opentelemetry-helm-charts icon indicating copy to clipboard operation
opentelemetry-helm-charts copied to clipboard

Support configuring and disabling healthcheck extension

Open jsirianni opened this issue 1 year ago • 7 comments

The healthcheck probes are configured like this, without customization options.

    livenessProbe:
      httpGet:
        path: /
        port: 13133
    readinessProbe:
      httpGet:
        path: /
        port: 13133

I would like the following options

  • ability to change path, port
  • ability to set other options like periodSeconds
  • ability to disable healthcheck extension all together.

While I agree that generally healthcheck extension + healthcheck probes should always be in use, I do think this should be up to the user. I think defaulting to the current configuration (enabled with sane defaults) is great.

I would be willing to take on this work if the community agrees.

jsirianni avatar Mar 22 '23 17:03 jsirianni

@jsirianni this topic has come up before, here is the related issue for context: https://github.com/open-telemetry/opentelemetry-helm-charts/issues/242

TylerHelmuth avatar Mar 22 '23 17:03 TylerHelmuth

Do you have a use case in mind that requires more configurability to the readiness and liveness sections?

TylerHelmuth avatar Mar 22 '23 17:03 TylerHelmuth

Thanks for linking that. I am not sure I agree with the statement that health checks are mandatory in k8s, but I do agree that they are highly recommended. My argument has already been discussed there by the OP, so I can concede. It does feel "forced" but I understand the reasoning.

I do not have a concrete usecase, but I can imagine wanting to dial in probe settings for a service that might be scaling up and down frequently. To me, this was less important than having the ability to disable the checks.

If the stanza to absolutely always require the healthcheck extension, feel free to close this issue.

jsirianni avatar Mar 22 '23 17:03 jsirianni

I'd like to see the ability to extend the health check timeout as if you have a big configuration with a lot of processors it can take a while for the collector to report back healthy causing issues causing restarts again.

lewis-prior-thrivent avatar Apr 13 '23 13:04 lewis-prior-thrivent

Being able to configure these options sounds like a good idea.

TylerHelmuth avatar Apr 13 '23 14:04 TylerHelmuth

Being able to configure ports is also acceptable.

TylerHelmuth avatar May 02 '23 18:05 TylerHelmuth

Hi, would it be possible to also add option to disable readiness / liveness probe? At times I need to debug the otel collector using the delve debugger and I usually ran into the pod I am attached to getting killed by the liveness probe (when you pause the process, the liveness probe is not responding), forcing me to copy the helm chart and remove the liveness probe.

jmichalek132 avatar Jun 01 '23 09:06 jmichalek132