when it waits. |
| Not started | Made but not started yet. Start it with Start. |
| Finished | Ended normally: the time was reached, you pressed Stop, or the stop condition was met. |
| Stopped: limit | Ended early by a limit: the storage limit, or the free disk space the probe keeps. The result line says which. |
| Interrupted | The probe was stopped or restarted while it recorded. The part recorded is kept. |
| Missed | The probe was not running at the start time and the whole recording window had passed. |
| Failed | The recording could not be made. The result line gives the reason. |
The Result column explains how a recording ended, for example "Stopped by a user", "Stopped: duration reached", "Stopped: storage limit", "Stopped: disk space" or "Not started: storage limit". An event recording that kept packets from before its fault says "Includes 5 s before the event". A recording whose flow is no longer on the probe shows a Flow gone tag next to its name. The recording of one service of an MPTS stream shows "Service" with the name and program number (for example "Service ZDF HD (SID 28006)") on the flow line, with the flow below it (29.5.2).
A finished recording of a video flow shows a small picture taken while it recorded. A running recording shows the live picture.
29.3 Recording a Flow
Click New recording (or use Record this flow, see 29.5). One dialog opens; there are no steps.
- Name. Filled in from the flow and the time, for example "TS 124.2 2026-11-06 14:32". Type to change it (up to 80 characters). The file on disk is the name plus the start time, with a
.pcapngending. - Flow. Click the box and search by name, address or type. The flows are grouped by input, with the data rate of each; a flow without traffic is marked "inactive". Every row shows the destination port and a type chip (for example MPTS, RTCP or ST 2110-20), so flows of one address are easy to tell apart. An RTCP flow (the same address, port + 1) is listed right below its media flow, so the two cannot be mixed up: record the media flow, not its RTCP companion. When the dialog was opened from a flow, that flow is filled in and Change lets you pick another.
- Type. Manual, Scheduled or Event.
- The settings of that type (below).
- The line under the settings estimates the file size, for example "About 332 MB (44.3 Mbit/s for 1 min)", and how much room is left for recordings. It turns amber when the recording hardly fits, and red when it does not fit: then the button is disabled and a short reason is shown next to it.
- Press Start recording, Schedule or Arm.
29.3.1 Manual
Choose how long to record: 1 min, 5 min, 10 min, 15 min, 30 min, 60 min or Until I stop it. The first time the choice is 1 min; afterwards the dialog remembers your last choice in this browser. A quick record (29.5.1) always starts with 1 min and does not change the choice that is remembered. The recording starts at once and stops by itself after the time. To stop earlier, use Stop in the list.
29.3.2 Scheduled
- Repeat. Once, Daily, or Weekdays with the days you pick (Mon to Sun).
- Date (First date for a repeat) and Time, 24 hour clock, in the time zone of the probe. The dialog says so when your browser is in another zone.
- Length. Either a Duration (1, 5, 10, 15, 30 or 60 minutes), or an End time with an end date and time.
- A summary line shows the result, for example "Sat 7 Nov 2026, 20:00, 30 min". A start in the past, an end before the start or a date more than a year ahead is refused with a message.
The recording appears in the list as Scheduled. A repeating recording makes one new recording for every run (named with the date and time of the run) and stays in the list as the schedule with its next run. Use Cancel on the schedule to stop all further runs. Runs at the same clock time follow the clock of the probe, also across a daylight saving change.
If the probe is not running at the start time, a recording whose window is still open starts late and stops at the planned end. When the whole window has passed, the row says Missed.
29.3.3 Event
- Start when. The event that starts the recording, for example "RTP sequence gap" or "No data". The list holds the events that fit the type of the chosen flow (RTP, ST 2110, MPTS with its ETR 290 checks, PTP, and so on) in groups. Any error on this flow starts the recording on any of them.
- Keep before the event. The seconds recorded from before the event, 0 to 60 s, default 5 s. The probe keeps the newest packets of the flow in memory while it waits, so the file starts that many seconds before the fault.
- Record after the event. The seconds recorded after the event (default 10 s); the buttons give 10 s to 30 min.
The dialog shows how much memory the wait holds ("Memory held while waiting"). The memory is limited: when a very fast flow cannot hold the full time, a message tells you how many seconds it can keep ("Capped: ... holds only 3.2 s before the event"). When the memory for all waiting recordings is used up, a new event recording is refused.
An event recording records once: after the first event it ends as Finished. To wait for the next event, use Arm again on the finished row. A waiting event recording is kept across a restart of the probe.
29.4 Advanced
Below the settings, Advanced: several flows, a group, an input or all traffic opens extra choices:
- Record: One flow, Several flows, A group (the groups of the Groups page), An input or All traffic. All traffic on a busy probe makes very large files: use it for a short time.
- One service (the record choice for an MPTS flow): records one program of the transport stream instead of the whole stream. A Service picker lists the programs of the flow with their names (29.5.2).
- Stop after N packets instead. A recording that stops at a packet count instead of a time.
Advanced is not available when you edit a recording.
29.5 Record from a Flow
You can start a recording from the place where you see the problem, in two clicks:
- Flows page. Right-click a row (on a phone: press and hold), or click the three dots at the end of the row, and choose Record this flow.... The menu also has Open full page.
- Flow drawer. Click the Record this flow button (a red dot) before the Close button.
- Flow detail page. The same button is in the header.
The dialog opens with the flow filled in. After you press Start recording, the dialog closes, you stay where you were, and a message says the recording started, with View, which opens the Recordings page for that flow.
While a recording of the flow runs, a REC tag with the time recorded (for example "REC 0:42") shows in the drawer and the detail header. Click it to see the recordings of this flow. A flow with an event recording waiting is marked Armed. These entries are only shown to logins that may record.
29.5.1 Quick Record
Next to the pages where you look at a stream, the Record button (a red dot, on narrow rows only the dot) records what you are looking at. It is on the flow detail page and drawer, the stream bar and the Services rows of the MPTS page, and the Teletext, Subtitles, Audio, Video, AV Monitor and Packet Timing pages. Logins that may not record do not see it.
The New recording dialog opens ready to go:
- The flow or service you were looking at is filled in. For an ST 2110 sender on the Video or AV Monitor page the video flow and the audio flow that belongs to it are both recorded; on the Audio page only the audio flow.
- The length is 1 min.
- The name comes from the page: the service name (or "Program" and its number), and on the Teletext and Subtitles pages the page, type or language, followed by the date and time.
Check the name and press Start recording. A quick record does not change the length that the normal New recording dialog remembers, so the next normal recording still opens with your own last choice. You can still change the length, the name or the type in the dialog before you start.
29.5.2 Service Recordings
A transport stream of a broadcaster often holds many programs, but a problem is usually in one. A service recording keeps only that program, so the file is small and shows only what you want to look at. Choose it by pressing Record on a service (29.5.1), or in the dialog under Advanced with Record: One service and the Service picker (29.4). When the dialog opens for a service it shows "Service" with the name and program number, and the Whole stream switch turns the choice back into a recording of the whole flow.
The file holds, for that program only:
- the PAT, rewritten so that it lists only this program (and the NIT entry);
- the PMT of the program, the PCR and every audio, video, subtitle and teletext PID it uses;
- the NIT, SDT, EIT and TDT, so a player still sees the names and the time;
- when the service is scrambled, the ECM PIDs named in its PMT. The EMM PIDs and the CAT are not kept, and neither are the other programs.
Each datagram keeps its headers, VLAN tag and arrival time, but is shortened to the packets of the program. A datagram without any packet of the program is left out, and the RTP numbering continues without a hole for it, so only the real gaps of the input show as gaps in the file. A damaged datagram (one that does not hold a usable transport stream) is left out too. The file replays and opens in Wireshark like every other recording.
If the program changes while recording (a new PMT version, or the PMT moves to another PID), the recording follows it. The list shows a service recording with "Service" and its name and program number on the flow line (29.2). Record again and Edit and record again keep the program.
29.6 Working with the List
The buttons at the end of a row depend on the state and on your role. A button that is not available is grayed out and says why when you point at it.
| Recording | Buttons |
|---|---|
| Running | Stop |
| Scheduled, waiting for an event | Edit, Cancel (asks "Cancel this recording?") |
| Not started | Start, Edit |
| Finished, limit, failed | Download, Record again (Arm again for an event recording) |
| Missed | Record again |
The three dots (More) hold Rename, Delete and, on finished recordings, Edit and record again..., which opens the dialog with the settings of the old recording so you can change them first.
- Edit opens the same dialog with the recording's settings. You can change a recording only while it waits. A running or finished recording can only be renamed; the file keeps its name.
- Record again starts a new recording with the same flow and settings, straight away. It is disabled when the flow is no longer on the probe.
- Download saves the pcapng file through your browser. Files can be many gigabytes.
- Delete removes the recording, its file and its picture after a confirmation. It cannot be undone.
- Several at once. Tick the boxes of finished recordings (the boxes of the others are disabled) and click Delete in the bar that appears; Clear drops the selection.
Recording files can be large. A 1 Gbit/s stream makes about 7.5 GB per minute. The dialog shows an estimate before you start.
29.7 Storage
The probe protects its disk in two ways, for every user:
- A storage limit for all recordings together. When a new recording needs room, the oldest finished recordings are deleted first. If the running recordings alone reach the limit, they stop with "Stopped: storage limit". When the limit is used up, new recordings are refused until old ones are deleted.
- A free disk space the probe keeps (the larger of a size in GB and a percentage of the disk, by default 2 GB or 5 percent). Below it, running recordings stop with "Stopped: disk space" and new ones are refused. This limit never deletes anything.
An Admin changes them with the gear button in the disk banner (Storage settings): Storage limit for recordings (GB), Keep this much disk free (GB) and Keep this much disk free (percent of the disk). The new values also apply to recordings that are already running.
Recordings stay on the probe until you delete them or the storage limit removes them. The list of recordings is part of the backup; the pcapng files are not.
29.8 REST API
Recordings can also be managed through the REST API. The complete description, with all fields, is in the REST API reference (docs/rest-api.md, "Recordings"). The main calls:
| Method | Endpoint | Description |
|---|---|---|
GET | /api/v1/recordings | List all recordings with state, progress and flows |
POST | /api/v1/recordings | Create a recording; "start": true also starts it |
PUT | /api/v1/recordings/{id} | Rename a recording, or change a waiting one |
POST | /api/v1/recordings/{id}/start | Start a recording |
POST | /api/v1/recordings/{id}/stop | Stop a running recording |
DELETE | /api/v1/recordings/{id} | Delete a recording, or cancel a waiting one |
POST | /api/v1/recordings/{id}/download-ticket | Get a one-time link to download the pcapng file |
GET | /api/v1/recordings/events | The events that can start a recording, per flow |
GET | /api/v1/recordings/storage | Storage limit, space used and the state of the disk |
PUT | /api/v1/recordings/quota | Set the storage limit and the free disk space (Admin) |
All calls need a Bearer token from POST /api/login. A script can also download the file directly with the token: GET /api/recordings/{id}/download.
Example: Record a Flow for Five Minutes
# Login
TOKEN=$(curl -sk https://probe:2300/api/login \
-X POST -H "Content-Type: application/json" \
-d '{"username":"admin","password":"admin"}' | jq -r .token)
# Create and start a manual recording of flow 12 in one call
curl -sk -H "Authorization: Bearer $TOKEN" \
-H "Content-Type: application/json" \
https://probe:2300/api/v1/recordings \
-d '{
"name": "TS 124.2 capture",
"scope": { "type": "flow", "flow_ids": [12] },
"trigger": { "type": "manual" },
"stop": { "type": "timed", "duration_sec": 300 },
"start": true
}'
29.9 Typical Use Cases
Incident investigation Arm an event recording on a flow with intermittent packet loss: choose "RTP sequence gap" and keep 10 seconds before the event. When the next loss happens, you get a pcapng file that starts 10 seconds before it, ideal for finding the cause in Wireshark.
Compliance archiving Schedule a daily 5-minute recording of your playout chain during prime time. Keep the files as evidence that the signal was clean at the point of contribution.
Network commissioning Record All traffic under Advanced for 60 seconds during a network change. Compare the captured traffic with your expected flow plan to verify routing, multicast joins and VLAN tagging.
Remote troubleshooting Use the REST API to start a recording from a remote NOC, then download the pcapng file for analysis without access to the probe.
30 Diagnostics
The Diagnostics page shows IP-Probe's internal processing for support engineers: the processing pipeline and the load of each processing core. It changes nothing on the probe. The page is for Admins only and is not in the sidebar: open it with Diagnostics at the top of Settings > System > Support (chapter 31.15). Other roles see Admin access required; a demo login never sees it.
The page has two tabs:
- Pipeline - the topology of the cores and the rings between them, the fill level of each ring, and the backend timing, port, worker and drop statistics.
- Cores - one line per processing core with its role, utilization, packet rate, ring fill and drops (the view that was the Cores widget on the Dashboard).
30.1 Core Layout
IP-Probe uses a multi-core architecture with dedicated cores for specific tasks:
| Core Assignment | Function |
|---|---|
| Master core (0) | Statistics aggregation, TimeWheel scheduling, periodic housekeeping. |
| TM / WebServer core (1) | Task dispatch, HTTP/WebSocket serving, REST API. |
| Worker cores (2..N) | Multi-threaded packet processing - protocol parsing, flow tracking, incident detection. |
| Input cores (N+1..) | Packet capture - one core per physical input port. |
| RecWriter core | Recording writer - pcapng file output. |
30.2 Core Utilization
The Cores tab displays each core with:
- Core ID - The CPU core number.
- Role - The function assigned to this core (Input, Worker, Master, etc.).
- Utilization - Current CPU usage as a percentage and a visual bar.
- Packets/s - The packet processing rate on this core (for Input and Worker cores).
- Ring fill and Drops/s - How full the ring in front of the core is, and the packets dropped there per second.
High utilization on Worker cores may indicate that the system is approaching its processing limit. High utilization on Input cores is expected for high-bandwidth links.
30.3 Idle Sleep Status
The page also shows the current idle sleep configuration:
- Enabled/Disabled - Whether adaptive idle sleep is active.
- Current state - Which sleep stage each core is in (Active, Light Sleep, Deep Sleep).
See Settings > System > Idle Sleep (chapter 31) for configuring idle sleep parameters.
31 Settings
The Settings section provides all system configuration and management functions. It is accessible from the sidebar under System > Settings and is organized into categories, listed in a side list at the left of the page under four headings: General (System, Inputs, Users, Web server, Audit log, Backup / restore, Update, Versions), Monitoring (Aliases, Groups, Checks, ETR 290, System health), System (Power; Admin only) and Tools (Signal generator). Every category you may use is listed; click one to open it, or reach it with the Tab key and press Enter. The current category is highlighted, and a category that needs a license option you do not have shows a lock. On a narrow window the list shrinks to icons (the name appears when you point at one), and on a phone it becomes a drop-down at the top of the page with the same headings. Inside a category, pages such as Checks keep their own tabs. The ETR 290 category holds the profiles, limits and behavior of the TR 101 290 checks of the MPTS streams (see section 18.5).
Settings written by a newer version. After a manual downgrade, a settings file may come from a newer IPProbe version. The page of those settings then shows a notice: "These settings come from IPProbe <version> (or: a newer IPProbe version). They are read as far as this version understands them and cannot be changed here." The page stays readable but its controls are inactive, and a save is refused with the same explanation. An Admin can choose Reset to defaults in the notice (after a confirmation in the page) to replace those settings with the defaults of this version; restoring a backup (31.9) replaces them as well. The users cannot be reset this way: restore a backup instead. The About page lists the settings concerned.
31.1 System
The System page (Settings > General > System) contains general probe settings, display options and thread tuning. Restart, reboot and power off are on their own page, Settings > System > Power (31.13); the System page links to it. The input names and their order have their own page, Settings > Inputs (31.1.8).
31.1.1 Probe Name
| Setting | Description | Access |
|---|---|---|
| Probe Name | The display name shown in the header bar (e.g. "Studio A", "Truck 2"). Maximum 32 characters, counted as characters (an umlaut counts as one). The default is "IPProbe". It is also the name of the probe in every alarm the alarm export sends (31.19), so choose a name that tells your probes apart. A change applies at once. | Admin only |
31.1.2 Login (Admin only)
| Setting | Description | Default |
|---|---|---|
| Browser Page Timeout | Logs out an open page that has no user activity for this long: Never, 1, 2, 3, 5, 10 or 12 hours. | Never |
With Never, a page that stays open stays logged in. With a time set, a page where nobody does anything for that long goes to the login page, which shows the reason (for example "Logged out after 2 h without activity"). The login ends for the REST API too, and pages opened with the same login are logged out together.
What counts as activity: everything you do on a page that sends a request to the probe, such as opening another page, starting or stopping audio or video, resetting a measurement or saving a setting. Data the probe sends, the regular login check and requests a page sends by itself (after a reconnect, or a page that only reads a status) do not count. Moving the mouse or scrolling within a page does not count either.
31.1.3 Incidents (Admin only)
| Setting | Description | Default |
|---|---|---|
| Run-In After Start or Link-Up (s) | For this many seconds after the probe starts, and after the link of a port comes up (also when system health restarts the port), the incidents of the flows of that input are counted but not reported. 0 to 60 seconds; 0 = off. | 5 |
Why the first seconds are quiet: while a network card starts to receive, it can lose packets for a short time (at 25 Gbit/s with hundreds of streams, up to some ten thousand packets in the first two seconds). The streams themselves are fine; the loss is the start of the capture. Without a run-in, that loss would show as RTP sequence errors, CC errors and ETR 290 errors on many streams at once, with log lines, history entries and possibly recordings started by an incident.
During the run-in of an input:
- the counters keep counting: packet counts, the totals of each incident type and the NIC loss counters (missed packets on the Inputs page) stay correct and visible;
- there are no log lines, no red severity, no incident history entries, no recordings started by an incident and no No Data reports for the flows of that input, and no "NIC Rx loss" log line for the port;
- ETR 290 does not evaluate the streams of that input: the ETR 290 page shows them as starting (start-up grace) and counts no errors, errored seconds or history episodes for that time. After a probe start or a restart by system health this holds from the first packet. After a link-up it holds from the first packet of a stream that comes back while the probe still sees the link as down, otherwise from the moment the probe sees the link up (it checks once a second); the first half second of every stream is in the start-up grace anyway. A link that drops and comes back within a second can therefore show a few ETR 290 errors at its return. When the run-in ends, ETR 290 starts its checks afresh: a table, section or PID that was missing during part of the run-in is reported only when it is still missing one full interval later.
- after a link-down, the first packet of each PID that comes back starts the continuity count anew: the packets lost while the link was down are not reported as CC errors (the outage itself is reported as No Data).
The run-in is timed with the probe's own running clock, not the time of day: setting or correcting the system clock (for example by NTP or PTP at boot) does not make it longer or shorter.
When the run-in ends, the log shows "Run-in done on input 1: incidents reported from now", and from the next second every incident is reported as usual. A stream that is still missing after the run-in (it did not come back after a link-up) is reported as No Data then. Streams that start later in the run are reported from their first packet. A new length is used from the next start or link-up; 0 also ends a run-in that is going on.
31.1.4 Display
| Setting | Description | Default |
|---|---|---|
| Dark Mode | Switches between light and dark color schemes. | On |
| Summary | Shows or hides the summary panel on the right side. | On |
| Show PIDs as | Shows PIDs as Decimal (256) or Hexadecimal (0x0100) on every page that names a PID. Saved per user. | Decimal |
| Activity hold (s) | How many seconds (1 to 60) a flow, IP address or MAC address stays shown as active after it was last seen in the once-a-second statistics. Slow traffic such as STP or LLDP (about one packet per second) would otherwise blink between active and inactive and make the lists jump. A stream that really stopped is shown as inactive after this time. Saved per user. | 5 |
31.1.5 Visibility
| Setting | Description | Default |
|---|---|---|
| Show Broadcast Icon | Displays a broadcast badge on broadcast flows. | On |
| Show Multicast Icon | Displays a multicast badge on multicast flows. | On |
31.1.6 Idle Sleep (Admin only)
Controls the CPU idle behavior of Input and Worker threads. When enabled, threads progressively back off when no packets are arriving, reducing CPU usage.
| Setting | Description | Default |
|---|---|---|
| Enable idle sleep | Master toggle for idle back-off. | On |
| Pause threshold | Idle iterations before switching to light sleep. | 10 |
| Sleep threshold | Idle iterations before switching to deep sleep. | 100 |
| Light sleep | Duration of light sleep pause (microseconds). | 10 |
| Deep sleep | Duration of deep sleep pause (microseconds). | 100 |
31.1.7 History Database
| Setting | Description | Default |
|---|---|---|
| ETR 290 history (days) | How long the episodes and the per-minute counts are kept (1 to 365). | 30 |
| Hourly totals (days) | How long the data summed per hour is kept (7 to 3650). | 365 |
| Maximum size (MB) | Size limit of the database. 0 is automatic: the smaller of 2 GB and 10 % of the disk. The oldest data goes first. | 0 |
| Incident history (days) | How long the incident episodes of every flow are kept (1 to 3650). See chapter 27.1. | 90 |
| Audit log (days) | How long the audit log is kept (7 to 3650). The size limit never deletes audit entries. | 365 |
| Configuration versions (count) | At least this many newest saved versions are kept per settings file (5 to 1000). See 31.10. | 50 |
| Configuration versions (days) | Every version of this many last days is kept as well (1 to 3650). | 90 |
| Log files: keep for (days) | How long the application log files of the probe (in its log folder) are kept (1 to 3650). They are not part of the database. | 30 |
| Log files: total size limit (MB) | Size limit of all application log files together (10 to 100000). The oldest files go first; the file in use is always kept. | 500 |
The probe starts a new log file every day at midnight (local time) and when a file reaches 100 MB; the new file names the previous one at its top. Old log files are deleted at start and with each new log file, with the values set at that time. Other files in the log folder are not touched.
History database of another version. A history database written by a newer IPProbe version (after a downgrade) is kept unchanged: the page says "This history database was written by a newer IPProbe version (schema N). It is kept unchanged; history is off until that version is installed again or the history is started fresh." Before an update converts an older database, the probe makes a copy of it; if that copy cannot be made (for example the disk is full), the database is not converted and the history is off for that start, which the page states. Copies of other versions lists these copies and the databases that a Go back kept aside, with the schema, the period they cover, the size and the IPProbe version that wrote them; an Admin can download or delete them (at most 2 of each kind are kept). When a period was recorded by another version and is only in such a copy, the page says "History from <from> to <to> is in a copy kept from IPProbe <version>"; the Update page shows the same for a Go back.
Only an Admin changes the settings; everyone can read them and the Database status: whether the history is recording, the size against the limit (a bar and a percentage, with a warning from 80 %), the file size, the oldest data (also the oldest incident and the oldest audit entry), the check at start, the last full check, the records waiting to be stored and the records dropped, the numbers of ETR 290 episodes, incidents, audit entries and configuration versions (counts stay complete; only episode detail is capped; an audit entry that could not be stored is kept in a file on the probe and stored at the next start), the last cleanup and the last error. A value outside the limits is refused with the reason under its field.
Maintenance (Admin only), below the status:
- Check now checks every page of the database file. History keeps being recorded meanwhile; a large file takes some minutes (the check stops after 9 minutes). The result (OK, Damaged with the problems found, Stopped or Failed) and the time it took are shown and kept as the Last full check. A damaged file is not replaced by the check: use Recover or Start fresh.
- Start fresh replaces the database with a new, empty one, after a confirmation in the page. The current file is not deleted: it is kept beside the new one as
history.db.old-<time>. The incidents, audit log, ETR 290 history and configuration versions start empty; recording pauses for a moment, monitoring continues. - Kept copies lists the copies kept beside the database, newest first, with their size and time: Damaged copies were renamed at start because the file was damaged; Replaced copies are the files that Start fresh, Recover or a restore replaced. At most 3 of each are kept, and all copies together take no more disk space than the size limit of the database; the oldest goes first. The newest damaged copy (for Recover) and the newest copy stay even beyond that. Download saves a copy as a zip (for support), Delete removes it after a confirmation.
- Recover (on a damaged copy, after a confirmation) copies what is still readable from that copy into a new database, adds the history recorded since the damage, and makes that the database in use. The copy itself stays as it is; the database in use before is kept as a replaced copy. The result lists, per table, the rows read back and the rows added. Rows the damage destroyed cannot be read back. Recover needs a function of the database library of the probe; where the library lacks it, the button is disabled and the page says so.
Only one of these actions runs at a time, and not while a backup or restore runs: the page shows what is running, and a refused action says so.
If the probe finds the database damaged, an Admin sees a notice below the header until it is acknowledged. At start the damaged file is renamed (its name is in the notice) and a fresh database is made. If the damage is found while running, history recording stops until the next start; monitoring continues. Acknowledge hides the notice, also after a restart.
31.1.8 Inputs
The Inputs page (Settings > General > Inputs) lists all detected network inputs (capture ports) and lets an Admin rename, describe, reorder and switch them:
| Column | Description |
|---|---|
| Input | The current display name, a link to the card of the input on the Inputs page, and below it the network card and its PCI address (or "Virtual input"). |
| Link | The link state of the port with its speed, for example UP - 10G, No link, Link lost or STOPPED, as on the Inputs page. |
| Name | Editable short name (max 64 characters). Leave empty for the default name Input_<n>. |
| Description | Editable description (max 128 characters). It is shown under the name on the card of the input on the Inputs page. |
| Active | Toggle to include or exclude this input from the monitoring view. |
Order. The rows are in the order the Inputs page and every other list of inputs show. An Admin drags a row by its handle at the left to a new place, or moves it with the up and down arrow buttons (which also work with the keyboard), and selects Save; the position of the row is the order. Other users see the order but cannot change it.
Show transmit. The check box above the rows shows the Transmit chart next to Receive on every card of the Inputs page, for all users. It is off by default, because a capture port does not transmit; the Receive panel then uses the full width of the card. Like the order, it changes when an Admin selects Save; Discard changes puts it back.
The name of an input is shown everywhere the input appears: the Flows, Endpoints and Devices pages, the filter bar, the dashboard widgets, the AV Monitor, recordings and reports. A rename shows on all open pages as soon as it is saved, without reloading the page. An input without a name is shown as Input_<n>, where <n> is the input number. On the Flows page, flows grouped by input get one group per input, so two inputs with the same name stay separate groups.
31.2 Aliases
The Aliases page manages user-defined names for IP addresses, MAC addresses, flows, and VLANs. Aliases provide human-readable labels that appear throughout the interface.
Aliases can be configured for:
| Entity | Fields |
|---|---|
| IP addresses | Alias (name), Description. |
| MAC addresses | Alias (name), Description. |
| Flows | Alias (name), Description. |
Click Add to create a new alias, or click an existing alias to edit it. Click Save to persist changes.
31.3 Groups
The Groups page manages flow groups. Groups allow operators to organize flows into logical collections (e.g. "Channel 1", "Studio A", "Backup feeds") for:
- Filtering the Flows page by group tab.
- Scoping recordings to a group of related flows.
- Organizing dashboard views.
Each group has:
| Field | Description |
|---|---|
| Name | The group display name. |
| Description | Optional descriptive text. |
| Members | The list of flow IDs or matching criteria that belong to this group. |
A group can also have the rule Do not monitor the flows of this group: its flows are then Not Monitored (see 31.11).
31.4 Checks
The Checks page decides when IP-Probe raises an incident and how severe it is. It has two tabs: Checks, with every check of the probe and its value for all flows, and Overrides, with what differs for a group or a flow (and, for Subtitles stopped, an MPTS program or a PID). The page has the version history and Undo last change of the other settings pages.
31.4.1 The Checks tab
The checks are grouped in cards in the order of the signal path: Signal and network, Packets and timing, Redundancy (ST 2022-7), PTP, Picture (the video transport of ST 2110-20 and -22), Audio, Timecode and ANC, Captions and subtitles and Transport stream (MPTS). The ETR 290 checks of a transport stream have their own page (Open ETR 290 in the Transport stream card).
Every check is one row:
| Column | What it does |
|---|---|
| Check | The name and what it detects. The info icon opens its section below; clicking the name opens its details. |
| On | Switches the check on or off for all flows. |
| Severity | Critical, Major, Minor or Info (section 31.4.5). A check with two limits shows Minor / Major. |
| Settings (all flows) | The values with their units. An empty field uses the built-in default, shown in gray in a dashed box; a value out of range is outlined in red with the allowed range. |
| Overrides | How many overrides change this check; a click shows them on the Overrides tab. |
Under a check its profiles are listed (section 31.4.2), with Add profile. A check whose all-flows value, on/off, severity or trigger differs from the built-in default has a bar on its left and a changed tag.
Above the cards: Find a check, the filters Changed from default, Off and With overrides, chips that jump to a card, and the severity legend. A card where nothing differs from the defaults and no override changes a check starts collapsed and shows how many checks it has of each severity. Defaults in a card header puts the all-flows values, on/off, severities and triggers of its checks back to the built-in defaults; profiles and overrides stay.
The details of a check. Clicking the name of a check opens its details next to the list of checks: what it detects, its all-flows values and severity, its trigger, its profiles (values, severity, where they are used and on how many flows), Where it differs (every override that changes it and what it wins over), and Last 24 hours, a lane of its incidents (muted where an incident was below its trigger). On a phone the details fill the screen; All checks goes back.
Raise the alarm only after several times. A check that raises short events (lost packets, a TTL change, an AES3 error, the PTP checks and similar) has Raise at once in its row. Open it and switch on Only after several times to raise the alarm only when the check fires a number of Times (2 to 100) within a Time window (1 s to 24 h, in s, min or h). The first times within the window are listed on the Incidents page marked below trigger: they are not an alarm, are not exported and give no mark on a multiviewer tile. The trigger is off by default. A profile and an override can set their own trigger.
Changes count only after Save. Until then the bar at the bottom shows how many checks changed and which; Discard puts the page back to the checks as they were last loaded or saved. A value the probe would refuse is listed above the bar, and Save waits until it is corrected.
31.4.2 Profiles
A profile is a named second setting of a check, for example "IPTV slow" for No data with a silence limit of 5 s. It has its own values (empty fields use the all-flows value) and can have its own trigger. A profile applies only where an override uses it. The severity in the profile dialog is set on every override that uses the profile. Click a profile chip to edit or delete it; the overrides of a deleted profile fall back to the all-flows values.
31.4.3 The Overrides tab
An override changes some checks for one target: a group (Settings > Groups), a flow, an MPTS program of a flow or a PID of a flow. The most specific setting wins: All flows, then Group, then Flow, then MPTS program, then PID. An override changes only the checks it names; every other check comes from the level above. MPTS program and PID overrides apply to Subtitles stopped (the DVB subtitle PIDs of an MPTS stream) and to the picture checks Black picture, Frozen picture, Frozen and silent and Picture area judged (the video of a program); the other checks are judged per flow, so for them use a group or a flow override. The target choices MPTS program and PID are offered only while every check of the override is one of those, and they then list only those checks. For each check an override can:
- use a profile,
- set own values (empty fields keep the value of the level above),
- change only the severity,
- switch the check Off for the target.
Add override asks for the target (for a flow, program or PID: pick the flow from the flow picker; the override keeps the flow by its identity, so it stays with the flow after a restart), an optional note, and the checks it changes. The overrides are listed by kind of target. Each row shows its changes as chips and a switch that turns the whole override off without deleting it; the arrow opens the row, which shows the all-flows value struck through next to the new one, and X removes one change. By check lists, per check, every override that changes it.
Group order. A flow can be in more than one group. Where two groups set a value for the same check, the group higher in the list wins. Drag a group row, or use its up and down arrows (also with the keyboard), to change the order; an override for several groups moves its groups together. Two groups with the same priority are listed as the probe decides between them: the override with the higher rule priority first, then the one whose override id sorts first. The order is stored as the group priority in Settings > Groups and saved with the checks.
What one flow gets. Pick a flow to see, for every check, whether it is on, its severity and values, and where each comes from: Built-in, All flows, a group, this flow, a program or a PID (Subtitles stopped only), and the profile. The groups of the flow that have an override are shown in their order. By default only the checks that differ from All flows are listed (a value from a group, the flow, a program or a PID); show all lists every check. It shows the saved settings: save first to see a change there. The History tab of a flow has a link that opens this panel for that flow.
31.4.4 Check types
Every check type of the probe is described below, with its name in checks.json in brackets. On the Checks tab the info icon of a check shows a short description; clicking it opens its section here.
31.4.4.1 No data (no_data)
Raises an incident when a flow sends no packets for longer than the silence limit, for example because a sender or a link stopped. max_silence_ms is the silence limit in milliseconds (default 1000); 0 turns the check off. The incident ends when packets arrive again.
31.4.4.2 Bitrate low (bitrate_low)
Raises an incident when the bitrate of a flow drops under min_bps (bit/s), for example when an encoder fails or a stream thins out. The incident ends when the bitrate is back above the minimum plus hysteresis_bps, so a bitrate that hovers around the limit does not raise one incident after the other.
The bitrate is the one the Flows page shows: the bits of the flow in the last second. It is judged once a second for a flow that sent packets in that second, and not in the first second after the flow appears (that second is only partly counted). A bitrate between the minimum and the minimum plus the hysteresis keeps the state it had. min_bps of 0 (or no min_bps) turns the check off; hysteresis_bps is optional and 0 by default. A flow that stops sending is reported by the No Data check; the bitrate incident stays open until the flow sends again with a bitrate back over the limit.
31.4.4.3 Bitrate high (bitrate_high)
Raises an incident when the bitrate of a flow goes over max_bps (bit/s), for example a sender that sends too much and could flood the network. The incident ends when the bitrate is back under the maximum minus hysteresis_bps. The bitrate is judged as for bitrate_low; max_bps of 0 (or no max_bps) turns the check off.
31.4.4.4 TTL changed (ttl_changed)
Raises an incident when the TTL of the packets of a flow changes. A new TTL usually means that the route through the network changed (another router hop, a failover). Discovery and local control multicast are left out by default (see 31.4.5).
31.4.4.5 AES3 errors (aes3_errors)
Raises an incident on parity, flag, block start and channel status errors of ST 2110-31 AES3 audio. These errors show that the AES3 transport is damaged, so the audio or the data in it (for example Dolby E) can be corrupt. "enabled": false turns the check off.
31.4.4.6 AES3 status changed (aes3_status_changed)
Raises an incident when the channel status of an ST 2110-31 AES3 pair changes, for example a switch between PCM audio and data, or another sampling frequency. Use it to notice a source that changes its output unexpectedly. "enabled": false turns the check off.
31.4.4.7 ST 2022-7 class exceeded (st2022_7_class_exceeded)
Raises an incident when the path differential of an ST 2022-7 pair (the time between the same packet on the two paths) goes over the limit of the receiver class in class ("A", "B", "C" or "D", default "B"). A receiver of that class could then no longer merge the two paths without a gap.
31.4.4.8 Timecode jump (anc_tc_jump)
Raises an incident when the ancillary timecode (ST 2110-40) jumps or goes backwards between two frames. This catches edits, source switches and a broken timecode generator. See 16.2 for the timecode panel.
31.4.4.9 Timecode stall (anc_tc_stall)
Raises an incident when the ancillary timecode repeats while the video moves on, which shows that the timecode generator stopped. frames is the number of repeated frames before the incident is raised (1 to 250, default 2).
31.4.4.10 Timecode missing (anc_tc_missing)
Raises an incident when a timecode type that was present is no longer sent, for example a timecode that is lost on the way through the plant. missing_ms is the time without that timecode before the incident is raised (40 to 60000 ms, default 1000).
31.4.4.11 Timecode off PTP time (anc_tc_ptp_offset)
Raises an incident when the timecode differs from the time address that PTP time gives the frame, which shows a timecode generator that is not locked to the house clock. max_frames is the largest accepted offset in frames (default 1), hold_s the seconds over that limit before the incident is raised (default 5). utc_offset_min (local time minus UTC in minutes, 60 = UTC+1) and dst (true adds one hour for daylight saving time) set the local time the timecode is compared with; without them it is compared with UTC.
31.4.4.12 Subtitles stopped (subtitles_stopped)
Raises an incident when a DVB subtitle PID of an MPTS flow that sent subtitles before sends none for the delay while the rest of the flow goes on. The incident ends after 2 seconds of subtitles again; a single subtitle after the stop does not end it. A PID that never sent subtitles does not raise it, and a flow that stops altogether is No data, not this. delay_s is the time without subtitles before the incident is raised (5 to 600 s, default 30). The incident names the PID and stays open in the incident history until the subtitles are back. This is the only check that takes an MPTS program or PID override (31.4.3): for example a longer delay on one subtitle PID, or Off for a program without subtitles at night.
31.4.4.13 Timing limits
The six timing checks below judge the packet timing of a flow once a second against two limits, Minor above and Major above (in checks.json the warning_* and error_* values). A value above a limit reaches its level; a value at the limit does not. A limit of 0 turns that level off, and an empty limit keeps the built-in default of the flow type.
The PCR frequency offset and drift checks are off until they are switched on (as is the ETR 290 PCR accuracy check 2.4, set under Settings > ETR 290), because the probe's clock is not locked to PTP yet. The incident opens as Minor in the first second over the lower limit, goes to Major when a second passes the upper limit, and ends after 60 seconds in a row under the lower limit (a lower level comes after 60 seconds under the higher one), so a value that hovers around a limit does not raise one incident after the other. The Packet Timing page and the Timing tab of the MPTS page color the values and draw the limit lines with the same limits.
31.4.4.14 IAT peak-to-peak (iat_pp_high)
The IAT peak-to-peak of a second (the longest minus the shortest inter-arrival time) against warning_us and error_us (microseconds), for every flow type. For flows with a packet time (ST 2110-30 and -31) warning_pct and error_pct set the limits in percent of the packet time; when both a time and a percentage are set, the lower limit applies. A time alone replaces the default percentage of that level, so it can also loosen the limit. Defaults: ST 2110-30 / -31 Minor above 50 % and Major above 100 % of the packet time; every other type off (an ST 2110-20 sender is judged by its ST 2110-21 verdict instead, see 21).
31.4.4.15 RTP jitter (rtp_jitter_high)
The RTP interarrival jitter (RFC 3550, the highest value of the second) against warning_us and error_us (microseconds), for flows that carry RTP with a known clock rate. Off by default.
31.4.4.16 MDI delay factor (mdi_df_high)
The MDI delay factor (RFC 4445) of a transport stream against warning_us and error_us (microseconds). Defaults: Minor above 150 ms and Major above 300 ms. The delay factor of a variable bitrate stream is indicative only and not judged. Streams that are normal bursty (a remultiplexer or a software encoder that sends in bursts) stay under these values; for a managed network with a tight receiver buffer, lower the limits with an override for that flow, for example warning_us 20000 and error_us 50000.
31.4.4.17 MDI media loss (mdi_mlr)
The MDI media loss rate (RFC 4445) of a transport stream: the TS packets lost in a second. warning_pkts and error_pkts are the number of lost packets in a second at which the level is reached. Default: Major at 1 lost packet (any loss), no Minor limit.
31.4.4.18 PCR frequency offset (pcr_freq_offset)
The frequency offset of the PCR clock of a transport stream from 27 MHz (TR 101 290 PCR_FO, the PCR PID with the largest offset), against warning_hz and error_hz (Hz, either sign). Off by default: the offset is measured against the packet clock of the probe (the clock of the network card), which has no 27 MHz or PTP reference, so an offset of a few tens of ppm may be the probe's own clock. Switch it on on a probe whose clock is locked to PTP; the defaults are then Major above 810 Hz (TR 101 290) and no Minor limit. The PCR table on the Packet Timing and MPTS Timing pages shows the offset with the TR 101 290 colors either way.
31.4.4.19 PCR drift (pcr_drift)
The drift rate of the PCR clock (TR 101 290 PCR_DR, the change of the frequency offset per second, the PCR PID with the largest drift), against warning_mhz_s and error_mhz_s (millihertz per second, either sign). Off by default, for the reason given at the PCR frequency offset: enable it on a probe whose clock is locked to PTP. The defaults are then Minor above 0.075 Hz/s, the TR 101 290 limit of the broadcast profile, and no Major limit. The PCR table shows the drift with the TR 101 290 colors either way.
31.4.4.20 Captions stopped (captions_stopped)
Raises an incident when a caption service of a flow that was present for at least 10 seconds has no captions for the delay while the rest of the flow goes on: CEA-608 field 1 or 2, CEA-708 or OP-47 subtitles on ST 2110-40 and ST 2022-6, and the 608 and 708 closed captions in the video of an MPTS. The incident ends after 2 seconds of captions again. A flow that stops is No data, not this; a service that was never seen does not raise it, and Stop expecting on the flow page (Admin and NOC) takes a service out until it has been present for 10 seconds again. delay_s is the time without captions before the incident is raised (5 to 600 s, default 30). See 25.5.
31.4.4.21 Checks without settings (rtp_seq_gap and others)
These checks have only on/off, a severity and the trigger "only after several times"; they are on by default.
| Check | In checks.json | Raises an incident when |
|---|---|---|
| Source changed | ssrc_collision | the RTP SSRC of the flow changes |
| Payload type changed | pt_changed | the RTP payload type changes |
| DSCP changed | dscp_changed | the DSCP (QoS class) of the packets changes |
| Packets lost | rtp_seq_gap | RTP sequence numbers are missing |
| Extended sequence break | st2110_seq_gap | the ST 2110 extended sequence number breaks |
| Duplicate packets | rtp_seq_duplicate | the same sequence number arrives twice |
| RTP time jump | rtp_ts_jump | the RTP timestamp leaves the arrival time |
| Loss after merge | st2022_7_loss_after_merge | a packet is missing on both ST 2022-7 paths |
| Path loss | st2022_7_path_loss | one ST 2022-7 path loses packets the other still has |
| Path differential step | st2022_7_skew_change | the delay between the two paths steps |
| Sync lost | ptp_sync_loss | the Sync messages of the grandmaster stop |
| Grandmaster changed | ptp_gm_change | another clock becomes grandmaster |
| More than one master | ptp_multi_gm | two masters send in one domain |
| Sync messages lost | ptp_seq_gap | Sync messages are missing |
| Correction field high | ptp_corr_high | the correction field is large |
| Marker error | st2110_marker_error | an ST 2110-20 frame ends at a wrong packet count |
| Receiver buffer overrun | st2110_vrx_overrun | the packets are too bursty for the wide profile (ST 2110-21) |
| JPEG XS frame gap | st2110_22_f_gap | the ST 2110-22 frame counter skips |
| JPEG XS marker missing | st2110_22_marker_missing | a frame ends without a marker |
| JPEG XS size mismatch | st2110_22_size_mismatch | the codestream is not the announced size |
| ANC checksum error | anc_checksum_error | an ST 291 checksum word is wrong |
| ANC checksum convention | anc_checksum_convention | the checksums use another parity convention |
31.4.4.22 Black picture (video_black)
Raises an incident when the picture of an ST 2110-20 or ST 2022-6 flow stays black. The probe judges one picture of each flow once a second, on a grid of sampled luma values (on a 1080 picture every 4th pixel of every 8th line), inside the area set under Picture area judged (31.4.4.26). A sample is black when its luma is at or under the black level: level 90 (the default) means 10 % luma or less (0 to 100). The picture is black when area_pct of the judged samples are black. The field is called Area that must be black on the Checks page: 1 to 100 % of the judged area, default 100 % (all of it must be black, as DVBMosaic judges). The 98 % chip next to the number lets a small logo or clock on an otherwise black picture still raise Black; other tools use 95 to 98 % for the same reason. Any whole number from 1 to 100 can be typed. The incident evidence lists the value under the settings used (when it is below 100 %). The incident is raised when the picture stays black for duration_ms (1 to 600 s, default 5 s). The alarm clears at the first picture that is not black. A picture that is black and does not move is reported as frozen only (31.4.4.23), not as both. Major by default; it takes the trigger "only after several times".
31.4.4.23 Frozen picture (video_freeze)
Raises an incident when the picture of an ST 2110-20 or ST 2022-6 flow does not move. Once a second the sampled luma grid of the judged area is compared with the previous picture and with an older anchor picture, so a slow fade or drift is seen as motion too. A sample moves when its luma changes by more than diff_pct (1 to 50 %); the picture is still when fewer than min_moving_samples samples move (1 to 10000). The incident is raised when the picture stays still for duration_ms (1 to 600 s, default 10 s) and clears at the first moving picture. From 4 seconds of a still picture the Multiviewer and the AV Monitor show the outlined hint still N s; the hint is never an alarm, it is not counted and not exported.
Sensitivity (sensitivity) chooses the two numbers at once; the duration stays its own value:
| Sensitivity | Motion threshold | Moving samples needed | Use |
|---|---|---|---|
| Seldom | 5 % | 1 | any small motion counts as moving: fewer alarms |
| Normal (default) | 10 % | 1 | most pictures |
| Often | 15 % | 20 | a small moving clock or ticker does not hide a freeze: more alarms |
| Custom | as typed | as typed | the two numbers are editable |
Slate colors (slate_colors). Up to 4 named solid colors, for example the station slate or the color bars a source sends while it is idle. Each has a name, a Y' Cb Cr code in 10-bit values (Y' 64 to 940, Cb and Cr 64 to 960, BT.709), a tolerance (1 to 200 codes per component) and a severity. A still picture whose samples all lie within the tolerance of a named color is reported as Frozen on name, with the severity of that color; Off raises no incident at all, for a slate that is expected. The list shows each color as a swatch.
A frozen black picture is reported as frozen only. Major by default; it takes the trigger "only after several times".
31.4.4.24 Frozen and silent (video_freeze_silent)
Raises its own incident, next to the freeze incident, when a frozen picture comes with silent audio: often a stalled playout or a dead source. It works only for a video flow with a manual AV pair (Settings > Aliases, AV pairs, or Pair in the AV Monitor, see 24); the automatic matching of the AV Monitor is not used for alarms. While the freeze alarm of the video is raised, the peak of the first channel of the paired audio must stay under silence_dbfs (-120 to -20 dBFS, default -60) for duration_ms (1 to 600 s, default 5 s). The incident clears when the picture moves or the audio is back. The freeze incident keeps its own severity; a tile shows the stronger of the two, and the alarm export sends both. Critical by default.
31.4.4.25 Out of gamut (EBU R103) (video_gamut)
Raises an incident when too much of the judged area of an ST 2110-20 or ST 2022-6 picture is outside the limits of EBU R103, judged on the 10-bit values once a second, inside the area set under Picture area judged (31.4.4.26). A pixel is outside when R', G' or B' is below rgb_min_pct (default -5 %) or above rgb_max_pct (default 105 %), or Y' is below y_min_pct (default -5 %) or above y_max_pct (default 105 %), the limits of EBU R103 version 3.0 (20 to 984 at 10 bits). The incident is raised when more than area_pct of the judged area (Area outside, 0.1 to 100 %, default 1 %) stays outside for duration_ms (1 to 600 s, default 2 s), and clears when it stays under for clear_ms (default 2 s). matrix and range (in the detail pane) say how Y'CbCr is turned into R'G'B': Automatic takes the matrix from the picture size and the narrow range; set them for a flow that uses another one. The incident Picture out of gamut says how much of the picture is outside and which limit, for example "1.8 % of the picture outside R103 (R' above 105 %)". R 103 filter (lowpass, on by default) filters the picture as EBU R 103 recommends before it is judged, so single overshooting pixels at sharp edges do not count; off judges every pixel as sent. Off by default; compressed video (ST 2110-22, MPTS programs) is not judged. Minor by default. False color, Gamut in the AV Monitor shows where the pixels are (24.3).
31.4.4.26 Picture area judged (video_area)
A setting for every picture check (black, frozen, frozen and silent), not a check of its own: it has no switch, no severity and raises nothing. skip_top_lines leaves out the top lines of the active picture, where closed caption and VITC lines are carried (0 to 100, default 10). The margins margin_l, margin_r, margin_t and margin_b leave out a part of each side in percent of the width or height (0 to 45 %, default 0), for example a logo, a ticker or a clock, which would otherwise keep a frozen picture "moving". With a top margin and skipped lines, the larger one counts. Like any check it can be overridden for a group or a flow.
31.4.4.27 Evidence per episode (video_evidence)
A setting, like the picture area: what the probe keeps with every black or freeze episode.
pictures(default on): three small JPEG pictures. The last moving (last good) picture, at most 5 seconds older than the start of the episode; the first frozen (first black) picture; and the latest picture, refreshed while the episode is open.trace_s(30 to 600 s, default 120): the trace of moving samples and mean luma per second before the raise; it goes on until the clear.
The evidence is shown when you open the episode on the Incidents page (27.1.1) and is removed together with the episode. Only flows whose check and evidence are on pay for the pictures; with Off the trace is still kept.
31.4.4.28 Detector decodes (video_decode)
A setting for the whole probe, in the box Compressed video (ST 2110-22, MPTS) of the Picture card: how many pictures a second (decodes_per_s, 1 to 50, default 10) the probe decodes to judge the black and freeze checks of compressed video, the video of the MPTS programs and ST 2110-22 (JPEG XS) flows. Only All flows counts; there are no profiles or overrides for it.
Compressed video is judged on I-frames. A picture the probe already decodes for a viewer (a thumbnail, the multiviewer) is reused; the other programs get a reduced decode in turn. With 48 programs on and 10 decodes a second each program is judged about every 5 s, so an alarm comes about 10 to 15 s after the picture stopped (the freeze raise time plus one round). The text under the field works this out for the programs judged now and the value you type. More decodes judge each program more often and cost more CPU.
Incidents of an MPTS program name the program and the service ("program 13001 ORF 2: ...") and carry the video PID; the evidence says Judged on with how often the stream sent an I-frame and how often the probe judged one ("I-frames every 1 s, judged about every 5 s"; only the second while the I-frame rhythm is not known).
The text under the field also estimates the CPU the decodes use: the value times the average time of one decode as measured now ("At 10 a second the decodes use about 0.05 of a core"). A judged I-frame of HD H.264 takes about 5 ms, one of HEVC 2160p about 40 ms, so 50 decodes a second of UHD HEVC need about two cores.
Frozen and silent of an MPTS program uses the program's own first audio track: the probe decodes it only while the picture is frozen, for at most two programs at once (one audio decoder place always stays free for a player). A program whose audio stays live for 10 seconds gives its decoder back and is listened to again after 30 seconds, so a third frozen program also gets its turn.
31.4.4.29 Loudness standard (loudness_standard)
A setting for all loudness checks of the Audio card. The loudness of every ST 2110-30, ST 2110-31 and AES67 audio track is measured as ITU-R BS.1770-5 and EBU Tech 3341 / 3342 describe it, one track per channel group of Settings > Audio channels, else one per channel pair. A pair of an ST 2110-30 flow that carries Dolby E or AC-3 data is measured only while you listen to it. The audio of an MPTS program (MPEG audio, AAC, AC-3, E-AC-3 and SMPTE 302M with PCM, Dolby E or AC-3 pairs) is measured where Measured is set to Every audio track, MPTS too: each audio PID is one track (a 302M PID one track per pair or Dolby E program), decoded with dynamic range control off, at about 0.3 % of one CPU core per audio PID.
- Standard: EBU R128 (-23 LUFS +/-1 LU, -1 dBTP, the default), ATSC A/85 (-24 LKFS +/-2 LU, -2 dBTP), ARIB TR-B32 (-24 LKFS +/-1 LU, -1 dBTP), Free TV OP-59 (-24 LKFS +/-1 LU, -2 dBTP) or Custom. The standard fills Target, Tolerance and Max true peak; with Custom you set them yourself.
- Short-term window: 3 s (EBU) or 10 s (ATSC).
- Measured: ST 2110 audio (the default: every ST 2110-30, ST 2110-31 and AES67 track, no MPTS audio), Every audio track, MPTS too, or Off. MPTS audio is off by default: switch it on in an override for a group, a flow or one MPTS program (or one audio PID) that you want judged unattended. Set Off in an override for a group whose tracks need no loudness checks; nothing is measured for them then. At most 32 audio PIDs of one MPTS flow are measured.
Every loudness limit is written relative to the target ("target +5 LU") with the absolute value beside it ("= -18 LUFS"), so a group that switches to another standard moves all its limits at once. The ladder under the row draws the limits on one LUFS scale. Profiles (for example an A/85 profile for a group of US affiliates) sit on this row.
31.4.4.30 Too loud (loudness_short_high)
Raises an incident when the short-term loudness of a track stays above the limit: Above target +5 LU (-18 LUFS with R128) for Raise after 10 s. The alarm clears when the short-term loudness is under Clear below (target +3 LU, at most the Above value). Major by default.
31.4.4.31 Too quiet (loudness_short_low)
Raises an incident when the short-term loudness stays under Below target -10 LU (-33 LUFS with R128) for Raise after 60 s. Audio under the silence level is left to Silence (31.4.4.37), so one quiet scene is never two incidents: Too quiet ends when the track goes silent. Minor by default.
31.4.4.32 Off target (loudness_integrated)
Raises an incident when the integrated loudness of the Window is outside target +/- Tolerance (empty: the standard's tolerance) for Hold (1 min). Window: Sliding 10 min (the default), Sliding 30 min or Since the last reset (a long silence starts it anew). The window is judged once it holds 60 s. Each ad break (SCTE-104) judges every ad break as a whole; it needs the SCTE-104 check, which this version does not have yet, so the choice is not available. Minor by default.
31.4.4.33 Momentary peaks (loudness_momentary_high)
Off by default. Raises an incident when the momentary loudness (400 ms) rises above Above target +8 LU Times times (5) Within a window (60 s); it clears when fewer rises remain in the window. Minor.
31.4.4.34 True peak over (true_peak_high)
Raises an incident when the true peak (the peak between the samples, 4x oversampled) rises above Above (empty: the standard's maximum, -1 dBTP with R128) Times times (3) Within 60 s; it clears when fewer overs remain in the window. Minor by default.
31.4.4.35 Loudness range (loudness_range)
Off by default. Raises an incident when the loudness range (LRA, EBU Tech 3342) of the Off target window stays under Below (3 LU) or over Above (20 LU) for a minute. Minor.
31.4.4.36 Dialnorm mismatch (dialnorm_mismatch)
Raises an incident when the dialogue level an AC-3 or E-AC-3 track announces (dialnorm) differs from its measured integrated loudness by More than 2 LU for a minute, measured with dynamic range control off over the Window. PCM, MPEG audio and AAC tracks have no dialnorm; an AC-3 or E-AC-3 PID of an MPTS program whose loudness is measured is judged unattended (AC-3 carried in SMPTE 302M is not). Minor by default.
31.4.4.37 Silence (audio_silence)
Under the sub-heading Silence and transport. Raises an incident when every channel of a track stays at or under Below (-60 dBFS) for Raise after 10 s; it clears with the first audible audio. Channels: Any pair silent (any track of the flow) or All channels silent (only when every track is silent). Channels that never carried audio are not judged. Major by default.
31.4.4.38 Teletext lost (teletext_lost)
One check for both kinds of teletext source; its settings apply to both. Raises an incident when a teletext PID of an MPTS flow that sent PES before sends none for the delay while the rest of the flow goes on, or when an ST 2110-40 flow that carried teletext (OP-47) sends no teletext packets for the delay. The incident names the PID or the flow and ends after 2 seconds of data again. A flow that stops altogether is No data, not this. delay_s is the time without teletext data before the incident is raised (5 to 600 s, default 5). Major by default.
31.4.4.39 Teletext page missing (teletext_page_missing)
Raises an incident per page when a subtitle page that the PMT signals as a subtitle page (teletext type 2 or 5) is not sent for 30 seconds while other subtitle pages of the PID are (the rule of the standard), or has not been sent once 10 minutes after the PID started. The incident names the page and ends when the page is sent again. It has no settings besides the switch and the severity. Minor by default.
31.4.4.40 Teletext subtitle hidden (teletext_subtitle_hidden)
Raises an incident per page when a subtitle page with text is sent with the Inhibit Display bit (C10) set, so a receiver does not show it. By default the first occurrence raises (count 1) and the incident ends 30 seconds (window_s) after the last. Major by default.
31.4.4.41 Teletext subtitle late (teletext_subtitle_late)
Raises an incident per page when subtitle updates are complete more than 100 ms after their PTS (EN 300 472, clause 5) count times within window_s (default 3 times in 60 seconds). The incident says how late the last update was. retention (off by default) counts updates that were held before their PTS for longer than the decoder model of the standard allows too. An update that is late by 100 ms or less is counted on the flow page only. Minor by default.
31.4.4.42 Teletext subtitle flags (teletext_subtitle_flags)
Raises an incident per page when a page that the PMT signals as a subtitle page is sent without the Subtitle bit (C6), when a subtitle page is sent with the Newsflash bit (C5), or when a subtitle page is carried in a data unit meant for non-subtitle data. By default the first occurrence raises and the incident ends 30 seconds after the last. Minor by default.
31.4.4.43 Teletext text outside the box (teletext_unboxed)
Raises an incident per page when a subtitle page (C6) has text outside the box set by Start Box and End Box (the warning of other analyzers is called unboxed). By default 3 times in 60 seconds raise it. Minor by default.
31.4.4.44 Teletext subtitle erase (teletext_erase)
Raises an incident per page when a subtitle page is sent with no rows and no Erase Page (C4) after a page with text, so a receiver keeps showing the old text, or is sent again without C4 with only some rows so that rows of the text before stay on the screen. By default 3 times in 10 minutes (600 seconds) raise it. A repeat of the same text with C4 is not an error and is only counted. Minor by default.
31.4.4.45 Teletext transmission errors (teletext_transmission)
Raises an incident per PID when the data units of the teletext PID are damaged or out of order: a framing code that is not 0xE4, rows of a magazine after its page header was lost to a Hamming error, a packet of another magazine within a page sent in serial mode, or a change of the serial mode (C11) between the headers of the service. Single bit errors that are corrected are counted only. By default 5 times in 60 seconds raise it. Minor by default.
31.4.4.46 Teletext parity errors (teletext_parity)
Raises an incident per PID when characters of the text rows have a parity error. The characters are counted, not the rows: by default 20 characters within 60 seconds raise it. The incident gives the number of characters and the page, for example Parity 3 on page 888. Minor by default.
31.4.4.47 Teletext PES layout (teletext_pes_layout)
Raises an incident per PID when the PES or the data units of the teletext PID do not follow EN 300 472: a PES_packet_length that is not a multiple of 184 minus 6, a header data length that is not 0x24, a data_identifier outside 0x10 to 0x1F, a data_unit_id other than 0x02, 0x03 or 0xFF, a data unit length other than 0x2C, a line_offset outside 0 and 7 to 22, or a PES without a PTS. By default 5 times in 60 seconds raise it. Minor by default.
31.4.4.48 Teletext checks and the standards
The ten teletext checks group the single rules of the standards by what an operator does about them. The incident names the rule that was broken and its clause, for example "Late 380 ms on page 888 (EN 300 472, clause 5)", and the teletext viewer and the Subtitles page show the same names. All ten are on by default, run on every teletext source (also while no viewer is open), and take their severity, count, window and per-PID overrides from the Checks page (the Teletext card).
| Check (id) | Default | Rules it judges and their clauses |
|---|---|---|
Teletext lost (teletext_lost) | Major | No teletext data for 5 s while the flow goes on (no clause; a loss alarm). |
Teletext page missing (teletext_page_missing) | Minor | Page not sent: a page the teletext descriptor signals is transmitted (EN 300 468, 6.2.43). |
Teletext subtitle hidden (teletext_subtitle_hidden) | Major | Inhibit Display (C10) set on a subtitle page (EN 300 706, table 2). |
Teletext subtitle late (teletext_subtitle_late) | Minor | Late: the last packet arrives after the PTS (EN 300 472, clause 5); Retention, off by default: data waits more than 40 ms plus one field before the PTS (EN 300 472, clause 5). |
Teletext subtitle flags (teletext_subtitle_flags) | Minor | Subtitle bit (C6) missing and Newsflash bit (C5) set on a subtitle page (EN 300 706, table 2); a subtitle in a data unit for non-subtitle data (EN 300 472, 4.4). |
Teletext text outside the box (teletext_unboxed) | Minor | Text outside Start Box and End Box of a page with C5 or C6 (EN 300 706, 12.2). |
Teletext subtitle erase (teletext_erase) | Minor | Page cleared or rows kept without Erase Page (C4) (EN 300 706, A.2). |
Teletext transmission errors (teletext_transmission) | Minor | Framing code (EN 300 706, 6.2); missing page header (7.1); wrong magazine in serial mode (B.3); serial mode changed, C11 (table 2). |
Teletext parity errors (teletext_parity) | Minor | Odd parity of the character bytes (EN 300 706, 8.1). |
Teletext PES layout (teletext_pes_layout) | Minor | PES layout (EN 300 472, 4.2); data identifier, data unit id, length and line offset (EN 300 472, 4.4); PTS present (EN 300 472, annex A). |
Counted only, never an incident: corrected Hamming errors (EN 300 706, 8.2), a repeat with Erase Page (EN 300 706, A.2), an update that is late within the margin, and the updates the probe dropped because it was busy.
31.4.5 Severity
Every check has a severity, on the same scale on every page (Incidents, the dashboard, the multiviewer and the alarm export):
| Severity | Meaning |
|---|---|
| Critical (red) | The service is off air or unusable. |
| Major (amber) | Viewers see or hear it. |
| Minor (blue) | Degraded, no effect on air yet. |
| Info (gray) | Listed on the Incidents page, but not an alarm: not exported, no mark on a multiviewer tile. |
A check with two limits (packet spacing, RTP jitter, MDI delay factor, MDI media loss, PCR frequency offset and drift) raises Minor above its lower limit and Major above its upper limit; one incident moves between the two. Settings saved by an older version with the severity Warning read as Minor.
TTL change is on for every flow except discovery and local control multicast, whose senders use different TTL values by design: the local control block 224.0.0.0/24 (with mDNS 224.0.0.251 and LLMNR 224.0.0.252), SSDP 239.255.255.250, and ff02::c, ff02::fb and ff02::1:3. This default is built into the probe. An override that switches the check on for a group decides for its flows instead, for example to check SSDP flows after all.
31.4.6 Sending alarms to other systems
Checks decide when an incident is raised; they do not send it anywhere themselves. To send alarms to the Alert Center, syslog, SNMP managers, email, Zabbix, Nagios or webhooks, use Settings > Alarm export (section 31.19), where each target has its own filter.
31.5 Users
The Users page manages user accounts and access control.
| Field | Description |
|---|---|
| Username | The login name. |
| Role | The access level: Admin (full access), NOC (operational access), Management (read-only). |
| Password | Set or change the user's password. |
The password is entered twice, in Password and Confirm password. When the two differ, the confirm field shows "The passwords do not match" and Save stays disabled until they are the same. A new user needs a password: Save stays disabled until one is entered (the probe refuses a new user without one too). When editing an existing user, leave both fields empty to keep the current password; a new password typed there replaces it.
Each change is stored when you confirm it. Save in the dialog sends the change to the probe at once; the dialog shows Saving... and closes when the probe has accepted it, and the table then shows the stored list with a short Saved note. When the probe refuses the change (for example a second Admin on a demo site), the dialog stays open and shows the reason, and nothing in the table changes. Delete asks first ("Delete user ...? This cannot be undone.") and then removes the user at once; the Active box in the table also saves at once. There is no Save button under the table. To change your own password without this page, use Change password in your Profile (3.8), which takes effect at once.
Admin users can create, edit, and delete user accounts. Non-admin users can view the user list but cannot modify it.
Passkeys (Admin). The Passkeys column shows how many passkeys each user has, with a warning symbol when one of them may be copied. All passkeys under the table lists every passkey with its owner, state and last use, and Revoke removes one after a confirmation in the page; sessions signed in with it end. The edit dialog of an existing user lists that user's passkeys with Rename and Revoke. An Admin cannot add a passkey for another user, because a passkey is created on the user's own device (see Profile in 3.8). These actions take effect at once, without Save, and open pages follow the change.
31.6 Signal Generator
The Signal Generator page controls IP-Probe's built-in SMPTE ST 2110 test stream generator. The generator can produce standards-compliant video and audio test streams.
This feature requires the 2110 license option.
31.6.1 Operating Modes
| Mode | Description |
|---|---|
| NIC mode | Transmits real packets on the network via a NIC port with hardware-clock pacing. |
| Demo mode | Generates packets internally and feeds them into IP-Probe's processing pipeline without network hardware. |
| Hybrid mode | Combines real NIC inputs with a virtual demo port for simultaneous live and generated traffic. |
31.6.2 Config Tab
Configure the generator before starting:
- General - Source IP address, source MAC address, port index.
- Video Stream - Enable/disable, format selection (1080i50, 1080p29.97, 1080p50, 720p59.94), test pattern (Color Bars 75%/100%, Flat Field, Luma Ramp, Moving Bars, Frame Counter), destination multicast IP, UDP port, RTP payload type.
- Audio Stream - Enable/disable, sample rate (48/96 kHz), channel count (2/4/8/16), tone pattern (Silence, Sine, Sine Odd Only), frequency, destination address and port.
- Ancillary Stream - Enable/disable, destination address and port.
31.6.3 Stats Tab
While running, displays live transmission statistics:
| Metric | Description |
|---|---|
| TX pps | Current packet transmission rate. |
| Frames Sent | Total video frames transmitted. |
| Drops | Packets dropped due to buffer overflow or TX queue congestion. |
31.6.4 Start / Stop
| Indicator | State | Meaning |
|---|---|---|
| Gray | Stopped | Generator is idle. |
| Orange | Starting/Stopping | Generator is initializing or shutting down. |
| Green | Running | Actively transmitting. |
31.7 Audit Log
The Audit log (Admin only; other users do not see it in the list) shows who changed what on the probe, newest first, kept for the days set under System > History Database (default 365 days).
What is recorded (the Action filter lists these in groups):
- Configuration: every change from the web interface or the REST API (preferences, aliases and video/audio pairs, groups, checks, users, system health and ETR 290 limits, history settings, SRT and RIST monitors), backup downloads, restores and their undo, and configuration versions restored.
- Users and sessions: logins, failed logins, logouts (by the user, or by the "Log out idle pages" limit), users created and deleted, a password set by an Admin, a user's own password change (also a refused one), and passkeys added, renamed, removed or marked suspect. A login with a passkey shows Method: Passkey and the passkey name; a failed one gives the reason in words (for example unknown passkey, wrong address, signature not valid, counter went back, took too long).
- Recordings: recordings created, started, stopped and deleted, and a change of the recording quota.
- System: restart, shutdown, reboot and power off, the signal generator and the test stream started and stopped, port actions of System Health, a changed web address and the certificate actions of Settings > Web server (generated, uploaded, back to the generated one).
- History: history database maintenance (check, Start fresh, Recover, deleted copies) and an acknowledged history database notice.
- Other: license uploads and license requests, and ETR 290 counter resets.
Each line shows the time, the user (with the role and the address of the computer the request came from), the action and how it came in (REST or WebSocket; the probe for actions of the probe itself), the domain (the part of the configuration) and what in it, a summary, the result (OK, or Failed with the reason in words, on a red line) and the configuration revision. Passwords, PINs, keys and passphrases never appear: where they changed, the entry shows ***.
Details opens more of an entry:
- For a configuration change, a table of the differences: the Setting, the Change (added, removed or changed), and the value Before and After. A list entry is named by its name or id, for example
groups[name="Studio A"].priority. At most 50 differences are kept per change; a note says when there were more, or when only the names of the changed settings were kept because the values were too long. - For a failed login, the reason is in the result column (wrong password, unknown or inactive user, too many attempts) and the method (password) in the details.
- For a logout by the idle limit, how long the page was without activity and the limit.
Failed logins. So that someone trying many passwords cannot flood the log, the first 3 failed logins a minute from one address are recorded one by one (at most 20 a minute in total). The rest of that minute become one summary line when the minute ends, with the result Failed: repeated: its Details show how many attempts there were and between which times, the address, the user names tried (at most 5, then "and N more") and how many attempts failed for each reason. Failures from more than 32 addresses at once are summed into one line for "other addresses" (with up to 5 of them). A minute of attempts therefore gives at most about 50 lines.
Finding entries.
- Time range: last 24 h, 7 days, 30 days, All (the default: everything kept) or a custom from / to.
- Filters: user name, domain, action, and a text search in the summary and details. Clear filters resets them.
- Load more reads the next 100 entries.
- Export saves the selection as CSV or JSON (at most 100000 rows; the message says when the file was cut).
- When someone saves a change while the page is open, a button "N new entries, show" appears; click it to read the list again.
31.8 System Health
Settings > System health (under Monitoring) sets the limits for the hardware health shown on the Inputs and About pages and in the health widgets. An Admin can change them; other users see the values that apply now.
The Thresholds table has one row per check, with a switch to turn the check on or off, a Warning limit and a Critical limit. Leave a limit empty to use the default.
| Check | Limits |
|---|---|
| NIC chip temperature | Warning in degrees C (default 95). Critical: Device limit (the limit the chip itself reports) or a number. |
| Module temperature | The module's own warning and alarm thresholds, or numbers. |
| Module optical and electrical | The module's warning and alarm flags (Rx and Tx power, Tx bias, supply). Never stops capture. Not judged while the port has no link. |
| PCIe correctable errors | Warning and critical counts within a window in minutes (default 10 and 100 in 10 minutes). Uncorrectable errors are always critical. |
| PCIe link below capability | Warns when the link runs slower or narrower than the card supports, for example x4 in an x8 slot. |
| Link lost | Warning when a link that was up goes down. A port that never had link is gray "No link" with no incident. Closes when the link is up again or with Reset on the Inputs page. |
| CPU, NVMe, GPU, memory module temperature | Warning and critical in degrees C; Device limit uses the critical limit of the drive or module. |
| Memory used | Warning and critical in percent (default 90 and 97). Hugepages count as used. |
| Disk space used | Warning and critical in percent of the disk that holds the probe data, and of the recordings disk when that is another disk (default 85 and 95). |
| Memory errors (ECC) | Warning when the number of corrected memory errors within the window in minutes reaches the limit (default 1 in 60 minutes). Any uncorrected error is always critical. No check applies to memory without ECC. |
| ipprobe memory (RSS) | Growth in GiB per hour, off by default; for finding memory leaks. |
Stop port on critical. The NIC chip and the module temperature rows have a switch Stop port on critical. When it is on, capture on that port stops at the critical limit and restarts by itself when the sensor has cooled down (see chapter 7.4). The host checks never stop capture.
The Behavior section holds the timing:
| Field | Meaning |
|---|---|
| Confirm | How many samples (one every 5 seconds) a value must stay past a limit before the state changes. |
| Clear margin | How far below a limit the value must fall before the state goes back to OK (degrees, and percent for memory and disk space). |
| Restart hold | How long a stopped port must stay below the warning limit, minus the margin, before it restarts. |
| Repeat | While a warning stays, one reminder line in the log per interval, and no new incident. |
| Incidents | Turns the incidents for health warnings on or off. |
Preview on this probe lists the sensors of this probe with their current value and state. Click Save to apply the limits (they take effect at the next sample, no restart); a value that cannot be saved is marked, and Reset all to defaults returns every limit to its default after a confirmation. Every change is written to the audit log with the old and the new value.
31.9 Backup / Restore
Settings > Backup / restore (Admin only) saves the whole probe configuration in one file and puts it back on the same probe or on another one.
Backup. Download backup saves a zip file with the preferences, aliases, audio channels, groups, checks, users, system health and ETR 290 limits, the signal generator, the SRT and RIST monitors and the recording sessions. Logs and capture files are not included. The file contains the users' password hashes: store it like a password list. Include the license adds the license; a license only works on the machine it was made for, so a restore on another probe skips it.
Include the history database adds the history of this probe: incidents, audit log, ETR 290 history and configuration versions. The page shows its size first. The probe makes a copy of the database while it keeps recording, so it needs as much free disk space; a large database makes a large file and takes a while. While the probe builds the file the page says so; then it shows how much has been downloaded.
Restore. Drop a backup file on the page or click Choose file. The probe checks every file first and changes nothing yet. The preview shows:
- where the backup comes from (probe name, time, software version, Unique System ID) and whether it was made on this machine or another machine;
- warnings, for example that the users and their passwords are replaced;
- whether the backup holds the history database;
- one line per file with its status: Changed (what differs: entries added, removed or changed, with names; settings that differ; a file converted from an older format; a file the probe does not have yet), Unchanged, Invalid (with the problems found; it cannot be applied), Skipped (for example the license of another machine) or Not in backup (the probe keeps its own file). Needs a restart marks a file that only takes effect after a restart (generator, recording sessions).
When the backup holds a history database, a History database box shows whether it is Valid (with the number and the time range of the incidents, audit entries and ETR 290 episodes, and the number of configuration versions), Invalid (with the problems; it cannot be restored) or Skipped. The history is restored only when Restore the history database (replaces the history of this probe) is ticked; it is off by default. It can be restored on its own, with no file chosen. The history of this probe is kept as a replaced copy (see 31.1.7); an undo of the restore does not bring it back.
Every changed file is chosen; clear the ones you want to keep. Apply asks for a confirmation in the page before anything is written. The current files are first saved as a pre-restore copy, then the chosen files are written together: if one fails, all files get their old content back. The result lists what was restored. When a restored file needs a restart, Restart now restarts the probe (after a confirmation). If the restore changed your own password, you are logged out and the login page says why.
A preview stays valid for 15 minutes. When a file changed on the probe since the preview, or the preview expired, the probe refuses the apply and nothing is written: upload the backup again. Other refusals have their own message, for example a file that is not an IPProbe backup, a backup of newer software, a file larger than the probe accepts (32 MB, plus the size limit of the history database when the probe keeps one), or another backup, restore or history database action that is still running. The page shows how much of a large file has been uploaded.
Undo a restore. The probe keeps the newest five pre-restore copies. Undo last restore (or Put back for an older one) restores that copy after a confirmation; the configuration before the undo is saved as a new copy, so an undo can be undone too. To put back one settings file, use the configuration versions (31.10).
31.10 Configuration Versions
Every saved settings file is kept as a version in the history database: each save from a settings page, each restore, the restore of a version, and every file as it is at each start. A version that equals the newest one of its file is not stored again. How many are kept is set in 31.1.7. Only an Admin sees and restores versions. User accounts are not kept as versions, because an old version would bring an old password back; they are in the backup file.
On a settings page. For an Admin, a line above System, Inputs, Aliases, Groups, Checks, ETR 290, System health and the Signal generator says how many versions of that file are kept and when the newest was saved:
- Undo last change puts back the newest version that differs from the current file, after a confirmation in the line. The current file is kept as a version too, so a second undo goes back again.
- Versions... opens the list of versions of that file in a dialog.
Settings > Versions shows the same list for every file (choose it under Settings file), also for the files without a settings page (audio channels, generator, SRT and RIST monitors, recording sessions). Per version: the time saved, who saved it, where it came from (Saved, Backup restore, Undo of a restore, Version restored, At start) and its size; current marks the version that equals the file on the probe now. Older versions reads the next part of the list.
- The eye button shows a version: whether it differs from the current file and what a restore would change (entries added, removed or changed, settings that differ, a format conversion), and its content. A version that no longer passes the checks of this software is marked Invalid with the reasons and cannot be restored.
- Compare with the current file (or tick two versions and click Compare) lists every difference: where (the setting, or the list entry by its name, address or number), whether it was added, removed or changed, and the values before and after.
- Restore this version makes it the current file, after a confirmation in the page. It goes through the same checks as a save and takes effect at once; the file it replaces stays in the list of versions, so the restore can be undone. A version restore makes no pre-restore copy, so the copies kept for Undo a restore (31.9) are not used up. A restore of the users can log you out when your own password changes.
Password hashes, passphrases and keys are always shown as ***; a restore puts back the real values. When a backup, a restore or a history database action is running, a restore or undo is refused, not queued: the page says so, and you try again when the other action has finished. Without a history database no versions are kept, and the page says so.
Open settings pages follow. After an undo, a version or backup restore, or a save by another user, an open settings page of that file reads it again. When the page has unsaved changes it asks first: Reload takes the file of the probe, Keep editing keeps your changes (a later save then writes them over the other change).
31.11 Flow Monitoring
Settings > Flow monitoring (Admin only) switches the monitoring of a UDP flow off and on: MPTS, ST 2110, ST 2022-6, SRT, RIST, PTP over UDP and other UDP or RTP flows. Use it for a spare feed, a test source or a flow that is known to be broken, so that it no longer fills the probe with incidents. Every flow is monitored by default. Flows that are not UDP (ARP, IGMP and similar) are always monitored.
Not Monitored. A flow that is switched off is marked Not Monitored everywhere it appears: the Flows page, the MPTS overview, the flow detail page and the status breakdown on the Summary (4.2), where it is a group of its own and is never counted as OK or as an error.
| Keeps running | Stops |
|---|---|
| The flow stays in every list. | The analysis: MPTS tables, services and ETR 290, thumbnails and decoding, ST 2110 analysis, media clock, audio meters and ancillary data. |
| Packet and bitrate counters, last seen time. | Incidents: no TTL, DSCP, RTP sequence, payload type or SSRC incidents and no "no data" check. |
| The type of the flow (for example ST 2110-20) stays current. | History rows, and recordings that are started by an incident. |
| Recording sessions you started yourself keep capturing. | Live video and audio of the flow: running viewers end with "Not Monitored" and new ones are refused. |
The probe also uses less CPU for a flow that is not monitored, because its packets are counted and then dropped before the analysis.
Switch a flow. Open Settings > Flow monitoring. The table lists every known flow, including flows that are not live now (a switched-off flow stays in the list; an alias with a destination address is listed too). Search for a name or an address, or filter with All, Not Monitored, Live and Not live. Click the switch in the first column. Switching off asks for a confirmation, with an optional reason (it is shown in the table and in the audit log); switching on happens at once. To change several flows, tick them and use Monitor or Stop monitoring above the table. Add flow... prepares a flow that is not live yet, for example a spare feed that will start later. Changes apply within a second, without a restart.
The State from column says why a flow is off: Flow (its own setting) or Group (see below). For a flow that is off because of a group, Monitor this flow anyway adds a setting of its own that overrides the group.
Switch a group. In Settings > Groups (31.3), the dialog of a group has Do not monitor the flows of this group. All flows in the group, now and in the future, are Not Monitored, for example a multicast range of spare feeds. A setting on one flow always wins over the group. A group with the rule is marked in the Groups list and listed at the bottom of the Flow monitoring page.
Incidents. When a flow is switched off, its open incident episodes end at once. The Incident History (27.1) shows them as Ended: monitoring switched off. Switching the flow on again starts fresh: new analysis, new ETR 290 state and no false gap from the time it was off.
Who can switch. Only an Admin. Other users see the Not Monitored tag and its reason, but no switch. On the flow detail page of a flow that is off, an Admin gets a link to the Flow monitoring page. Every change is written to the audit log (31.7) with the flow, the reason and the user, is part of a backup (31.9) and is kept in the configuration versions (31.10).
31.12 Web Server
Settings > Web server (Admin only, under General) sets the web address of the probe and manages the HTTPS certificate that browsers check when they open it. Passkeys depend on both: a browser only allows a passkey on a host name (not on an IP address) with a certificate it trusts, and a passkey is tied to the host name it was made for.
Web address.
| Field | Description |
|---|---|
| Host name | The DNS name users open, for example probe1.example.com, without https:// or a port. An IP address and localhost are refused. Empty turns passkeys off. |
| HTTPS port | The port of the web interface, 1 to 65535 (default 2300). |
| Extra addresses | Up to 8 more addresses users may open the probe by, for example behind a reverse proxy. Each one must be https:// and be the host name or a name below it (https://ipprobe.probe1.example.com). |
| Redirect http to https | A visit to the plain http address is sent to the https address. On by default. |
Errors are shown below the field they belong to, both while you type and when the probe refuses a value. Save applies the address at once. When the host name changes, the page warns first, for example "3 passkeys stop working on the new address. Users can sign in with their password and add new ones.", and asks Save anyway? in the page. The passkeys are kept and marked For another address; changing the host name back makes them work again. The versions line above the page keeps the earlier addresses (31.10).
The Passkeys line at the top says where things stand: ready at the address, set a host name, the certificate does not cover this host name, or open this page at the address to check (when you opened the page in another way, passkeys can only be tried at the web address).
HTTPS certificate. The card shows where the certificate comes from (generated by this probe, or uploaded), its subject, the host names it covers, whether it covers the host name set above, the issuer, the dates it is valid with the days left, and the SHA-256 fingerprint (the copy button copies it). A warning appears when 30 days or fewer are left, an error when it has expired, a warning when the certificate does not cover the host name, and a warning with the addresses when the probe has an address the generated certificate does not cover (the probe CA only vouches for the addresses the probe had when the CA was made). Then open the probe by its host name, or regenerate with a new CA; every computer then installs the new CA again.
- Download probe CA saves
ipprobe-ca.crt, the certificate of the authority that signed the generated server certificate. It is public and needs no login, so the link is also on the About page for every user. Install it once on each computer (see below) and the browser trusts the probe. - Regenerate server certificate makes a new certificate for the current host name with the same probe CA, so computers that trust the CA keep working.
- Make a new probe CA replaces the authority as well. Every computer and browser must install the new CA again, which the page says after the confirmation.
- Upload certificate... installs your own certificate: the PEM chain (server certificate first, then the intermediates) and its private key, chosen as files or pasted as text, and the passphrase for an encrypted key. The files may be 64 KiB together. The probe checks them and shows what is wrong at the field. The private key stays on the probe; it is never shown again and never part of a configuration backup. A certificate from your own authority needs no CA installation on computers that already trust that authority.
- Use the generated certificate again (shown while an uploaded certificate is in use) goes back to the certificate made by the probe.
How to trust the probe CA. The help under the card repeats these steps.
- Windows: open
ipprobe-ca.crtand choose Install Certificate... (or runcertmgr.msc); place it in Trusted Root Certification Authorities; restart the browser. - macOS: open the file to add it to Keychain Access, open the certificate, expand Trust and set When using this certificate to Always Trust.
- Linux, Chrome and Chromium: Settings > Privacy and security > Security > Manage certificates > Authorities > Import, and tick Trust this certificate for identifying websites. Firefox: Settings > Privacy and Security > View Certificates > Authorities > Import, and tick Trust this CA to identify websites.
Why passkeys need a host name. A passkey is made for one web address and works only there. A browser does not make passkeys for an IP address, and a passkey made for probe1.example.com cannot sign in on another name. That is why the probe needs a host name and a certificate the browser trusts before passkeys work.
Account recovery. The password always works: passkeys are an extra way to sign in and never replace or switch off the password.
- Lost or broken passkey (a lost phone or security key, a new computer): sign in with your password, open your Profile, remove the passkey and add a new one (3.8). If the passkey shows This passkey is blocked, do the same: remove it and add it again.
- Lost password, passkey still works: sign in with the passkey, then change the password in the Profile (3.8).
- Lost password and no working passkey: ask an Admin to set a new password for you in Settings > Users (31.5). The Admin can also remove your old passkeys there. Add a new passkey after you have signed in.
- Lost Admin account (no Admin can sign in): contact support. Do not change the files on the probe yourself: a wrong step can remove all users and their passkeys.
Passkeys added, renamed or removed, a changed web address and every certificate action are written to the audit log (31.7).
31.13 Power
Settings > System > Power (Admin only) restarts IPProbe, reboots the probe or switches it off. Other users do not see the page in the list; opening it shows Admin access required.
| Action | What happens | Confirmation |
|---|---|---|
| Restart IPProbe | IPProbe stops and starts again. Monitoring stops for about 30 s. | Confirm in a dialog. |
| Reboot | The machine restarts. Monitoring stops for about 3 minutes. | Confirm with your password. |
| Power off | The probe switches off. It must be switched on at the machine. | Confirm with your password. |
Each button opens a dialog that says what will happen. Running recordings are listed there, because they stop too. Reboot and power off ask for your current password: a wrong password shows The password is wrong. in the dialog, and you stay logged in and can try again. After the confirmation every logged-in user sees what is happening (see While the probe restarts in 3.8) and logs in again when the probe is back. Sessions are kept only while IPProbe runs, so every user must log in again after any of these actions. Each action is written to the audit log (31.7) with the user.
Which actions are available depends on the machine. Reboot and power off need the system service of IPProbe; when it does not answer, or on a Windows computer, the button is grayed out and the panel says why. Restart IPProbe always works. On a probe there is no button to stop IPProbe without starting it again, because the web interface would be gone with no way to start it. Only IPProbe on a Windows computer (a development PC) shows a Stop IPProbe panel. When the page cannot read which actions the probe offers, it says so and the buttons stay disabled.
31.14 Date and Time
Settings > System > Date and time (Admin only) shows the two clocks of the probe apart, and sets the NTP servers, the time zone and, when needed, the clock by hand. Other users do not see the page in the list; opening it shows Admin access required. The About page shows every user one line System clock, for example "Synchronized (NTP), time zone Europe/Amsterdam". A new probe has no NTP server: commissioning turns off the Ubuntu internet time servers, as a probe is normally offline, so the line says "Not synchronized: no NTP server is set" until an Admin enters the time server of your network here.
The probe has two different clocks:
| Clock | What it is for | Set by |
|---|---|---|
| System clock (NTP) | The time of the probe: log files, incidents, license dates. | NTP servers on the management network, or by hand. |
| Media clock (PTP) | The time of the media streams, measured on the media network. | Nothing on the probe: IPProbe only measures PTP. |
IPProbe does not set its clock from PTP. The arrival times of packets come from the network cards.
System clock (NTP). The panel shows a status tag: Synchronized, Not synchronized (with the reason, for example "No NTP server answers.") or No time service. Below it: the probe time in UTC and in the probe's time zone, the time of this computer, the difference between the two, the selected server, its offset and stratum, and when the clock was last corrected. The server table lists each NTP server with whether it answers, its stratum and its offset.
Media clock (PTP). For each input the PTP grandmaster, its domain and its state, from the same data as the PTP page, or "No PTP seen on the inputs." Open PTP opens the PTP page (13).
NTP servers and time zone.
- Use NTP switches the time service on or off.
- NTP servers: up to 4, each a host name or an IP address. The page marks an entry that is not valid or listed twice before saving.
- Time zone: type part of the name ("new york", "amsterdam") and pick the zone from the list. Log files and the system use this zone. IPProbe reads it when it starts, so after a change the page asks Restart IPProbe now to use the new time zone? Choose Restart (monitoring stops for about 30 s and everyone logs in again) or Later (the new zone is used after the next restart, see 31.13). The web pages keep showing times in the time zone of your computer.
- Save stores the settings. Each change is written to the audit log (31.7).
Set the clock by hand. Only shown while NTP is not synchronized, for a site without an NTP server. It sets the probe's clock to the clock of this computer and shows the difference first. Check that this computer shows the right time. The dialog asks for your current password; a wrong password shows The password is wrong., and you stay logged in. When the new time is more than a day away from the probe's clock, the dialog says how far the clock moves (for example "The clock moves 3 days forward.") and asks Set it anyway? Select Set it anyway to set it, or Cancel. Before you confirm, keep in mind:
- The history is kept. After a forward jump the probe waits one hour before it removes old data by its age, so a wrong setting can be undone in that time.
- The license is checked against the clock. A time after the license end date switches IPProbe to restricted mode (32.1) until the clock is right again.
The probe refuses:
- a time before the license was made or before this version of IPProbe was built ("This time is before the license was made; IPProbe would run without its options"): check the date of this computer;
- any time while NTP is synchronized ("The clock is synchronized by NTP and cannot be set by hand").
Setting the clock is written to the audit log with the time before and after.
When the page cannot change the clock. A probe that uses systemd-timesyncd instead of chrony, or has no time service, shows its settings read only. Without the system service of IPProbe the clock can be read but not changed, and the page says why. On a Windows computer IPProbe uses the clock of Windows; set the date, time and time zone in Windows.
31.15 Support
Settings > System > Support (Admin only) makes the support bundle: one zip file with what support needs to look at the probe. Other users do not see the page in the list; opening it shows Admin access required.
Contains: logs, settings without passwords, system and hardware facts (network, disks, kernel messages), health, the license state (not the license file) and the incidents of the period.
Never contains: passwords, keys (including the private keys of certificates), the license file, session tokens, backups and the history database. In log files, IPProbe replaces secret values with *** before they go into the bundle: values written as name=value (for example secret=, passphrase= or token= in a web address or a command), as "password": value, as an Authorization or Proxy-Authorization header (everything after the colon; the word Bearer, Basic or Digest stays readable), and as password: value or passphrase: value (also with a prefix such as db_password:); the file list in the zip says how many. A log file that holds a private key, a password hash or part of the web server's key is left out of the bundle completely. Every other text file is checked for secrets before it is added to the zip.
Make a bundle.
- Choose the Period: Last day, Last 7 days (default) or Last 30 days. It limits the logs and incidents.
- Decide on Include the audit log (user names and client addresses). The audit log holds personal data: the names of users and the addresses of the computers they used. Switch it off if you do not want to send these.
- The page shows the estimated size ("About 38 MB"); it changes with the choices (the probe is asked a moment after the last change, and Create and download waits for the size). A bundle is never larger than 200 MB: when the logs are too large, the oldest are left out and the manifest in the zip says so.
- Select Create and download. The probe collects the files, which can take a minute; the page shows the progress, and the browser then saves the file as
ipprobe-support-<probe name>-<date and time>.zip. Cancel stops waiting, but the probe may need up to a minute to finish the bundle it started; Create and download stays off until then. The browser keeps the whole file in memory while it downloads, so a large bundle (up to 200 MB) needs a computer with enough free memory; choose Last day or Last 7 days if the download fails.
Only one bundle is made at a time. If another Admin is making one, the page says The probe is busy with another job (a support bundle, a backup or a restore); try again in a minute. Each bundle is written to the audit log (31.7) with the options and the size.
Without the system service of IPProbe (a Windows computer, or a probe where the service is not installed) the page says that the system facts are left out; the other files are still included.
The panel This probe shows the version and the Unique System ID, which support often asks for. Copy copies the ID.
Diagnostics at the top of the page opens the Diagnostics page (chapter 30), with the internal processing views that support may ask about.
31.16 Network
Settings > System > Network (Admin only) sets the management network of the probe: the network cards you use to reach the web interface, not the media ports that capture the streams. Other users do not see the page in the list; opening it shows Admin access required.
Interfaces. Each network card has a panel with its name, its MAC address, a link tag (for example Link 2.5 Gb/s or No link) and what it has now: its addresses, gateway, DNS servers and, with DHCP, the lease. The tag This browser marks the card this page is open over: a change there can cut your browser off. A card IPProbe may not change is shown read only, with the reason, for example "Not found on this machine." or "Not set up for IPProbe at commissioning. Contact service to manage it here."
For each card you can set:
| Setting | What it does |
|---|---|
| IPv4: DHCP / Static | DHCP takes the address from the DHCP server; Static uses the addresses you enter. |
| Address and Prefix | Up to 4 IPv4 addresses, each with its prefix length (8 to 30), for example 192.168.10.50 / 24. |
| Gateway | The default gateway, in the subnet of one of the addresses. Empty for none. Only one card can have a gateway. |
| DNS servers | Up to 3, IPv4 or IPv6 addresses. |
| Search domains | Up to 6 domain names, separated by commas. |
| IPv6: Off / Automatic | Automatic takes IPv6 addresses from the network (router advertisements). |
The page marks a value that the probe would refuse before you save, for example "This is the broadcast address of the subnet". Undo changes puts back the current settings.
Two cards cannot share a subnet: a change that puts a card into a subnet another card already uses is refused on its address field ("... two interfaces cannot share a subnet."). Two cards that were already in the same subnet before IPProbe managed them are kept as they are: the page shows a note, and every change shows the warning Shared subnet with the two cards and the subnet, which you accept with I understand the warnings.
Host name. The panel shows the full name and the web address of the probe and lets you change the Host name (one name, such as probe1: letters, digits and hyphens, at most 63 characters) and the Domain (optional, for example studio.example.com). The page marks a value the probe would refuse. A host name change needs no test period: it applies at once and does not cut the connection.
The switch Use probe1.studio.example.com as the web address is off by default and needs a domain. Without it only the host name of the probe changes and the web address stays as it is (it can also be set under Settings > Web server, 31.12). With it the web address changes to the full name, and with it:
- Passkeys belong to the web address: the passkeys made for the old address stop working and must be made again at the new one. The confirmation says how many.
- Certificate: a generated certificate is made again for the new name, with a new probe CA that every computer installs once (the notice offers Download the probe CA). An uploaded certificate is never replaced: browsers warn until you upload one for the new name. If making the certificate fails, the old one stays in use and the notice points to Regenerate under Settings > Web server.
- Links and bookmarks to the old address stop working. Open the new address if this page stops answering.
Save... shows the changes and the warnings (for example Check your DNS: make sure the new name points to the probe) in a dialog. With warnings, tick I understand the warnings, enter your password and select Save. A wrong password shows The password is wrong.; you stay logged in and can try again. After five wrong passwords in a minute the probe asks you to wait.
Media ports (used by IPProbe capture). The capture ports are listed read only with their PCI address and card. They are not set here.
Apply a change safely. A wrong network setting can make the probe unreachable. So a change is first tested, and the probe goes back to the previous settings by itself unless the change is confirmed from the new address.
- Select Save.... The probe checks the change and shows it in a dialog: each setting before and after, and the warnings, for example "This browser loses the page", "Address not in the certificate" or "Check your DNS". With warnings, tick I understand the warnings.
- Choose the Confirm window: 60 to 600 seconds, 180 by default.
- Keep the old address during the test (on by default) leaves the old address on the card until the change is kept or restored, so this page stays reachable and can show the new address, also one that a DHCP server hands out.
- Enter your password and select Apply. A wrong password shows The password is wrong.; you stay logged in and can try again.
- The banner Testing the new network settings counts down and lists the confirm links, one for each new address and one for the web address. With DHCP it says "Waiting for an address from the DHCP server..." until the address is known. Open a link (or Copy link and open it on another computer that reaches the new address).
- The confirm page asks Keep the new network settings? with the same countdown. Select Keep. It says "Settings kept. Log in at this address." At a new IP address the browser may first warn about the certificate: the probe's certificate does not name that address yet. Continue to the page.
When no address changes (only DNS, search domains, gateway or IPv6), the links point to the addresses the probe keeps, and the change is confirmed the same way.
The confirm must reach the probe over a new address. A confirm over the old address (for example through the web address while your DNS still points to the old address) is refused with "Open the link with a new address to keep the settings", and the countdown goes on. Each link works once; a used link shows Nothing to confirm. The code in the link is removed from the address bar when the page opens.
Do not keep on the confirm page sends nothing: the previous settings come back when the time runs out. Restore the old settings now in the banner ends the test at once. When nobody confirms in time, the probe restores the previous settings by itself, also after a restart during the test, and the page says "The previous settings were restored at 10:42:13 because the change was not confirmed." until you select Dismiss.
If you logged in at the new address while the test runs (for example because the link got lost), the banner offers Keep these settings.
If the change takes this browser's address away (the old address is not kept), the page shows Waiting for the new address with the links to the new address. It opens the new address by itself when the probe answers there, and reloads when the previous settings come back.
Certificate after an address change. When the kept addresses are not in the probe's certificate, the probe makes a new certificate with a new probe CA by itself. The confirm page says "A new certificate is being made for the new address. Browsers and other clients must install the new probe CA once." with Download the probe CA. On the Network page a notice follows: the certificate was made (with the CA download), or it could not be made. Then the network change stays, the old certificate stays in use, and you use Regenerate under Settings > Web server (31.12).
Install the new probe CA. After a confirmed address change every computer and browser that opens the probe must install the new probe CA once. The old CA does not vouch for the new certificate, and a second CA from an earlier change may stay installed or be removed.
- Download the CA from the link on the confirm page or on the Network page, from Settings > Web server (Download probe CA, 31.12) or from the About page. You do not need to log in for the download. The file is
ipprobe-ca.crt. - Install it on each computer:
- Windows: double-click
ipprobe-ca.crtand choose Install Certificate.... Choose Local Machine (needs administrator rights) or Current User, then Place all certificates in the following store, select Trusted Root Certification Authorities, finish, and restart the browser. Edge and Chrome use this store. - macOS: double-click the file to add it to Keychain Access, open the certificate, expand Trust and set When using this certificate to Always Trust. Safari and Chrome use the keychain.
- Linux: Chrome and Chromium have their own list: Settings > Privacy and security > Security > Manage certificates > Authorities > Import, and tick Trust this certificate for identifying websites. For command-line tools on Ubuntu or Debian, copy the file to
/usr/local/share/ca-certificates/and runsudo update-ca-certificates. - Firefox (any system): Settings > Privacy and Security > View Certificates > Authorities > Import, and tick Trust this CA to identify websites.
- Windows: double-click
- Close the browser completely and open the probe again. The padlock shows no warning.
Until the CA is installed, a browser shows a certificate warning when it opens the probe (for example "Your connection is not private"), passkeys cannot be used, and programs that call the REST interface refuse the connection. Continuing past the warning works for the web interface, but then the connection is not checked. A probe whose certificate you uploaded from your own authority needs none of this on computers that already trust that authority.
Every apply, confirm and restore is written to the audit log (31.7). Without the system service of IPProbe, or on a Windows computer, the page says Not available on this machine and shows only the host name; nothing can be changed. A probe that is not commissioned yet shows its network cards read only with a note to contact service. When the list of capture ports cannot be read, the page says so and blocks every change, so a capture port is never changed by mistake.
31.17 The Probe's Screen
A screen connected to the probe (HDMI) shows its address. This is how you find the probe when you do not know its address, for example after a change of network or a DHCP address you were never told. The screen has no menu and no customer login: the probe is managed in a web browser. The login prompt below the text is for service staff.
The text looks like this:
IPProbe 2026.02 - probe1.studio.example.com
Management:
enp4s0 static 192.168.10.50
enp5s0 DHCP 10.0.0.23
Web: https://probe1.studio.example.com:2300 (or https://192.168.10.50:2300)
Manage this probe in a web browser. There is no customer login on this screen.
- The first line has the version and the full name of the probe (the host name, and the domain when one is set).
- Under Management each network card used for the web interface has one line: its name, static or DHCP, and its addresses. A static address is the one you entered. A DHCP address is the one the DHCP server gave and follows a new lease within a few seconds. no address yet means the DHCP server has not answered (check the cable and the DHCP server). While a network change is being tested, the line also shows the old address "until the change is confirmed".
- The Web line is the address to type in the browser. With a host name it shows the name and, in brackets, the first address. The port is left out when it is 443.
- not found on this probe (contact service) means the card is missing or was replaced.
- A probe that is not commissioned yet shows "Network (this probe is not commissioned yet):" and "No management interface was found. Contact service." when it has no card.
The screen follows every change made on the Network page (and its automatic restore) within a few seconds.
If you lost the address.
- Read the Web line on the screen and open it in a browser on a computer in the same network. A certificate warning is normal until the probe CA is installed (31.16).
- Log in as an Admin. The Network page (31.16) shows the settings and lets you change them.
- If the computer cannot reach the address (a different subnet, a wrong gateway), put a computer in the subnet shown on the screen, or ask your network administrator for a DHCP reservation.
- If no address works and no Admin can log in, contact support. Do not change the network files on the probe yourself.
A change on the Network page that cuts you off is undone by the probe itself when it is not confirmed in time (31.16), so the address on the screen after that time is the previous one.
31.18 Update
Settings > Update (Admin only) installs a new IPProbe version from a package file, and goes back to the previous version when you ask for it. Other users do not see the page in the list; opening it shows Admin access required. The probe needs no internet connection: you get the package from your supplier and upload it on this page.
An update works in this order: the probe checks the package and shows what will change, you confirm with your password, the probe saves your settings, stops, installs the new version, starts it and checks that it runs. If the new version does not start correctly, the probe goes back to the previous version by itself. Your settings, users, passkeys, certificates, license, history and recordings stay on the probe. Monitoring stops while the update runs, usually for about 2 minutes; the page tells you the time before you confirm.
What you need. The package file for your probe, for example ipprobe_2026.11.0_ubuntu26.04_amd64.ipu, and, when your supplier gives you one, the checksum file with the same name and the extension .sha256. The .ipu file is signed by the maker of IPProbe: the probe installs only a package whose signature it can check, so a changed or damaged file is refused. Download both from the place your supplier names. The package must be for the Ubuntu release of your probe (the number in the file name, 26.04 in the example), and your license must cover the version (see Updates until below).
31.18.1 The page
| Panel | What it shows |
|---|---|
| Installed | Version and Release date of the running IPProbe. Updates until: the last release date your license covers, with the days left, No end date, or No valid license. Last update: the result of the last attempt, its versions and the time. |
| Update to ... | Only while an update or a Go back runs, and after it ended: the steps, and the result (31.18.5). |
| Install a new version | Upload, checks, release notes, the warning and the Install button (31.18.2). |
| Go back | Returns to the previous version (31.18.6). |
| Previous updates | The last 20 attempts: time, from, to, result and who started it. |
A note at the top of the page tells about history that a Go back kept aside (31.18.7).
31.18.2 Install a new version
- Drop the package file on Install a new version, or click Choose file. A file larger than 512 MB is refused: it is not an IPProbe package. The page shows the progress; Cancel upload stops it. Only one package is on the probe at a time: a new upload replaces the earlier one. If you reload the page, the package is still there until you click Remove package or install it, unless the probe refused it: a package that did not pass its checks is removed when the page opens (an Admin's page does this; the page says so) and the drop zone is shown again. To use another package while one is loaded, click Choose another package above its version line: the loaded package is removed and the file dialog opens. You can also drop another
.ipufile anywhere on Install a new version; it replaces the loaded one. The checksum line takes only the.sha256file: an.ipufile dropped there is refused with the hint to use Choose another package. - The probe checks the package ("Checking the package..."). Nothing changes on the probe until you install. The page shows the version, the release date, the size and the SHA-256 checksum of the file, with two tags: the signature (Signature valid; Not signed or Signature does not match for a file that is refused) and No checksum file given. The probe checks the signature first: a file that is not signed by the maker of IPProbe is refused before anything else is looked at.
- Compare the checksum. Drop the
.sha256file on the line "Compare with the checksum file of the release", click Choose file, or paste the checksum into or paste the checksum and click Compare. The tag turns into Matches the checksum file or Does not match the checksum file. A package that does not match is refused. Without a checksum file the page only shows the checksum; compare it with the one your supplier published when you can. - Read the Checks (31.18.3), the Release notes (what is new since your version, in the order Implemented, Fixed, Changed, Updated) and What changes: settings that are converted to the new version, a conversion of the history database (a copy is kept for Go back), changes to the probe's system settings, and The probe restarts during this update. when a restart is needed.
- Read the warning box. It says how long monitoring stops (Monitoring stops for up to 2 minutes.), or, when the probe restarts, The probe restarts during this update. Monitoring stops for up to N minutes. It lists the Recordings in progress and the Open live viewers. The update may go ahead while they run; they stop with it.
- Click Install. The button is available only when every check passed and nothing else is running. The dialog repeats the warning and asks for Your password, the password of the Admin who is logged in; a wrong password shows The password is wrong. and nothing starts. Click Install in the dialog.
Before it stops anything, the probe saves a copy of your settings (without the history). It is listed with the saved copies under Settings > Backup / restore (31.9); the newest three are kept. The update itself keeps the previous version and an exact copy of the settings for Go back.
31.18.3 The checks
Every line is green (passed), amber (passed with a warning to read) or red (refused). If one line is red, This package cannot be installed. The red lines say why. and Install stays off. Fix the cause (usually: download the right package again) and upload it again.
| Check | Refused when | What to do |
|---|---|---|
| Signature | "The package signature is missing or does not match." The file is not a signed IPProbe package, or it was changed after it was signed. Then this is the only check shown. | Download the .ipu file of the release again from your supplier and upload it. A .deb file cannot be installed on this page. |
| Update package | "The file is not an IPProbe update package. Upload the .ipu file of the release." | Upload the .ipu file of the release, not the .sha256 file or an archive. |
| Package complete | "The package is damaged or incomplete. Download it again." | Download it again; check the size. |
| IPProbe package | "This package is not IPProbe." | Use the package of IPProbe. |
| Architecture | "This package is for the X architecture. This probe needs a package for amd64." | Download the amd64 package. |
| Ubuntu release | "This package is for Ubuntu X. This probe runs Ubuntu Y. Download the package for Ubuntu Y." | Download the package for the Ubuntu release of this probe. |
| Newer version | "This package is version X. IPProbe Y is installed. Older or equal versions cannot be installed; use Go back to return to the previous version." Also a test build, a development build, or a package with inconsistent version numbers ("Download it again."). | Upload a newer released version. To return to the previous version use Go back. |
| Checksum | "The package does not match its checksum file. Download both files again." Also "The package changed after it was checked. Upload it again." | Download the package and its .sha256 file again, and upload the package again. |
| License updates | "Your license includes updates until 2026-09-30. This version is from 2026-10-16. Renew the software maintenance to install it." | Renew the updates of your license (31.18.8) or install a version from before that date. |
| Free disk space | "Not enough free disk space for this update: X needed, Y free. Delete recordings you no longer need, or contact support." | Delete recordings you no longer need (an update needs several times the size of the package), then upload again. |
| System packages | "This package needs system packages that are not on this probe. Contact support." | Contact support; this is a problem of the release. |
| Probe setup | "This probe needs a change that only support can make before this version can be installed (X). Contact support before installing." | Contact support. |
Two checks can pass with a warning (amber), which the dialog of Install repeats:
- License updates: "No valid license: IPProbe starts with all licensed options off after the update, as it does now." The probe has no valid license; the update goes ahead and the probe starts restricted, as it does now (32.1).
- Probe setup: "The probe restarts during this update. Monitoring stops for about N minutes." The new version needs a restart of the probe. The update does it by itself and continues after the restart.
If the probe refuses the package once more when you click Install, the page says The probe checked the package again and refused it. The red lines below say why.
31.18.4 While the update runs
All users see a banner at the top of every page: IPProbe is being updated to 2026.11.0. Monitoring stops for up to 2 minutes. (or, when the probe restarts: IPProbe is being updated to 2026.11.0. The probe restarts; monitoring stops for up to N minutes.). Every user is logged out when IPProbe stops.
The Update page lists the steps: Queued, Stopping, Backing up, Installing, Restarting the probe (only when the update needs it), Starting and Checking. The current step is marked. The step Checking takes about a minute: the new version must run without a restart before the update counts as done.
When the connection to the probe is lost, every open page shows the screen IPProbe is being updated:
- Waiting for IPProbe to come back... This can take up to N minutes. The time is an upper limit, not a promise: Expected says up to. The screen stays for at least 30 seconds even when the probe answers sooner, so you can read it; the probe itself is not delayed.
- Elapsed counts the time. When the update takes longer than twice the expected time, the screen says This takes longer than usual. If the probe does not come back, check the screen connected to it for its address. (31.17).
- Checking whether IPProbe is back... shows that the page is trying. When it stopped trying, Check again tries once more.
- When IPProbe is back, the page opens the login. Everyone must log in again. The new web interface loads by itself; if a page was open for a long time, the banner This web page is version X, but IPProbe Y runs on the probe. offers Reload; if the message stays, clear the browser cache.
Do not switch the probe off or pull the cable while an update runs. If power fails, the probe finishes the update or goes back to the previous version when it starts again.
31.18.5 The result
After the update the Update page shows the result in the panel Update to ... (and as a banner for all users, which closes after 30 seconds):
| Result | Meaning |
|---|---|
| IPProbe 2026.11.0 is installed. | The update went well. Last update shows Installed. |
| The update did not start correctly and IPProbe went back to 2026.10.0. | The new version did not run correctly, so the probe returned to the previous version with its settings. The Reason says why. Last update shows Rolled back. |
| The update was stopped before it installed anything. IPProbe 2026.10.0 runs as before. | Something interrupted the update before the installation (for example a power failure). Nothing was changed. You can try again. |
| The probe refused the package when the update started. | The probe checked the package again at the start and refused it (the reason is shown), or the update service did not start the update. Nothing was changed. |
| The update failed. | The new version did not run and going back did not work either. Contact support. |
For every result except a successful one the page shows Contact support and send them a support bundle of this probe with a Support bundle button (31.15). Log of the update opens the last lines of the update log; send it to support too. Every update, Go back and refusal is written to the audit log (31.7) and, as IPProbe update, to the incidents (info when it went well, warning when it went back or was refused, error when it failed).
31.18.6 Go back
The probe keeps the version that was installed before the last update (one version). Go back shows Version 2026.10.0 is kept on the probe. and the button Go back to version 2026.10.0; without a kept version it says No previous version is kept on this probe.
Go back installs the kept version again with the settings it had before the update. The license stays as it is now, so a renewed license is not lost. The dialog shows What Go back undoes before you confirm:
- the settings changed since the update (Settings changed since the update are undone: Aliases, Checks.) or No settings were changed since the update.;
- the passkeys added since the update, which are removed (Passkeys added since the update are removed: ...);
- the history the newer version recorded, which is kept aside (31.18.7).
Confirm with Your password and Go back. Monitoring stops for a short time, as for an update; the banner and the screen IPProbe is being updated appear with "going back". The result is IPProbe went back to 2026.10.0., or Going back to 2026.10.0 failed. with the support advice above. If the kept files were changed, Go back is refused: The kept files of version X were changed; Go back is not possible. Contact support. To go further back than the kept version, contact support.
31.18.7 History and Go back
An update to a version that uses a newer format of the history database converts the database. Before it does, the probe makes a copy of the old database. A Go back puts that copy back, so the older version starts with its own format and all the history up to the update. The converted database is not deleted: it is kept aside, and the history that the newer version recorded between the update and the Go back is in it only.
While the older version runs, the Update page and Settings > System (the history database, 31.1.7) say History from 2026-10-23 10:15 to the Go back was recorded by IPProbe 2026.11.0 and is kept aside; it is shown again after updating to 2026.11.0. That period is missing from the history until then. When you update to that version again, the probe takes the kept database back in use, and the period you spent on the older version is then missing and kept aside in its turn. Nothing is deleted by the probe except the oldest copy when more than two of a kind exist; an Admin can download or delete the copies under Settings > System (Copies of other versions, 31.1.7).
31.18.8 Updates until, and license
The license says until which release date software updates are covered (Update expiration on the License page, 32.2). The Update page shows it as Updates until. The probe checks it before installing, so an update never makes a working probe restricted: a version released after that date is refused with Your license includes updates until .... Versions released up to that date stay available. A license with an update date in the year 2200 or later has no end date.
To renew: send the Unique System ID from the License page to sales, order the software maintenance, and upload the new license file on the License page (32.4). No restart is needed, and the refused package can be uploaded again.
IPProbe always starts. If a version is installed that the license does not cover (for example by service staff), or the license is missing or invalid, the probe starts restricted: all licensed options are off, the pages show the reason and the License page explains it (32.1). Installing a valid license ends this at once.
31.18.9 Settings from a newer version
If a settings file on the probe was written by a newer IPProbe version than the one that runs (this can only happen after service staff installed an older version by hand), the page of those settings shows These settings come from IPProbe X. They are read as far as this version understands them and cannot be changed here. The About page lists them as well. An Admin can choose Reset to defaults in the notice (after a confirmation) to replace those settings with the defaults of this version, or restore a backup (31.9); or contact support. The description is in chapter 31 (Settings written by a newer version).
A Go back does not cause this: it puts back the whole set of settings of the older version.
31.19 Alarm Export
Settings > Alarm export sends the alarms of the probe to other systems, so they reach people and tools that do not watch the IPProbe pages: the DVBControl Alert Center, a syslog server, an SNMP manager, email, Zabbix, Nagios, Slack, Microsoft Teams or any system that takes a webhook. An alarm is raised when a condition starts (for example an RTP sequence gap on a flow, a link down, No Data) and cleared when it ends. The same alarms are on the Incidents page.
An Admin sets the page up and tests the targets. NOC sees the state of every target and can download the files for monitoring systems, but cannot change anything. Management does not see the page. Nothing is sent until a target is switched on; all targets are off in a new probe.
Every alarm names the probe by its Probe Name (Settings > System, 31.1), not by the host name of the machine, in the email subject and text, the digest, the webhooks, the Alert Center, the syslog message and the SNMP traps. Syslog keeps the host name of the machine in its HOSTNAME field and has the Probe Name in the message. An alarm about a PID of a program in an MPTS stream also names the service ("service Das Erste HD (28106)": the SDT name and the program number); an alarm without a service, such as one about an SPTS or an ST 2110 flow, has no such part. In the Alert Center the service shows in its Service column.
How each system is connected in detail (message formats, OIDs, the Zabbix template, the Nagios plugin, signatures) is in the integrator guide (alarm-export.md in the IPProbe documentation).
31.19.1 The Page
The General section (31.19.5) is at the top and always in view. Below it each kind of target has its own tab: Alert Center, Syslog, SNMP, Email, Zabbix and Nagios, Webhooks and, for an Admin, API keys. The dot on a tab is green when the target (or one of its receivers) is switched on and gray when it is off. Each target is a card with Enabled, its settings, a Filter, the switch Send repeats and a status line. Send repeats (on by default) lets the target also get the repeats of an alarm that goes on, such as the ETR 290 repeats of 18.5.4; switched off, the target gets one raise and one clear per alarm. Email folds a repeat into the mail of its raise while that mail waits and never mails it on its own. Change what you need and click Save at the top; Undo changes puts the page back to what is saved. A value that is not valid is marked in the card, its tab shows a red mark with the number of values to correct (the page opens that tab after a refused save), and the page is not saved until it is corrected.
At the top the page shows how many alarms are active and how many were raised since the probe started. If the alarm export is not running on this probe, a notice says so and no alarm is sent.
Status line. Each target shows its state and counters:
| State | Meaning |
|---|---|
| Off | The target is switched off. |
| Connecting | The probe is connecting. |
| Connected / Ready | The target works. SNMP, email and webhooks show Ready: they connect when an alarm is sent. |
| Retry in N s | The target is not reachable; the probe tries again and keeps the alarms that can be kept. |
| Waiting for approval | Alert Center only: an Admin must approve this probe there. |
| Rejected | The receiver refused the probe or the message. The reason is shown. |
| Error | The settings cannot be used or the receiver reported an error. The text says which. |
The counters are sent, dropped (by the filter, the rate limit or a full queue; hover for the split), queued and active (the alarms of this target that are still open). A message below the state shows the last error, and a warning if a flow group in the filter no longer exists.
Test. The Send test alarm button of a target (Admin only) sends one test alarm and its clear to that target with the saved settings and shows whether it was accepted. Use it after every change of a target.
Test connection. The Test connection button (Admin only) checks whether the probe can reach the receiver with the values as they are typed, before you save them. Nothing is sent and nothing is saved; a password field you did not touch uses the stored password. The line below the button says Connected (12 ms) or which step failed and why:
| Step | What failed |
|---|---|
| DNS lookup | The host name does not resolve. |
| TCP connect | The receiver refused or did not answer the connection (wrong address or port, a firewall). |
| TLS / certificate | The TLS handshake or the certificate check failed. With a pin, the server's SHA-256 is shown to compare. |
| Proxy | The webhook proxy cannot be reached or refused the connection. |
| WebSocket upgrade | Alert Center: the server answered, but not as an Alert Center (wrong path or another service). |
| HTTP request | Webhook: the receiver did not answer, or answered 401 or 403 (reached, but refused: check the token or headers) or 404 (reached, but the path was not found). Any other HTTP answer counts as reached; its code is shown. |
| SMTP greeting, STARTTLS, SMTP login | Email: the mail server refused the session, does not offer STARTTLS, or refused the user name or password. |
| Settings | A value cannot be used (the field is named). |
SNMP traps and syslog over UDP have no connection: only the address is looked up. The Alert Center test does not ask for an approval, so it also works before the probe is approved. Each step waits at most 5 seconds, the whole test at most 10 seconds, and one connection test runs at a time.
Filter. Every target has its own filter, so a mail can go to the duty engineer for major alarms only while the syslog server gets everything:
- Minimum severity - the target gets alarms at or above this severity (default Warning). The severities are Info, Warning, Minor, Major and Critical.
- Alarm types - all types, or a chosen list of types to include or to leave out.
- Flow groups - only the flows in the chosen groups (Settings > Groups). System alarms are not filtered by group.
The start and stop of IPProbe are Info events; to receive them, set the minimum severity to Info.
SCTE-104 messages (type scte104) are Info events too, one per message the log lists (a duplicate the encoder marks is not sent). They are off in every target unless chosen: select SCTE-104 message in the target's alarm types (selecting all types does not include it) and set the minimum severity to Info.
Secrets. Passwords, communities, signature secrets and webhook URLs are write-only. After saving they show as "set" and are never displayed again. Type a new value to replace one, or use Clear to remove it. They are not part of a configuration backup: after a restore on another probe, enter them again.
31.19.2 Alert Center
| Setting | Description | Default |
|---|---|---|
| URL | Address of the Alert Center, wss://alertcenter:8443/ws/client. Without a path /ws/client is added. Any certificate is accepted. | empty |
| Machine name | How the probe is listed in the Alert Center. Shown only: it is always the Probe Name of Settings > System. When the Probe Name is changed, the probe connects again under the new name; the approval stays. | The Probe Name |
| Acknowledge from | Alarms at or above this severity must be acknowledged by an operator in the Alert Center. | Critical |
| Machine UUID | The identity of this probe, with a copy button. Made once. |
The first time, the state is Waiting for approval: an Admin of the Alert Center approves this probe (by its machine UUID) there. From then on the probe sends its active alarms every time the connection is made, so the Alert Center never shows an alarm that has ended. If the Alert Center snoozes the probe, its alarms are hidden there until it is unsnoozed.
31.19.3 Syslog
Up to four receivers. Per receiver: Host and Port (514, TLS 6514), Transport (UDP, TCP or TLS), Framing for TCP and TLS (octet counting or one line per message), Facility (local0 by default), App name (ipprobe) and for TLS the TLS trust:
- Certificate authority - the certificate must be issued by a CA you paste (PEM) or, when empty, by the CA store of the system (on Windows paste the CA).
- Pinned certificate (SHA-256) - enter the fingerprint of the receiver's certificate; fits a self-signed receiver.
- No check - any certificate is accepted.
A syslog message carries the Probe Name, alarm id, type, severity, state, flow, service and counters as structured data alarm@2696. Heartbeat messages with the number of active alarms are sent regularly (General).
31.19.4 Zabbix and Nagios
This card serves two ways of working:
- Polling - Zabbix and Nagios ask the probe for its alarm status with a read-only API key (31.19.8). The card offers the files to download: the Zabbix template, the check_ipprobe plugin and a sample Nagios configuration. Importing and configuring them is described in the integrator guide.
- Pushing - Zabbix sender (up to two receivers) sends the alarms to a Zabbix server or proxy: Zabbix server or proxy, Port (10051) and Host name in Zabbix, the host the template is linked to.
31.19.5 General
- Heartbeat (minutes) - syslog, SNMP and the Zabbix sender get one message with the number of active alarms every N minutes, so a silent probe can be told from a quiet one. 0 is off, up to 60. Default 5.
- Severity per alarm type - changes the severity of one alarm type for all targets (for example No Data from Major to Critical). The table shows the default of each type.
31.19.6 Email
| Setting | Description | Default |
|---|---|---|
| Server, Port | The SMTP server. | Port 587 (465 with TLS) |
| Security | None, STARTTLS or TLS. | STARTTLS |
| User, Password | The login. Empty user: no login. | empty |
| Sender address | The From address of the mails. | |
| TLS trust | As for syslog. | Certificate authority |
| Recipients | Up to 20. Per recipient: the address, the minimum severity (default Major), whether clears are mailed (default yes), and the daily digest and/or weekly digest. | |
| Digest | The time (probe local time, default 07:00) and the weekday of the weekly digest (Monday). |
Alarm mails. One mail per alarm; alarms that follow each other within 30 seconds are combined into one mail that lists them. A recipient gets at most 20 alarm mails an hour; the alarms over that are counted and reported in a summary mail every 15 minutes. A clear is mailed in the thread of its raise.
Digest. A daily or weekly overview of the alarms of the previous day or the previous 7 days: how many, how long, the mean time to recover, what is still open, the busiest flows and the longest alarms, compared with the period before. Send digest now sends the daily (yesterday) or weekly (last 7 days) digest at once. A digest that was missed because the probe was off is sent at the next start.
31.19.7 SNMP Traps and Webhooks
SNMP traps. Up to four targets, each with up to four managers (host and port, default 162). Choose SNMPv2c with a Community, or SNMPv3 with a User, a Security level, an Authentication (SHA-256 by default) and Privacy (AES-128) and their passwords (at least 8 characters). The probe sends traps only. For SNMPv3 the manager must be told the Engine ID shown on the card (copy button). Download MIB saves IPPROBE-MIB.txt, which a manager needs to show the traps by name.
Webhooks. Up to four. Format: generic JSON, Slack or Microsoft Teams. URL of the receiver (kept secret, because Slack and Teams put a token in it), a Timeout (5 to 60 s), Batch (JSON: up to 50 alarms in one request), Sign with HMAC-SHA256 with a Signature secret (JSON: the receiver can check that a request comes from this probe), an optional Proxy and up to four Extra headers (for example an API key). If the receiver is busy or down, the probe retries for up to an hour. Any certificate of an HTTPS receiver is accepted, and redirects are not followed.
31.19.8 API Keys
For systems that poll the alarm status (Zabbix, Nagios), an Admin creates a read-only key:
- Click Create key, enter a name that tells where it is used (for example
Zabbix) and, if wanted, an expiry date. - Copy the key at once (it starts with
ipk_). It is shown only once; the probe keeps only a fingerprint of it. - Enter the key in the monitoring system.
The list shows when each key was created, when it expires and when and from which address it was last used. Revoke stops a key at once. A key can only read the alarm status; it cannot change anything. Use one key per system. NOC users see the list but cannot create or revoke keys. Keys are not part of a configuration backup.
31.19.9 After a Restart and Troubleshooting
When the probe starts it sends the event "IPProbe started; alarms of earlier runs are void", and it sends its open alarms again to the Alert Center. A receiver should therefore treat everything it knew from before the start as over.
- A target shows Retry in N s or Error - read the text under the state: it names the host and the reason (address, port, firewall, login, certificate). Fix the setting and use Test.
- TLS: certificate not trusted - the text shows the SHA-256 fingerprint of the certificate of the receiver. If it is the right receiver, pin that fingerprint or paste its CA.
- Waiting for approval - approve the probe in the Alert Center.
- Alarms are missing at a target - check its minimum severity and type filter; the dropped counter shows how many the filter or a limit held back. A receiver that is slow or unreachable for long can lose alarms; the probe then sends one "Alarms were lost" event with the counts.
- Many alarms of the same kind - to protect the receivers the probe sends at most 20 alarms per flow, alarm type and hour as separate alarms; the rest is summed in one alarm called "+N more".
- Webhook rejected - the state shows the HTTP code. 404 or 410 means the URL is wrong or removed; 401 or 403 means the receiver wants a login (use an extra header).
32 License
The License page shows whether this probe is licensed, the Unique System ID that a license is made for, the details of the installed license and the options it includes. An administrator installs a new license file here.
Every logged-in user sees this page. Only an Admin can upload a license.
32.1 Status
| Item | Description |
|---|---|
| Status | Licensed, Evaluation license or Not licensed. Not licensed also gives the reason, for example "No license installed", "License is for another machine" or "License expired on 2026-09-30". |
| Unique System ID | The identity of this machine. A license only works on the machine it was made for. Click the copy button and send the ID to sales to order a license or an extra option. |
| Upload license... | Admin only. Installs a license file (.slc, or a .zip that holds it), see 32.4. |
Without a valid license the probe runs restricted: you can log in, open the License page and install a license, but the probe does not capture or analyze traffic. A license that has expired, or a probe software version released after the license's update expiration date, also makes the probe restricted. The analysis pages (Protocols, Flows, Endpoints, Devices, Network Map, Packet Timing, Compare, Reports and Recordings) are grayed out with a lock ("Requires a valid license") and open a page that says the probe has no valid license. Dashboard, Inputs, Logs, Settings, License and About stay available.
32.2 License
Shown for a valid or evaluation license; hidden when the probe is not licensed.
| Field | Description |
|---|---|
| Name | The person the license was issued to. |
| Company | The company the license was issued to. |
| Serial | The serial number of the license. |
| License type | Full, or Evaluation (shown in the info color). An evaluation license puts a DEMO mark in pictures and video. |
| Purchase date | The date of purchase. |
| Expiration date | The last day the license is valid, or Perpetual for a license that does not expire. |
| Update expiration | Software versions released up to this date run with this license, with the days left. |
A date within 30 days is shown in the warning color, a date that has passed in the error color.
Update expiration and updates. The update expiration date is the last release date of IP-Probe that this license covers. The Update page (31.18) checks it before a new version is installed and refuses a version released after that date, so a probe that works never drops into restricted mode through an update. A date in the year 2200 or later means updates have no end date. To renew the updates, send the Unique System ID to sales and upload the new license file (32.4). IP-Probe always starts: when a version is running that the license does not cover, or the license is missing or invalid, it starts restricted (32.1) with all licensed options off and the reason shown on this page, and a valid license ends this at once. An evaluation license gets the same check.
32.3 Entitlements
The license options, each marked Included or Not included:
| Option | What it enables |
|---|---|
| Base | The probe itself: flows, Endpoints, Devices, traffic, logs, reports, recordings, jitter and dashboards. |
| MPTS | MPEG-TS multi-program analysis: programs, PIDs, service information and thumbnails. |
| Streaming | SRT and RIST inputs and stream analysis. |
| 2110 | PTP, the ST 2110-10, -20, -30 and -40 pages and the test stream generator. |
| 2022-6 | ST 2022-6 uncompressed video over IP. |
| Audio | Audio meters, waveform, loudness and listening, on every page that has them. |
| Video | Pictures, thumbnails, video playback and scopes, on every page that has them. |
| Ancillary | Ancillary data: ST 2110-40, subtitles and captions. |
Pages and options not included
- A menu entry of an option that is not included is grayed out with a lock. Its tooltip names the option, for example "Requires the Streaming option".
- Opening such a page shows "... is not licensed" with the option it needs and an Open License button.
- A page that combines options opens with its protocol option and shows a lock in place of the parts of another option. For example, the ST 2110-30 page without the Audio option shows the stream analysis, but "Needs the Audio option" in place of the meters, the waveform and the listen buttons. The AV Monitor opens with the Video or the Audio option and shows only the part that is licensed.
- A dashboard widget whose data needs an option that is not included shows the same lock.
The navigation menu of a probe whose license includes Base, MPTS, 2110 and Video:
The page of an option that is not included, here SRT without the Streaming option:
Notice in the header
A notice at the top of every page links to the License page when:
- the probe is not licensed ("Not licensed: analysis is stopped");
- an evaluation license is installed ("Evaluation license");
- the license, or its update expiration, ends within 30 days.
32.4 Installing a License
- Copy the Unique System ID and send it to sales. You receive a license file (
.slc) made for this machine. - Log in as an Admin and open the License page.
- Click Upload license... (in the guide of the Status card, or at the top right when a license is installed) and choose the
.slcfile, or the.zipit came in: the probe takes the.slcout of it.
The probe checks the file and applies it at once, without a restart: the page shows "License installed", the menu and the pages follow the new options. If the file is not accepted, the page shows the reason (for example "License is for another machine") and the installed license stays as it was.
32.5 Requesting a License
While the probe has no valid license, the Status card shows a short guide under the reason and the Unique System ID:
- Request a license with the Request license... button.
- Send the file to your supplier.
- Upload the license you receive with the Upload license... button.
The buttons are for Admin users only. Other roles see the same three steps without buttons and the line "Ask your administrator to license this probe." Once the probe is licensed, the guide is gone and Upload license... is at the top right of the Status card.
To request a license:
- Click Request license....
- Fill in Company, Contact name, Email, Phone number and Occupation. All fields are required; the email address must contain an @ and no field may be longer than 128 characters (a character is one letter, also for non-Latin scripts).
- Click OK. The browser downloads the file
license_request.licand the page confirms: "License request file downloaded (license_request.lic). Send it to your supplier."
Cancel closes the dialog without making a file. If the probe refuses the request, the dialog stays open and shows the reason.
A license file must be installed within 31 days after it was made, and a license cannot be replaced by an older version of the same license.
33 About
The About page displays system and application information:
| Field | Description |
|---|---|
| Application | IP-Probe by RTSS B.V. (Real-Time Software Solutions). |
| Version | The current software version number (for example 2026.10.0), the release date and the build date. |
| Backend Version | The version of the C++ backend engine. |
| Frontend Version | The version of the Angular web interface. |
| System | Operating system, kernel version, and hostname. |
| Uptime | How long IP-Probe has been running since the last start. |
What's new. The card What's new in 2026.10.0 shows the release notes of the running version, in the order Implemented, Fixed, Changed and Updated. Show all versions opens the notes of this and every older version. The notes come with the installation; the card says so when a version has none. The update itself is on Settings > Update (31.18): there the same notes are shown for a new package before you install it.
Notes on this page: The web page is version X; reload to get Y with a Reload button appears when the web interface in your browser is older than the probe (after an update). For settings written by a newer version the page shows The X settings come from IPProbe Y. This version reads them but cannot save changes to them. (see chapter 31), and an Admin sees a note about history that a Go back kept aside (31.18.7).
Host health shows the probe's own hardware, live (it refreshes every 5 seconds). The header names the machine, its processor, the memory and the operating system. Under it are tiles for:
- CPU - the processor temperature.
- Memory - the percentage used, with the used, total and available amounts, the hugepages and the memory of the IP-Probe process. A badge next to the value tells whether the memory has ECC (error correction), see below.
- Storage - the percentage of the disk used that holds the probe data (the install folder with the settings, the history and the recordings), with the used, total and free amounts, the folder and the disk. When the recordings are on another disk, a second tile Storage - Recordings shows that disk. The tile warns at 85 percent and goes critical at 95 percent (Settings > System health); delete old recordings or move them off the probe.
- NVMe - the temperature of the system drive and its critical limit.
- GPU - the temperature of the graphics processor, when there is one.
- Memory modules - the temperature of the memory modules (DIMMs), when they have a sensor.
- NIC chip - the hottest NIC chip (or one in warning), its state and the port it belongs to. Click the port name to open it on the Inputs page. There is no NIC chip tile when no chip sensor can be read.
- CPU load - the load average, and the busy percentage of the cores of the operating system and web server and of the packet-processing cores (these sleep when idle, so a low value is normal).
Each temperature tile has a small graph of the last hour, and the state follows the limits under Settings > System health.
Memory ECC badge. The badge next to the memory value shows whether the memory of the probe corrects errors (ECC) and, on Linux with the EDAC driver loaded, how many errors it has corrected. Hover over the badge, or focus it, for the details. The same badge is in the Host health dashboard widget.
| Badge | Meaning |
|---|---|
| ECC (green) | The memory has ECC. The tooltip shows the corrected and uncorrected error counts since the probe booted ("ECC active, error counters not available" when the platform has no counters, as on Windows). |
| ECC: errors corrected (amber) | Corrected errors are rising: the tooltip names the number in the last period and the module that has most of them. A module may be failing; plan to replace it. |
| ECC: uncorrected errors (red) | The memory reported an error it could not correct. Replace the failing module. |
| No ECC (gray) | The memory has no error correction. No memory error check applies. |
| ECC unknown (gray) | The probe could not determine it. Check the BIOS setting, or load the EDAC driver, for example modprobe amd64_edac. |
The Network ports pane lists every NIC port that has link, with its card, chip temperature and speed; click a row to open the port on the Inputs page. The pane is not shown when the probe has no NIC ports.
The About page also provides links to:
- IP-Probe website - ip-probe.com
- RTSS website - rtss.eu
- Documentation - Online documentation.
- Support - Contact information for technical support.
The copyright notice at the bottom of the page always shows the current year.
IP-Probe is a trademark of RTSS B.V. (Real-Time Software Solutions).
34 Performance and System Requirements
This chapter shows what one IP-Probe handles, measured on real network traffic, and what hardware to choose for a probe. All numbers come from live tests with a separate machine sending the traffic over direct cables into the probe's network ports.
IP-Probe is delivered as an appliance, so the processor, memory and network cards are chosen and sized for you. This chapter shows what one probe handles. The probe has two network ports (25 Gbit/s and 10 Gbit/s) and was measured on 12 and 16 October 2026.
34.1 Measured performance
"No loss" means that every packet sent was received and analyzed: the network card dropped none after the run-in period (34.2).
| Test | Load | Result | Measured |
|---|---|---|---|
| ST 2110 studio traffic at 25 Gbit/s line rate | 24.5 Gbit/s, 2.2-2.3 million packets/s | No loss, in both receive modes | 12 Oct 2026 |
| ST 2022-6 at 25 Gbit/s line rate | 24.5 Gbit/s, 2.1 million packets/s | No loss | 12 Oct 2026 |
| Two 25 Gbit/s ports at once | 49.1 Gbit/s, 4.6 million packets/s | No loss, in both receive modes | 12 Oct 2026 |
| Small packets (64 bytes), 256 flows | 37.2 million packets/s (25 Gbit/s line rate) | No loss in rss mode; pipeline mode takes 10.5 million packets/s | 12 Oct 2026 |
| One single flow | up to 2.1 million packets/s | Pipeline mode: no loss; rss mode: about 1.5 million packets/s per flow | 12 Oct 2026 |
| 400 MPTS streams, all ETR 290 checks | 17.7 Gbit/s, 1.6 million packets/s | No loss, 0.8 GB of memory | 16 Oct 2026 |
| 800 MPTS streams on two ports, 30 minutes | 26.6 Gbit/s (a 25G and a 10G port), 2.4 million packets/s | No loss, control core 5% busy, 1.4-1.5 GB of memory | 16 Oct 2026 |
| ETR 290 cost | 400 MPTS streams, priorities 1-3 | +14 ns (11%) per TS packet | 12 Oct 2026 |
| Thumbnails of 100 MPTS streams, 10 minutes | 20 pictures per second | +121 MB of memory, back within 6% after closing the page, no loss | 16 Oct 2026 |
| 200 MPTS streams, monitoring switched off for some | 8.9 Gbit/s | The analysis load falls with the share switched off (half off: half the load) | 16 Oct 2026 |
| New flows: 2 million new flows per second | 10 Gbit/s port | Up to 17 000 packets lost in the first second, none after | 12 Oct 2026 |
Stream count: one probe of this class handles 800 MPTS streams with all ETR 290 checks; 400 streams leave plenty of room.
34.2 Run-in period after start and link-up
When the probe starts on a busy link, and each time the link of a port comes up, the network card can lose packets for a moment: at 25 Gbit/s with hundreds of streams, up to about 20 000 packets in the first two seconds. The streams themselves are fine; it is the start of the capture.
For this reason the first 5 seconds after a start or a link-up are a run-in period (Settings > System, Run-In After Start or Link-Up (s), see 31.1). During the run-in the probe counts every packet and every loss, but raises no incidents, no ETR 290 errors and no "No Data" alarms for that port. The loss stays visible in the port counters (missed packets on the Inputs page); the "NIC Rx loss" log line of the port is written again from the first second after the run-in. From the first second after the run-in, every fault is reported as usual. A stream that is still missing after a link-up is reported as No Data when the run-in ends.
34.3 Receive modes: pipeline and rss
IP-Probe receives its network ports in one of two modes, chosen at start (--rx-mode):
| Mode | How it works | Use it for |
|---|---|---|
| pipeline (default) | One receive core per port hands the packets to the analysis cores | Every broadcast load, and single flows above about 1 million packets/s (for example UHD ST 2110-20 or ST 2022-6) |
| rss | The network card spreads the flows over all analysis cores | Very many small packets (more than about 10 million packets/s per port) |
In rss mode one flow always stays on one core, which handles about 1.5 million packets per second. When a flow comes close to that, the probe writes a warning to the log and to System Health for the port: "Flow ... is near the rss limit of one Worker ...: use pipeline mode for this port."
34.4 Capacity
One IP-Probe appliance has two network ports (25 Gbit/s and 10 Gbit/s). It handles 25 Gbit/s of ST 2110 or ST 2022-6 traffic at line rate on a port, 49 Gbit/s on two ports at once, and 800 MPTS streams with all ETR 290 checks, without packet loss (34.1).
Every packet-processing task has a processor core of its own, which polls continuously: in the operating system these cores show as busy whatever the traffic. The real load of each core is on the Diagnostics page (chapter 30). The network cards produce most of their heat as soon as the ports are active; the appliance is cooled for this, and the card temperature is shown on the Inputs page and in System Health (31.8).
A Shortkeys
A.1 General
| Short-Key | Description |
|---|---|
| ALT + H | Toggle for Help pages. |
| ALT + F | Toggle to get Help pages, full screen. |
| ALT + S | Toggle for Settings. |
| ALT + T | Toggle between the 2 Themes: Light-Mode and Dark-Mode. |
| Short-Key | Description |
|---|---|
| ALT + D | Go to Dashboard view |
| ALT + F | Go to Flows view |
| ALT + I | Go to Endpoints view |
| ALT + V | Go to VLANs view |
| ALT + M | Go to Devices view |
| Short-Key | Description |
|---|---|
| ALT + 1 | Go to Card view |
| ALT + 2 | Go to Grid view |
| ALT + 3 | Go to List view |
B Supported protocols
- SMPTE ST 2059 (PTP)
- SMPTE ST 2110
- 2110-20: Uncompressed video
- 2110-21: Traffic shaping
- 2110-22: Compressed video (JPEG XS)
- 2110-30: PCM audio
- 2110-31: AES3 audio
- 2110-40: Ancillary data
B.1 SMPTE ST 2059 (PTP)
Timing is absolutely critical in live production. ST 2110 relies on Precision Time Protocol (PTP) to synchronize every device down to the microsecond.
PTP replaces:
- Black burst
- Tri-level sync
- Embedded timing in SDI
This enables:
- Frame-accurate switching
- Multi-camera sync
- Lip-sync between video and audio
Clocking must be extremely stable, otherwise sync issues appear.
B.2 SMPTE ST 2110
SMPTE ST 2110 is a suite of standards that defines how professional media - video, audio, and metadata - is transported over IP networks in real-time. It replaces traditional SDI infrastructures and enables highly flexible, scalable, and software-driven broadcast systems.
The core idea is essence-based transport:
Video, audio, and ANC data are sent in separate streams, not bundled together like SDI.
This allows devices to subscribe only to what they need, making routing and processing dramatically more flexible.
Goals of the ST 2110 Family
- Enable uncompressed, real-time IP media for live production
- Achieve SDI-equivalent timing performance on IP networks
- Support scalable, multi-vendor interoperability
- Allow independent routing of video, audio, metadata
- Use standard IT networking technologies (Ethernet, multicast, RTP)
Network Requirements
ST 2110 assumes a professional IP network, generally with:
- Multicast routing (IGMP, PIM)
- QoS (DiffServ)
- Non-blocking switches
- 10/25/40/100 GbE links
- Traffic shaping (per 2110-21)
- Redundancy using SMPTE 2022
2022-7 provides seamless protection switching:
Two identical streams are sent over two paths ? the receiver merges them ? zero-downtime even if one path fails.
Control and Discovery
ST 2110 does not define how devices find and manage streams.
This is handled by NMOS standards:
- IS-04: Discovery & registration
- IS-05: Connection management
- IS-08: Audio mapping
- IS-09: System parameters
- IS-11: Capabilities
Together they allow plug-and-play IP broadcast systems.
Typical Use Cases
- Live broadcast studios and control rooms
- Sports production / OB trucks
- Remote and distributed production
- Cloud-based or hybrid production
- Post-production and graphics systems
- Virtualized routers replacing SDI routers
High-Level Architecture
The ST 2110 ecosystem consists of:
1. Essence transport standards
- 2110-20: Uncompressed video
- 2110-21: Traffic shaping
- 2110-22: Compressed video (JPEG XS)
- 2110-30: PCM audio
- 2110-31: AES3 audio
- 2110-40: Ancillary data
2. Timing & synchronization
- SMPTE ST 2059 (PTP-based timing)
3. Control & discovery
- AMWA NMOS IS-04 (discovery)
- NMOS IS-05 (connection)
4. Redundancy & protection switching
- SMPTE 2022-7
Together these create a complete IP-based live-production environment.
B.2.1 ST 2110-20: Uncompressed Video
SMPTE ST 2110-20 is the part of the SMPTE 2110 family that defines how uncompressed video essence is transported over IP networks within professional live-production environments (broadcast, OB vans, production studios, etc).
Video is carried inside RTP (Real-Time Transport Protocol) using a well-defined packet structure.
RTP offers:
- Sequence numbers (frame order)
- Timestamps (synchronization)
- Flexibility for different frame rates and pixel formats
ST 2110-20 supports a wide range of professional Sampling Formats:
- YCbCr 4:2:2 10-bit (most common for broadcast)
- RGB 4:4:4 (graphics / high-end production)
- HDR formats (PQ, HLG)
- HD, 1080p, 1080i, 2160p/UHD, and beyond
Streams usually use IP multicast, allowing many receivers to consume the same video feed without duplication of traffic.
Precise timing comes from SMPTE ST 2059, a profile of IEEE-1588 PTP (Precision Time Protocol).
This ensures:
- Frame-accurate synchronization
- Multi-camera alignment
- Deterministic behavior even across switches and long fiber paths
Redundancy
ST 2110 workflows often use SMPTE 2022-7 seamless protection switching, providing two identical streams over different network paths.
Receivers merge them to avoid pixel dropouts.
B.2.1.1 Advantages
Scalability
Add or reroute streams without rewiring - just change IP subscriptions.
Flexibility
Any device can become a source or receiver at any time.
Interoperability
The 2110 family ensures multi-vendor systems work together (especially when combined with NMOS IS-04/IS-05).
Near-Zero Latency
No codec delay - essential for live sports, live news, and real-time switching.
B.2.1.2 Network Requirements and Engineering Considerations
Network engineering is complex
Requires:
- Proper QoS
- PTP architecture
- Multicast routing
- Traffic shaping
High Bandwidth
Because video is uncompressed, bandwidth can be very high:
| Image Size | Frame Rate | Bits per Pixel | Chroma Sampling | Bandwidth |
|---|---|---|---|---|
| 1280 x 720 | 60 fps | 24 (8-bit color) | 4:4:4 | 1 Gbps |
| 1920 x 1080 | 60 fps | 24 (8-bit color) | 4:4:4 | 1 Gbps |
| 1920 x 1080 | 30 fps | 10 | 4:2:2 | 3 Gbps |
| 3840 x 2160 | 60 fps | 10 | 4:2:2 | 12 Gbps |
This is why 10/25/40/100 GbE fabrics are common.
B.2.1.3 Typical Use Cases
Studio and Control Rooms
2110-20 feeds cameras, replay servers, multiviewers, and switchers with essentially zero latency.
Remote/Distributed Production
Cameras at a stadium, mixing console at a broadcast center - all synchronized over IP.
Post Production / VFX
High-quality uncompressed media can be switched and shared between systems without SDI infrastructure.
Software-Based Processing
Transcoders, graphics engines, and QC tools can run on COTS (off-the-shelf) hardware.
B.2.2 ST 2110-21: Traffic Shaping / Sender Behavior
Defines how senders must pace packets:
- Avoids bursts that congest switches
- Two main profiles:
- Narrow Senders: predictable, switch-friendly
- Wide Senders: more flexibility, typically software
This ensures networks remain stable under heavy loads.
Traffic Shaping
The standard includes rules for:
- Maximum packet rates
- Packet spacing (to avoid burstiness)
- Consistent flow profiles
This keeps networks predictable and prevents packet microbursts.
B.2.3 ST 2110-22: Compressed Video (JPEG XS)
- Allows use of low-latency compressed video, e.g. JPEG XS
- Maintains "production quality" while reducing bandwidth
- Often used when UHD over long distances is required
B.2.4 ST 2110-30: PCM Audio
- Carries uncompressed audio using AES67 rules
- Sample rates: 48 kHz (typical), 96/192 possible
- Supports multi-channel audio and mixing environments
B.2.5 ST 2110-31: AES3 Audio (Non-PCM)
- For Dolby E, AC-3, metadata-carrying audio formats
- Preserves bit-exact AES3 framing
- The probe recognizes ST 2110-31 flows, checks their AES3 layer (parity, block starts, channel status and its checksum) and recognizes and decodes Dolby E, AC-3 and E-AC-3 pairs in them as in ST 2110-30 flows (sections 23.7 and 15.9)
B.2.6 ST 2110-40: Ancillary Data
- Carries metadata that used to be embedded in SDI VANC/HANC
Examples:
- Closed captions
- Timecode
- Teletext
- HDR metadata
- SCTE markers
Each type becomes an independent IP stream.
B.3 MPTS
C Background Network Traffic
Background network traffic is any data sent or received by a device, application, or service that is not directly initiated by the user's immediate action. It runs "behind the scenes," supporting system functions, apps, and network infrastructure. Understanding it requires separating types, causes, measurement methods, and impacts.
background LAN chatter
Definition and common types
- System-level traffic: OS updates, time synchronization (NTP), DNS queries, certificate revocation/OCSP checks, telemetry and crash reports.
- Application-level background traffic: app updates, push notification keep-alives, cloud sync (photos, documents), mail/calendar fetches, chat presence and message prefetch.
- Infrastructure and protocol traffic: ARP, DHCP, multicast/IPv6 neighbor discovery, routing protocol exchanges (in enterprise networks).
- Third-party/embedded traffic: ad networks, analytics SDKs, CDN prefetches, tracking beacons, and software that checks licensing or telemetry.
- Malicious or unwanted background traffic: botnet activity, data exfiltration, crypto-mining, scanning and lateral movement.
Why it exists
- Maintain connectivity and state (keep-alives, heartbeats).
- Ensure security and integrity (updates, certificate checks).
- Improve user experience (prefetching, caching, sync).
- Provide telemetry for developers, ISPs, and services.
- Support routing and address resolution on local networks.
How to measure and identify it
- Endpoint tools: resource monitors, packet captures (Wireshark/tcpdump), netstat, ss, Windows Resource Monitor.
- Network-wide: flow records (NetFlow/IPFIX), IDS/IPS logs, network performance monitors, DPI appliances.
- Analysis steps:
- Capture baseline traffic volume and common endpoints.
- Map frequent flows to processes/services (lsof, Windows Sysinternals Tcpview).
- Classify by protocol, port, and destination IP/hostname.
- Use DNS and TLS SNI to identify services when payloads are encrypted.
- Correlate with system logs and scheduled tasks to find periodic jobs.
Performance, cost, and security impacts
- Bandwidth consumption: reduces available capacity, affects metered connections and mobile data.
- Latency and jitter: competing flows can increase application latency, especially on congested links.
- Battery and CPU: background syncs and radio wakeups drain mobile device battery.
- Privacy and compliance: telemetry and third-party tracking may leak PII or violate policy.
- Attack surface: background services can provide persistence, command-and-control channels, or leaked credentials.
Practical controls and mitigation
- On devices:
- Disable unnecessary auto-updates, background app refresh, and telemetry where policy permits.
- Use application-level controls: set sync intervals, restrict background data on mobile.
- Harden privacy settings and remove/uninstall apps that generate excessive traffic.
- On networks:
- Implement per-user or per-device QoS and rate limits.
- Use DNS filtering, proxying, and firewall rules to block unwanted endpoints.
- Enforce allowlists for critical services in sensitive environments.
- Monitor flows with NetFlow/IPFIX and alert on deviations from baseline.
- For enterprises:
- Centralize updates and content caching (WSUS, apt/proxy caches, CDN edge).
- Deploy network segmentation and egress filtering to limit lateral movement and data exfiltration.
- Use EDR/IDS to detect anomalous background behavior (e.g., periodic beacons).
When background traffic is normal vs. suspicious
- Normal: periodic system updates, keep-alives to known services, backup/sync windows, low-bandwidth telemetry.
- Suspicious: unexplained high-volume uploads, persistent connections to unknown or flagged domains, regular beacons at odd intervals, encrypted traffic to unusual destinations.
Quick diagnostics checklist
- Identify top talkers by IP/process.
- Check schedule/cron, Windows Task Scheduler, and installed services.
- Resolve domains/WHOIS for unfamiliar endpoints.
- Temporarily block or rate-limit and observe user impact.
- Scan devices for malware and review recent software installs/updates.
Summary
Background network traffic is the unavoidable, often useful stream of non-user-initiated network activity produced by operating systems, applications, and network protocols. It must be measured, classified, and controlled to optimize performance, protect privacy, and reduce security risk. Practical management combines device settings, network controls, centralized update/caching strategies, and continuous monitoring.
D Measuring on different Network Topologies
This section describes the recommended methods and locations for connecting IP-Probe to a network in order to achieve reliable and accurate traffic visibility. The appropriate connection method depends on the network topology, traffic volume, and monitoring objectives.
D.1 Connection Methods Overview
IP-Probe can be connected using one or more of the following methods:
- Network TAP (Test Access Point) - passive, out-of-band monitoring
- Switch SPAN / Mirror Port - switch-based traffic mirroring
- Multicast Join - like a receiver
For continuous, high-accuracy monitoring, especially in broadcast and real-time media networks, a TAP-based connection is strongly recommended.
D.1.1 Using Network TAPs (Recommended for 24/7 Monitoring)
A Network TAP is a passive device installed inline on a network link that provides a copy of all traffic traversing that link. IP-Probe receives an exact replica of the traffic without influencing network behavior.
Explanation: The TAP sits inline between two network devices and provides a perfect copy of traffic to your probe without impacting the production network.
When to use
- 24/7 monitoring
- High-bandwidth links
- Broadcast and multicast traffic
- SMPTE ST 2110 / PTP environments
- Situations requiring loss-free and timing-accurate analysis
Advantages
- Complete visibility of all packets
- No packet loss (within link capacity)
- No impact on network latency or behavior
- Accurate timing and packet order
Typical connection points
- Core or aggregation uplinks
- Leaf-spine interconnects
- PTP Grandmaster distribution paths
- Contribution and distribution ingress/egress links
- Multicast replication points
Notes
- Ensure the probe interface speed matches or exceeds the monitored link speed.
- For fiber TAPs, verify optical power budgets.
- For high-speed links, a packet broker may be used to aggregate and filter traffic before analysis.
D.1.2 Using Switch SPAN / Mirror Ports
A SPAN (Switched Port Analyzer) or mirror port is a switch feature that copies traffic from selected ports, VLANs, or trunks to a designated monitoring port.
If you want to monitor all VLANs passing through a switch, the cleanest method is:
SPAN Source = a TRUNK port (carries many VLANs) SPAN Destination = your monitoring device "IP-Probe"
Because a trunk port already has all VLAN tags, your monitoring tool sees:
- VLAN tags preserved (if your tool supports 802.1Q decoding)
- Traffic from all VLANs the trunk carries
- No need to monitor individual VLANs one by one
Example (Cisco):
monitor session 1 source interface Gi0/1
monitor session 1 destination interface Gi0/10
Where Gi0/1 is the trunk.
^ This is the industry-preferred way if your analyzer can handle tagged frames.
When to use
- Temporary troubleshooting
- Access or edge monitoring
- Low-utilization links
- Environments where TAP installation is not possible
Advantages
- No additional hardware required
- Quick and flexible configuration
Limitations
- Mirrored traffic may be dropped under load
- Timing and packet order may be altered
- Not suitable for loss-sensitive or timing-critical analysis
Recommended SPAN usage
- Mirror trunk ports to preserve VLAN and multicast visibility
- Avoid mirroring multiple high-speed sources to a single destination
- Do not rely on SPAN as the sole data source for SLA or compliance monitoring
D.1.2.1 Remote and Encapsulated Mirroring (RSPAN / ERSPAN)
RSPAN and ERSPAN extend SPAN functionality across multiple switches or routed networks by encapsulating mirrored traffic.
RSPAN
- Encapsulates mirrored traffic in a special VLAN
- Sends it across your network
- Your analyzer connects to that VLAN elsewhere
ERSPAN
- Encapsulates SPAN data in GRE
- Can be routed across Layer-3 networks
When to use
- Analyzer is not physically near the monitored link
- Temporary or distributed monitoring
Considerations
- Encapsulation increases bandwidth usage
- Additional latency and packet loss may occur
- Not recommended for high-bitrate or timing-critical traffic
D.1.3 Join Multicast Groups
What it is
- Analyzer joins the RTP multicast via IGMP
- Receives the same stream as real endpoints
Why it's powerful
- Zero packet capture infrastructure
- No special switch config
- Sees exactly what receivers see
- Scales beautifully
But here's the key nuance
You only see:
- What the multicast delivers
- After network loss, shaping, policing, QoS
You cannot see:
- Packets that were dropped before replication
- Sender-side timing errors
- Other receivers' experience
Best for
- IPTV / broadcast RTP
- Monitoring delivered quality
- "What did the audience actually get?"
This is not packet sniffing - it's service monitoring.
Joining multicast is analysis from the receiver's point of view, not an observation of the network
D.1.4 Recommended Best Practice
For professional environments, a hybrid approach is recommended:
- TAPs for core, production, contribution, and distribution monitoring
- SPAN for edge visibility and ad-hoc troubleshooting
This approach ensures both accuracy and flexibility while minimizing operational risk.
D.1.5 Summary
- Use Network TAPs for continuous, accurate, broadcast-grade monitoring
- Use SPAN ports only where convenience outweighs precision
- Match probe capacity to expected traffic volume
- Always consider timing and packet integrity requirements when selecting a connection method
One Subtle but Important Distinction Method Sees "network truth" Sees "receiver truth"
That's why high-end systems often combine them:
- TAP/SPAN -> detect network impairments
- Multicast join -> validate delivered quality
E MAC usage
E.1 Addresses
E.2 Usage
E.3 Reserved
F IP usage
F.1 Addresses
F.2 Usage
F.3 Reserved
G VLAN usage
G.1 Introduction
VLANs (Virtual Local Area Networks) are used throughout modern Layer-2 and Layer-3 networks to segment traffic logically, regardless of the physical layout.
They appear typically in:
G.1.1 Switches (Layer 2)
VLANs are mainly implemented on managed Ethernet switches.
You create VLANs on switches to separate devices into different broadcast domains (e.g., VLAN 10 for Video, VLAN 20 for Audio).
Common usage on switches:
- Creating isolated networks on the same physical switch
- Trunk ports that carry multiple VLANs between switches
- Access ports that assign endpoints to a single VLAN
G.1.2 Routers and Layer-3 Switches
VLANs require routing to communicate with each other. This happens through:
- Router-on-a-stick (router subinterfaces for each VLAN)
- Layer-3 switches performing inter-VLAN routing
- Firewalls that route or filter between VLANs
G.1.3 Wireless infrastructure
Wi-Fi networks often map each SSID to a VLAN:
- SSID "Guest" -> VLAN 30
- SSID "Corporate"-> VLAN 10
Wireless access points typically use 802.1Q trunking to carry multiple VLANs back to the controller or switch.
G.1.4 Data centers
VLANs are heavily used in:
- Server segmentation (web, app, DB tiers)
- Hypervisors / virtual switches (VMware vSwitch, Hyper-V, KVM)
- Storage networks (iSCSI VLANs)
- Management VLANs for infrastructure
G.1.5 VoIP networks
Most IP phone deployments use separate VLANs:
- Voice VLAN for phones
- Data VLAN for PCs connected via phones
G.1.6 ISP and carrier networks
Service providers use extended VLAN concepts like:
- Q-in-Q (802.1ad)
- Provider Bridges
- Metro Ethernet services
These allow ISPs to isolate customer traffic.
G.1.6.1 Q-in-Q (802.1ad) - "Double VLAN Tagging"
Q-in-Q lets a service provider stack a second VLAN tag onto customer traffic.
Why?
Because customers already use VLANs internally, and service providers also need VLANs. To avoid conflicts, the provider adds its own outer VLAN tag.
How it works:
- Customer frames arrive with: [Customer VLAN tag = C-Tag]
- Provider adds: [Provider VLAN tag = S-Tag]
- Resulting frame has two tags (Q-in-Q).
| S-Tag (service provider VLAN) | C-Tag (customer VLAN) | Ethernet frame... |
Benefit:
The provider can transport many customers' VLANs through the same physical network without interfering with each other.
G.1.7 Security zones / micro-segmentation
Organizations use VLANs to:
- Isolate IoT devices
- Separate guest access
- Restrict sensitive systems
- Limit broadcast domains
Summary
VLANs are used in:
- Switches
- Routers / L3 switches
- Firewalls
- Wireless APs and controllers
- Data centers & virtual switches
- ISP carrier networks
- VoIP deployments
- Security segmentation
VLANs are used anywhere you need traffic separation on a network
G.2 VLAN monitoring
To monitor multiple VLANs using a SPAN port, you must make sure the traffic arrives at the SPAN destination unfiltered.
H FAQ
Why is packet loss 0%, but the video stutters?
Packet loss is not the only cause. Check the packet timing of the flow (chapter 21, Packet Timing, and the timing panels of the ST 2110 pages). Packets that arrive in bursts, in the wrong order or too late make a receiver stutter without any packet being lost.
Does it support encrypted streams?
The probe analyzes the transport of an encrypted stream (bitrate, packet loss, jitter, and for SRT the handshake parameters), but it cannot decode the content. A scrambled MPTS program is marked with a lock and cannot be played (section 18.2).
Can it monitor multicast?
Yes. The probe shows the multicast flows that reach its monitoring port, so make sure the network switch forwards them to that port (see appendix D).
Tip: The Manual button in the manual header opens this manual as one page in a new tab, ready to print or save as PDF.
Shaded: This part should have a shaded background.
Note: The HDMI output supports auto DVI mode over HDMI...
Warning: Don't unplug the device while updating firmware.
Danger: High voltage inside. Disconnect power first.
Tip: Use a short HDMI cable for best signal integrity.
| Field Name | Description (Summary) | Example |
|---|---|---|
| Address | The Ethernet (MAC) address of the endpoint. | 00:1A:2B:3C:4D:5E |
| Alias | A configurable, human-readable name for the endpoint. | Encoder-1-Eth0 |
| Description | Configurable descriptive text providing additional context. | Primary encoder uplink |
| Manufacturer | The hardware vendor associated with the MAC address. | Cisco Systems |
| Usage (Short) | A brief summary of the endpoint's typical role. | STP |
| Usage (Long) | A detailed description of the endpoint's role within the network. | Spanning Tree Protocol |
| Field Name | Description (Summary) | Example |
|---|---|---|
| Address | The Ethernet (MAC) address of the endpoint. | 00:1A:2B:3C:4D:5E |
| Alias | A configurable, human-readable name for the endpoint. | Encoder-1-Eth0 |
| Manufacturer | The hardware vendor associated with the MAC address. | Cisco Systems |
| Usage (Short) | A brief summary of the endpoint's typical role. | STP |
| Usage (Long) | A detailed description of the endpoint's role within the network. | Spanning Tree Protocol |
| Vulnerability | Level | Metric | Deviation | Description | |
|---|---|---|---|---|---|
| 1 | Robustness | major | Fail rate = 1.000 | Adding special chars \r in question can make the model to produce unexpected outputs. | Injecting long sequences of control characters \r in the value of question can alter the model's output significantly, producing unexpected or off-topic outputs. |
| 2 | Robustness | major | Fail rate = 1.000 | Adding special chars \x08 in question can make the model to produce unexpected outputs. | Injecting long sequences of control characters \x08 in the value of question can alter the model's output significantly, producing unexpected or off-topic outputs. |
| Vulnerability | Level | Metric | Deviation | Description | |
|---|---|---|---|---|---|
| 1 | Robustness | major | Fail rate = 1.000 | Adding special chars \r in question can make the model to produce unexpected outputs. | Injecting long sequences of control characters \r in the value of question can alter the model's output significantly, producing unexpected or off-topic outputs. |
| 2 | Robustness | major | Fail rate = 1.000 | Adding special chars \x08 in question can make the model to produce unexpected outputs. | Injecting long sequences of control characters \x08 in the value of question can alter the model's output significantly, producing unexpected or off-topic outputs. |
| Vulnerability | Level | Metric | Deviation | Description | |
|---|---|---|---|---|---|
| 1 | Robustness | major | Fail rate = 1.000 | Adding special chars \r in question can make the model to produce unexpected outputs. | Injecting long sequences of control characters \r in the value of question can alter the model's output significantly, producing unexpected or off-topic outputs. |
| 2 | Robustness | major | Fail rate = 1.000 | Adding special chars \x08 in question can make the model to produce unexpected outputs. | Injecting long sequences of control characters \x08 in the value of question can alter the model's output significantly, producing unexpected or off-topic outputs. |