search

ocf / rt

Request Tracker service
https://rt.ocf.berkeley.edu
Other
11 stars 5 forks source link

Enable healthcheck in kubernetes #13

Closed dkess closed 5 years ago

dkess commented 5 years ago

I recently realized we haven't been using this healthcheck since migrating the service to Kubernetes. This is tested in prod and appears to work.

dkess commented 5 years ago

Actually, looks like this should be a readinessProbe, and we should just not have a liveness probe in the first place.

dkess commented 5 years ago

After doing some more reading, I actually don't think a readiness probe is necessary here. I believe the original intent of the healthcheck script was to restart the service if it fails, which is what liveness checking does here. But I guess it doesn't hurt to specify it as both.

cg505 commented 5 years ago

A readiness probe will avoid bringing down the old RT service before the new one comes up when we deploy a new version. However, it might make sense to have the readiness probe be more simple (like just getting a 200 against /).