search

SparebankenVest / public-helm-charts

Public Helm charts provided by Sparebanken Vest
https://charts.spvapi.no
28 stars 51 forks source link

Improve probes for akv2k8s #97

Closed slushysnowman closed 12 months ago

slushysnowman commented 1 year ago

In akv2k8s:

It would be good if these both had both sorts of probes so that:

KenADev commented 1 year ago

The probes are apparently dependent of the prometheus metrics endpoints, causing the controller probes to fail when global.metrics.enabled is set to false, which is currently the default in values.yaml

$ kubectl describe pod/akv2k8s-controller-77dbc456bd-k5n5b
Events:
  Type     Reason     Age                  From               Message
  ----     ------     ----                 ----               -------
  Normal   Scheduled  118s                 default-scheduler  Successfully assigned akv2k8s/akv2k8s-controller-77dbc456bd-k5n5b to aks-nodepool1-60850117-vmss000000
  Normal   Pulled     88s (x2 over 118s)   kubelet            Container image "spvest/azure-keyvault-controller:1.4.0" already present on machine
  Normal   Created    88s (x2 over 118s)   kubelet            Created container controller
  Normal   Started    88s (x2 over 118s)   kubelet            Started container controller
  Normal   Killing    88s                  kubelet            Container controller failed liveness probe, will be restarted
  Warning  Unhealthy  68s (x12 over 117s)  kubelet            Readiness probe failed: Get "http://10.244.0.29:9000/healthz": dial tcp 10.244.0.29:9000: connect: connection refused
  Warning  Unhealthy  58s (x6 over 108s)   kubelet            Liveness probe failed: Get "http://10.244.0.29:9000/healthz": dial tcp 10.244.0.29:9000: connect: connection refused

Edit: See also https://github.com/SparebankenVest/public-helm-charts/issues/105

181192 commented 12 months ago

Resolved by SparebankenVest/azure-key-vault-to-kubernetes#562