Used Claude code to refactor #7929

sadohert · 2025-05-07T16:35:23Z

Sub pages were added. Claude seemed to expand on the documentation for calls configuration, including deployment, metrics monitoring, RTC setup, and troubleshooting.

sadohert · 2025-05-07T17:07:02Z

@streamer45 @cwarnermm - You can ignore this for the moment. I'm just trying to get it to publish a preview environment so i can look at how things lay out from the Claude Code changes. I'm not sure why I can't get this PR to build a preview

cwarnermm · 2025-05-07T17:09:55Z

Ah, I do, @sadohert. It's because it's a fork, not a branch. Only branches (via authorized users) trigger automated previews. For forks, we need to add the preview environment label to kick these off (each time a commit is made). Happy to monitor and help ensure the label is applied as needed so that you can see the pages as intended in production.

github-actions · 2025-05-07T17:12:13Z

Newest code from sadohert has been published to preview environment for Git SHA d0ee389

cwarnermm · 2025-05-07T17:17:38Z

@sadohert - Preview now available.

github-actions · 2025-05-07T21:18:27Z

Newest code from sadohert has been published to preview environment for Git SHA 0a722e5

github-actions · 2025-06-11T14:27:30Z

Newest code from sadohert has been published to preview environment for Git SHA 39abca9

sadohert · 2025-06-11T14:43:19Z

I'm transitioning these back to markdown. My bad. I thought RST was the standard we were moving to.

I'll also resolve the merge conflict.

github-actions · 2025-06-11T19:00:07Z

Newest code from sadohert has been published to preview environment for Git SHA 304864a

github-actions · 2025-06-11T20:55:09Z

Newest code from sadohert has been published to preview environment for Git SHA 52cc483

sadohert · 2025-06-12T13:59:57Z

TODOs:

Remove RTCD pprof configuration
Ensure references to Prometheus setup always have the label
Verify all "Getting Starting" docs within the RTCD and Offloader "Getting STarted" pages are reflected in the docs, then modify the projects to point to the docs (for a single source of truth)
Modify the Calls dashboard to include new constants for Offloader and RTCD OS Metrics ports (default 9100)

github-actions · 2025-06-24T13:49:02Z

Newest code from sadohert has been published to preview environment for Git SHA ad56133

github-actions · 2025-06-27T12:41:57Z

Newest code from sadohert has been published to preview environment for Git SHA 4048a34

github-actions · 2025-07-09T15:11:05Z

Newest code from sadohert has been published to preview environment for Git SHA 90d72d9

streamer45

Thanks for the effort so far! I gave it a first pass and left some comments (quite a few, to be honest 😅 ) for further tweaking and discussion.

streamer45 · 2025-07-09T16:12:52Z

source/configure/calls-deployment.md

@@ -88,15 +103,15 @@ This document provides information on how to successfully make the Calls plugin
 <td>RTC (Calls plugin or <code>rtcd</code>)</td>
 <td>8443</td>
 <td>UDP (incoming)</td>
-<td>Mattermost clients (Web/Desktop/Mobile)</td>
+<td>Mattermost clients (Web/Desktop/Mobile) and calls-offloader</td>


Technically, it's the jobs spawned by the offloader, such as the recorder or transcriber, that connect back on this port. It’s not necessarily the same instance either; for example, in Kubernetes, they could be scheduled on a different node.

Makes sense. Thanks

Will push a change

@streamer45 thoughts on this edit:

ff5b61d

source/configure/calls-deployment.md

streamer45 · 2025-07-09T16:25:30Z

source/configure/calls-deployment.md


-Media (audio/video) is encrypted using security standards as part of WebRTC. It's mainly a combination of DTLS and SRTP. It's not e2e encrypted in the sense that in the current design all media needs to go through Mattermost which acts as a media router and has complete access to it. Media is then encrypted back to the clients so it's secured during transit. In short: only the participant clients and the Mattermost server have access to unencrypted call data.
+**Is calls traffic encrypted?**  
+Yes, using WebRTC security standards (DTLS/SRTP). Traffic is encrypted in transit.


Did we mean to turn the existing paragraph into a single sentence?

streamer45 · 2025-07-09T19:27:19Z

source/configure/calls-troubleshooting.md

+### Browser Console Logs
+
+Instruct users to check their browser console logs:
+
+1. **In Chrome/Edge**:
+   - Press F12 to open Developer Tools
+   - Go to the Console tab
+   - Look for errors related to WebRTC, Calls, or media permissions
+
+2. **Specific patterns to look for**:
+
+   - "getUserMedia" errors (microphone permission issues)
+   - "ICE connection" failures
+   - WebSocket connection errors


I'm not sure how actionable these would be. Overall, it feels more like internal-facing documentation (for support) rather than something meant for customers.

For me this is an admin-facing document. I envision the admin going through early testing (test mode==true) and any helpful client side debug steps would be useful.

Is there anything incorrect in here?

Is there anything incorrect in here?

Not really, but at the same time, these aren't real examples of logs. Would an admin know what a WebSocket connection error refers to? We have at least two WebSocket connections to Mattermost, how would they distinguish between them? What I'm trying to say is that this section could benefit from much more technical investment to become truly valuable (i.e. lower pressure on our support/engineering teams).

streamer45 · 2025-07-09T19:28:42Z

source/configure/calls-troubleshooting.md

+To view real-time logs from calls-offloader containers:
+
+```bash
+# Find and follow logs from all calls-related containers
+docker ps --format "{{.ID}} {{.Image}}" | grep "calls" | awk '{print $1}' | xargs -I {} docker logs -f {}
+```


Similar to a previous comment, using the dedicated config setting seems simpler.

streamer45 · 2025-07-09T19:38:28Z

source/configure/calls-kubernetes.md

+For Kubernetes deployments, you need to ensure:
+
+1. UDP traffic is properly routed to RTCD pods (for media)
+2. TCP traffic can reach both the Mattermost pods and RTCD pods


This reads as slightly ambiguous or overly generic. There needs to be a clear connectivity path between Mattermost and RTCD on the API port (TCP 8045).

streamer45 · 2025-07-09T19:39:50Z

source/configure/calls-kubernetes.md

+Recommended annotations for AWS environments:
+
+```yaml
+service.beta.kubernetes.io/aws-load-balancer-backend-protocol: udp
+service.beta.kubernetes.io/aws-load-balancer-type: nlb
+```


Do these come from official recommendations? Asking because it's the first time I'm seeing them.

These were written by Claude. I don't know if these are valid, and I'm not setup to validate right now. Goign to remove.

streamer45 · 2025-07-09T19:50:59Z

source/configure/calls-offloader-setup.md

+- The calls-offloader service runs in a different network segment than clients
+- Internal DNS resolution differs from external URLs
+- You need to use internal load balancer endpoints for job communication


It's also quite useful if the Mattermost SiteURL is HTTPs, since it avoids the need to deal with client-side certificates, essentially bypassing all of that complexity.

@streamer45 - Can you elaborate on what you're getting at here?

HTTPS is required for Calls afaik... so there isn't really an "if the Mattermost SiteURL is HTTPS" case, is there?

Is the advantage in the ability to have the Offloader jobs simply connect to the non-TLS endpoints, which is the point behind:

avoids the need to deal with client-side certificates

Correct, a secure connection is required for browsers to initiate calls. The key point here is that we can bypass all of that, including client-side certificates.

github-actions · 2025-07-10T19:58:56Z

Newest code from sadohert has been published to preview environment for Git SHA 0f32682

cpoile

First pass, thanks Stu and Claudio

cpoile · 2025-07-09T21:13:19Z

source/configure/calls-rtcd-setup.md

+   ice_port_udp = 8443
+   ice_address_tcp = ""
+   ice_port_tcp = 8443
+   ice_host_override = "YOUR_RTCD_SERVER_PUBLIC_IP"


What's the best way to indicate that this setting should be set only if you're using ice_host_override ? Or if we can't do that and if we're only putting the defaults, then this should be left blank.

Right or wrong, I've mentally come to the conclusion that ice_host_override will be needed much more often than not.

Is that valid?

Well, when using STUN, the override is handled automatically, at least, that was the intention. In practice, though, things are more complex, so explicitly setting the override is the preferred approach in most cases.

Met with @cpoile .

@cwarnermm - Do you know if there's a way we could adapt this part of the docs to pull the active version of the sample config file directly into the docs for the reader?

cpoile · 2025-07-09T21:15:32Z

source/configure/calls-rtcd-setup.md

+   ice_host_override = "YOUR_RTCD_SERVER_PUBLIC_IP"
+
+   # UDP port range for WebRTC connections
+   ice.port_range.min = 9000
+   ice.port_range.max = 10000
+
+   # STUN/TURN server configuration
+   ice_servers = [
+     { urls = ["stun:stun.global.calls.mattermost.com:3478"] }
+   ]


Maybe we should just keep this to be the defaults, and put a big "NOTE" at the start that you should look at the config.sample.toml for in-place documentation for each of the settings?

cpoile · 2025-07-09T21:20:40Z

source/configure/calls-rtcd-setup.md

+     --set ingress.enabled=true \
+     --set ingress.host=rtcd.example.com \
+     --set service.annotations."service\\.beta\\.kubernetes\\.io/aws-load-balancer-backend-protocol"=udp \
+     --set rtcd.ice.hostOverride=rtcd.example.com


Is this correct? (the ice hostOverride=rtcd.example.com bit)

I don't think so because the override would be set through the RTCD_RTC_ICEHOSTOVERRIDE env variable.

cpoile · 2025-07-09T21:26:03Z

source/configure/calls-rtcd-setup.md

+### RTCD Configuration File
+
+The RTCD service uses a TOML configuration file. Here's an example with commonly used settings:
+
+```toml


100% better to just link. High potential to forget to update this.

@streamer45 What do you think about the example toml file above in the systemd installation section?

cpoile · 2025-07-09T21:27:49Z

source/configure/calls-rtcd-setup.md

+
+### Required Mattermost Server Configuration
+
+When using RTCD, you must configure the Mattermost server's CORS settings to allow proper communication between the server and the RTCD service.


Right, probably hallucinated. It does need to be in the offloader setup file though (if it isn't already -- haven't reached it yet)

cpoile · 2025-07-10T21:13:14Z

source/configure/calls-troubleshooting.md

+**Possible causes and solutions**:
+
+1. **Network connectivity issues**:
+   - Verify that UDP port 8443 (or your configured port) is open between clients and RTCD servers


Suggested change

- Verify that UDP port 8443 (or your configured port) is open between clients and RTCD servers

- Verify that UDP port 8443 (or your configured port) is open between clients (web, desktop, mobile) and RTCD servers

Makes sense... should we add calls-offloader to that list of client examples?

To be thorough yes, as they need to connect to RTCD, same as any other client.

cpoile · 2025-07-10T21:20:22Z

source/configure/calls-troubleshooting.md

+
+1. **HTTP API connectivity test**:
+
+   Test if the RTCD API is reachable:


We need to specify that this is from the Mattermost server's machine (each one, in an HA setup).

cpoile · 2025-07-10T21:21:32Z

source/configure/calls-troubleshooting.md

+   curl http://YOUR_RTCD_SERVER:8045/version
+   # Example response: {"buildDate":"2025-04-02 21:33","buildVersion":"v1.1.0","buildHash":"7bc1f7a","goVersion":"go1.23.6","goOS":"linux","goArch":"amd64"}   ```
+
+2. **UDP connectivity test**:


Like Claudio mentioned above, this has to be done when RTCD is not running (no service bound to 8443)

source/configure/calls-troubleshooting.md

Co-authored-by: Christopher Poile <[email protected]>

sadohert · 2025-07-11T15:01:47Z

source/configure/calls-deployment.md

@@ -185,13 +185,25 @@ RTCD is the only officially supported approach for Kubernetes deployments. For d

 ## When to Use RTCD

-The dedicated RTCD service (available with Enterprise license) is recommended for:
+This section will help you understand when and why your organization would want to use the dedicated RTCD service.


@streamer45 - I believe this essentially restores all the previous content on "Why" that we had before?

"Caveat" edits

sadohert · 2025-07-11T15:03:47Z

source/configure/calls-deployment.md

@@ -219,15 +231,16 @@ Mattermost Calls can function in air-gapped environments. Exposing Calls to the
 Calls performance primarily depends on:

 - **CPU resources**: More participants require more processing power
- **Network bandwidth**: Both incoming and outgoing traffic increases with participant count
+- **Network bandwidth**: Both incoming and outgoing traffic increases with participant count. Due to the nature of the service, the bottleneck is always going to be the outgoing/egress path


"Bottleneck"

Edited based on @streamer45 feedback here - #7929 (comment)

sadohert · 2025-07-11T15:05:52Z

source/configure/calls-deployment.md

 - **Active speakers**: Unmuted participants require significantly more resources 
+- **Presenters**: Screen sharing participants require even more resources than active speakers 


Continuing from this comment

How's this extra bullet @streamer45

Looks good!

Co-authored-by: Christopher Poile <[email protected]>

Co-authored-by: Claudio Costa <[email protected]>

sadohert marked this pull request as ready for review May 7, 2025 16:35

sadohert self-assigned this May 7, 2025

sadohert requested a review from cwarnermm May 7, 2025 16:37

sadohert mentioned this pull request May 7, 2025

Stu calls claude 05 01 #7911

Closed

sadohert requested a review from streamer45 May 7, 2025 17:06

cwarnermm added the preview-environment Allow the preview environment to be generated for Pull Requests coming from fork repositories label May 7, 2025

cwarnermm added the Work In Progress Not yet ready for review label May 7, 2025

sadohert added preview-environment Allow the preview environment to be generated for Pull Requests coming from fork repositories and removed preview-environment Allow the preview environment to be generated for Pull Requests coming from fork repositories labels May 7, 2025

sadohert marked this pull request as draft May 13, 2025 13:17

cwarnermm added preview-environment Allow the preview environment to be generated for Pull Requests coming from fork repositories and removed preview-environment Allow the preview environment to be generated for Pull Requests coming from fork repositories labels Jun 11, 2025

sadohert force-pushed the stu-calls-claude-05-07 branch from a1d3830 to 304864a Compare June 11, 2025 18:40

sadohert added preview-environment Allow the preview environment to be generated for Pull Requests coming from fork repositories and removed preview-environment Allow the preview environment to be generated for Pull Requests coming from fork repositories labels Jun 11, 2025

sadohert added preview-environment Allow the preview environment to be generated for Pull Requests coming from fork repositories and removed preview-environment Allow the preview environment to be generated for Pull Requests coming from fork repositories labels Jun 14, 2025

sadohert force-pushed the stu-calls-claude-05-07 branch from 700d283 to ad56133 Compare June 24, 2025 13:45

sadohert added preview-environment Allow the preview environment to be generated for Pull Requests coming from fork repositories and removed preview-environment Allow the preview environment to be generated for Pull Requests coming from fork repositories labels Jun 24, 2025

sadohert added 3 commits June 24, 2025 10:13

Edits from reviews and rebase conflict resolution

5c26d62

Minor tweaks and adjustments from review

aa3202d

Expanding direction for installing RTCD binary

4048a34

sadohert added preview-environment Allow the preview environment to be generated for Pull Requests coming from fork repositories and removed preview-environment Allow the preview environment to be generated for Pull Requests coming from fork repositories labels Jun 27, 2025

Minor edits to config example and section header

90d72d9

sadohert added preview-environment Allow the preview environment to be generated for Pull Requests coming from fork repositories and removed preview-environment Allow the preview environment to be generated for Pull Requests coming from fork repositories labels Jul 9, 2025

cwarnermm added the Guidance label Jul 9, 2025

streamer45 reviewed Jul 9, 2025

View reviewed changes

streamer45 requested a review from cpoile July 9, 2025 19:57

Edits from @streamer45 review

0f32682

sadohert added preview-environment Allow the preview environment to be generated for Pull Requests coming from fork repositories and removed preview-environment Allow the preview environment to be generated for Pull Requests coming from fork repositories labels Jul 10, 2025

cpoile reviewed Jul 10, 2025

View reviewed changes

sadohert and others added 3 commits July 11, 2025 09:40

Update source/configure/calls-troubleshooting.md

a3c879d

Co-authored-by: Christopher Poile <[email protected]>

Update source/configure/calls-troubleshooting.md

6093b67

Co-authored-by: Christopher Poile <[email protected]>

format update

ff5b61d

sadohert commented Jul 11, 2025

View reviewed changes

sadohert and others added 2 commits July 11, 2025 12:24

Update source/configure/calls-offloader-setup.md

8d92056

Co-authored-by: Christopher Poile <[email protected]>

Update source/configure/calls-metrics-monitoring.md

88b0d4c

Co-authored-by: Claudio Costa <[email protected]>


		### Required Mattermost Server Configuration

		When using RTCD, you must configure the Mattermost server's CORS settings to allow proper communication between the server and the RTCD service.

	- Verify that UDP port 8443 (or your configured port) is open between clients and RTCD servers
	- Verify that UDP port 8443 (or your configured port) is open between clients (web, desktop, mobile) and RTCD servers


		1. HTTP API connectivity test:

		Test if the RTCD API is reachable:

		- Active speakers: Unmuted participants require significantly more resources
		- Presenters: Screen sharing participants require even more resources than active speakers

Used Claude code to refactor #7929

Are you sure you want to change the base?

Used Claude code to refactor #7929

Conversation

sadohert commented May 7, 2025

Uh oh!

sadohert commented May 7, 2025

Uh oh!

cwarnermm commented May 7, 2025

Uh oh!

github-actions bot commented May 7, 2025

Uh oh!

cwarnermm commented May 7, 2025

Uh oh!

github-actions bot commented May 7, 2025

Uh oh!

github-actions bot commented Jun 11, 2025

Uh oh!

sadohert commented Jun 11, 2025

Uh oh!

github-actions bot commented Jun 11, 2025

Uh oh!

github-actions bot commented Jun 11, 2025

Uh oh!

sadohert commented Jun 12, 2025

Uh oh!

github-actions bot commented Jun 24, 2025

Uh oh!

github-actions bot commented Jun 27, 2025

Uh oh!

github-actions bot commented Jul 9, 2025

Uh oh!

streamer45 left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

github-actions bot commented Jul 10, 2025

Uh oh!

cpoile left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment