consul-k8s
consul-k8s copied to clipboard
hashicorp
helm:Configurability of readiness probes
It would be very useful if it was possible to configure the parameters of the readiness and liveness probes implemented by the chart. At times of high load, I frequently find Consul servers and even occasionally agents being terminated due to failures of readiness probes for their pods, and so far as I can tell, there is no actual underlying problem that would justify this termination - merely a slow response due to load.
Additionally, when using a GKE Ingress, it has to model the health checks from the readiness probe, and since it is using an "exec" instead of httpGet, they can't model the healthchecks automatically. Is there a specific reason that they health checks are not using httpGet? Maybe back in the olden days it didn't support https? 🤷
In my opinion, the liveness (readiness/startup) probes should all be in values.yaml so they can be easily customized instead of being "hard coded" in the template files.
REF:
- https://cloud.google.com/kubernetes-engine/docs/concepts/ingress#health_checks
- https://kubernetes.io/docs/tasks/configure-pod-container/configure-liveness-readiness-startup-probes/
@cerebrate or @TJM Are the issues still occurring with the latest Consul 1.12 or 1.13.x releases? If so, do you have a PR that we could review that could perhaps address the sort of issues you are seeing during Consul Server startup?
I am no longer experiencing the issue, but unfortunately I can't tell you if it would otherwise have gone away with the latest releases since, for me, it went away with a hardware upgrade. Sorry I don't have more useful information.
Closing as no feedback was given based on current releases of Consul K8s and Consul. Please open a new issue if more attention is needed.