"Failed to get user login" error loop, overloading a Synapse sync worker

[verymilan]

Yesterday a sync worker started pretty much continuously running at 100% cpu and the problem was tracked down to the Whatsapp bridge. Turning the Whatsapp bridge off, fixed the issue(!).

It appears somehow a Signal bridge bot ended up in a Whatsapp room. Both bots threw Failed to get user login to handle Matrix event for either bot user in the according log. Due to lack of a better idea, I have opted to purge the room, which has fixed the issue for me, for now. (but some requests are flooding the "in flight" graph in grafana now as the bridge keeps trying to join that room)

For now that's all I know. If this was some sort of conflict with the two bridges, perhaps something could be done to prevent this situation, that's why I have created the issue here, just in case.