We've been seeing a lot of readiness/liveness probe failures in our k8s setup, but every other indication is that our app is functioning healthily so let's update these timings to something more liberal since we know it can take some number of seconds for the canary to start and run migrations, import features, etc
Let's consider moving migraitons and feature config loading into a separate kubernetes 'job' so that the request handler pods are responsible only for collectstatic and starting gunicorn.
We've been seeing a lot of readiness/liveness probe failures in our k8s setup, but every other indication is that our app is functioning healthily so let's update these timings to something more liberal since we know it can take some number of seconds for the canary to start and run migrations, import features, etc
https://github.com/mozilla-services/cloudops-infra/blob/master/projects/experimenter/k8s/charts/experimenter/templates/deployment.yaml#L64-L85
┆Issue is synchronized with this Jira Task