Gummikavalier

We are preparing to upgrade MongoDB from 6.0 to 7.0 this evening to see if it has any effect. If not, we'll use the parameter `DISABLE_DB_WATCHERS=false` to enable oplog support...

I can confirm that issues continued in under 24 hours from switching from MongoDB 6.0 to 7.0. At the beginning everything seemed to work ok, but after the initial rush...

> > Also we have a restart process that restarts all 4 docker instances every night. > > [@Gummikavalier](https://github.com/Gummikavalier) does this process restarts all of them at once or is...

So far I'm uncertain of the results. We have some reports of messages not being delivered or to be exact, when you refresh, you suddenly get more channels in unread...

@sampaiodiego I can now confirm that issue remains even when using sticky (IP based) sessions and slow staggered start of the instances in the nightly restart. We switched back to...

I don't dare to try out 7.7.1 without using `DISABLE_DB_WATCHERS=false` in our production. We are a scientific organization (CSC - IT Center for Science Ltd.) which means our users are...

Ok, I can reproduce the issue (also in RC 7.7.1) pretty consistently by doing following: EDIT: Removed the need for multi instance setup and sticky sessions to reproduce this issue....

@pierre-lehnen-rc I think that slowdown issue was a separate and may have been part of that race condition thing sampaiodiego mentioned with several instances starting at the same time. I...

@pierre-lehnen-rc Ok, test done. I'm using Firefox. 1. On Bob's view, after the issue has been triggered (required only 3 - 4 messages this time). Highlighting stops working. I can...

Regarding the amount of websocket connections; I did get two of them after few reloads couple of times, but the triggering factor was always the reload itself. The highlighting /...