Open Half-Shot opened 11 months ago
Hm, this is potentially wrong. We actually respond to /health with an OK as soon as we've done the initial checks, as health is available as soon as the listener has started. I am unsure why the health checks were failing.
For larger deployments, we often need to do quite a bit of work before the bridge is fully "up". Rather than having the liveness probe fail until everything is ready, it's easier to move the probe so that we report ready when the bridge has reached it's dependencies (homeserver, db, redis etc).