Details
-
Task
-
Resolution: Fixed
-
Major
-
None
-
None
Description
When an 'ephemeral' pod fails we log the failed pod and its events - https://github.com/couchbase/couchbase-operator/blob/1.2.x/pkg/util/k8sutil/k8sutil.go#L436-L473.
However, we only seem to do this when the pod creation fails.
Instead, it would be good to capture this information for any pod that we have detected has failed (has been failed over by Couchbase or is down), even if it has PVs and will be re-attached.
This means that we will more easily be able to debug issues where Kubernetes is terminating a pod externally (for example, due to resource issues or draining) as these events will be captured in the logs.