Replies: 1 comment 1 reply
-
Unfortunately the fabric8 retry is currently only for network / IOExceptions. 5xx errors would require an enhancement to issue use the retry logic. |
Beta Was this translation helpful? Give feedback.
1 reply
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Once every couple of days we see errors like below in Strimzi log and Operator turns the status of the Kafka resource to "Error", and in the next reconciliation cycle the error is corrected and the system comes back to normal. I suspect the issue is probably with
kube-apiserver
not honoring requests made for the resource ( we have seen resource type differ each time as it could be PVC, StatefulSet, Secret etc). This, in turn, generates some false alerts for us.my question is there a case to be made to be flexible in retrying the resource before flipping the Status to an Error? for example, fabric8 Kube client API provides a property like
kubernetes.request.retry.backoffLimit
to define retry count before status is changed?Beta Was this translation helpful? Give feedback.
All reactions