digitalocean
Kubernetes.DeleteDangerous() does not delete PVC Volumes
According to the doctl kubernetes cluster delete docs, the --dangerous flag should delete PVC volumes
However, godo.Client.Kubernetes.DeleteDangerous() does not delete PVC Volumes
Using the latest version: v1.113.0
Hey @regnaio, thanks for the report. I've reached out to our Kubernetes team for more details, and will report back when I hear back from them.
hi @regnaio 👋 I work on the DOKS (managed Kubernetes) team at DO.
I just verified that the flag / godo function you mentioned is working as expected: a cluster which had a PVC got cleaned up properly, and the backing DO block storage volume was gone at the end when --dangerous was specified.
In case you were expecting to find code in godo that explicitly looks for PVCs/PVs in your cluster to delete them: that is actually not how the functionality is implemented. Instead, our backend keeps track of all managed resources and makes sure they get deleted when a cluster is (dangerously) deleted. I'm not sure if interactions against the API server is what you were referring when you quoted the godo function, but I figured I should add clarity here.
In case you do find that your cluster(s) aren't getting their volumes cleaned up when using the flag in question, could I ask you to file a DO support ticket? As mentioned the functionality seems to be working correctly per se, so any remaining issues may require taking a deeper look. Our support team is happy to help you on that should it be needed.
Thanks in advance!
I can try this again the next time I spin up a DigitalOcean k8s cluster
However, I know for sure that I had XFS PVC's bound to Pods in the cluster (immediately before cluster deletion), so they should have been considered part of the cluster. godo.Client.Kubernetes.DeleteDangerous() did not delete the XFS PVC's