question: liveness and readiness probes for config-reloader

[nvictor]

Hello,

I am trying to add liveness and readiness probes to my deployment of config-reloader and have a few questions.

The version of config-reloader I am using (0.4.3-gke.0) doesn't have the ready-url flag. That flag was added in a later version. Does ready-url specify the readiness endpoint for config-reloader?

That brings me to the second question. I don't quite see an endpoint for liveness. What approach would you recommend for implementing liveness for config-reloader?

Thank you

[maxamins]

ready-url specifies if the Prometheus instance the config-reloader is watching is ready. Check https://prometheus.io/docs/prometheus/latest/management_api/ fore more info.

That brings me to the second question. I don't quite see an endpoint for liveness. What approach would you recommend for implementing liveness for config-reloader?

We don't have a ready endpoint for config-reloader. A workaround is you can check and see if /metrics is returning data.

Why are you trying to add liveness and readiness probes?

[nvictor]

Thank you so much for the fast response and for explaining.

To answer your question, config-reloader came up after running a check on our cluster to see deployments without liveness or readiness probes.

I will set both probes to the /metrics endpoint for now. What do you think?

[maxamins]

I would set the probers to the Prometheus readiness/healthy endpoints, since that's what you care about. The /metrics endpoint would just be a hacky workaround.

[nvictor]

Got it. Thanks

Closing this as well.