[MM-62178] Fix signaling after WS reconnect in non-RTCD HA

[streamer45]

Summary

Our new HA-capable e2e tests have surfaced a bug causing signaling issues for sessions reconnecting in a non-RTCD HA deployment.

The optimal path would still be to pull the trigger on https://mattermost.atlassian.net/browse/MM-53376, but we'll have to defer it to the next major version bump (MM v11 / Calls v2).

Ticket Link

https://mattermost.atlassian.net/browse/MM-62178

[streamer45]

🎉

[cpoile]

I guess these are covered under the existing e2e tests? Will this bring down our test/code ratio? :)

I would argue that we don't really need to wait for v2 -- it's a breaking change in that it's removing functionality, but it's not breaking an API or a version compatibility agreement, right? Maybe I'm a little fuzzy on what a "breaking change" really means...

[streamer45]

I guess these are covered under the existing e2e tests? Will this bring down our test/code ratio? :)

Yes, for the time being, they are. If we move to using rtcd in our pipeline (https://mattermost.atlassian.net/browse/MM-62090) then we won't be stressing these paths any longer unless we test both scenarios at the expense of likely doubling execution time (or cost).

I would argue that we don't really need to wait for v2 -- it's a breaking change in that it's removing functionality, but it's not breaking an API or a version compatibility agreement, right? Maybe I'm a little fuzzy on what a "breaking change" really means...

Well, it's totally breaking because if even a single customer is using this today, as soon as they upgrade it will break Calls. And replacing with rtcd is far from being a config setting away.