Kubernetes CRD Deletion Hang

Posted on | Edited on | In

It is possible that the deletion of k8s CRD hang for indefinite time due to dependency issue.

If it happens, manually edit the CRD and empty the finalizers value block:

1
2
finalizers:
- xxx.finalizer.networking.gke.io

For programmatic way, what I did is running delete kubectl command in background and followed by a kubectl patch command to remove the finalizer block, for example the patch command:

1
2
3
kubectl path <crd type name> <crd resource name> \
{"metadata":{"finalizers":[]}} \
--type merge

The background delete command can be forcely killed at end or after some period of time if it does not fulfill.

0%