feat: user presence supports k8s

[mattkrick]

Description

fix #7204

User presence is currently reset during a postDeploy step. This won't work when we move to k8s and have multiple servers deploying at different times.

This PR will wipe out user presence for all users connected to a given socket server during shutdown. In the event of an ungraceful shutdown, when a server starts up it performs a health check on all other servers. It then removes any user presence for servers that were not present during the health check.

Testing scenarios

• [ ] tbd

[sonarqubecloud[bot]]

SonarCloud Quality Gate failed.   

1 Bug
0 Vulnerabilities
0 Security Hotspots
0 Code Smells

No Coverage information
0.0% Duplication

[mattkrick]

@Dschoordsch you're too kind! thanks for the review. I'm stuck on testing that SIGINT handler. i could call kill -s SIGINT, but i'd like to just ctrl+c, but that's not triggering it! I'm not sure what is swallowing the signal...

[mattkrick]

i misunderstood the SIGINT handler, it doesn't handle promises. instead of doing async work in the SIGINT handler, I'll have the dying node publish a message to all living nodes asking them to kill it in redis :grimacing:

[mattkrick]

@Dschoordsch would you mind giving this old thing a fresh review?

[mattkrick]

I think this is good to go, we're got a lot of risky PRs going on right now, so I'll hold off on merging this until things calm down in a week or so.