Unable to delete pvc volume

October 25, 2018 2.3k views
Kubernetes Block Storage CoreOS

I've been testing/learning Kubernetes using the DO managed Kubernetes cluster and am having some issues with some persistent volumes.

I was testing a deployment of a MSSQL Server Express service but needed more RAM to run the container. I went to the settings on the cluster and deleted the 1GB/1vCPU pool and created a new pool with more memory.

Once I was done testing I resized the node pool again back to a single node 1GB/1vCPU pool.

A few days later I noticed that there were two pvc volumnes listed in the DO control panel but I am unable to delete, attach, or detach them. I deleted the pvc and pv resources using kubectl but they still remain.

If I try to detach one of them it just spins and eventually does nothing.

If I try to attach one of them I get an error toast notification that says "Attachment not found".

I ran kubectl logs -f csi-do-controller-0 csi-attacher --namespace=kube-system and it keeps looping the following errors:

I1025 00:51:35.671304       1 reflector.go:286] github.com/kubernetes-csi/external-attacher/vendor/k8s.io/client-go/informers/factory.go:87: forcing resync
I1025 00:51:35.671387       1 controller.go:167] Started VA processing "csi-4ee97f129eafa27066c9e111ea25bd0d40dfacbd126e4f20e4eaf400d91b256f"
I1025 00:51:35.671400       1 csi_handler.go:76] CSIHandler: processing VA "csi-4ee97f129eafa27066c9e111ea25bd0d40dfacbd126e4f20e4eaf400d91b256f"
I1025 00:51:35.671406       1 csi_handler.go:127] Starting detach operation for "csi-4ee97f129eafa27066c9e111ea25bd0d40dfacbd126e4f20e4eaf400d91b256f"
I1025 00:51:35.671472       1 csi_handler.go:134] Detaching "csi-4ee97f129eafa27066c9e111ea25bd0d40dfacbd126e4f20e4eaf400d91b256f"
I1025 00:51:35.671485       1 csi_handler.go:335] Saving detach error to "csi-4ee97f129eafa27066c9e111ea25bd0d40dfacbd126e4f20e4eaf400d91b256f"
I1025 00:51:35.680875       1 csi_handler.go:345] Saved detach error to "csi-4ee97f129eafa27066c9e111ea25bd0d40dfacbd126e4f20e4eaf400d91b256f"
I1025 00:51:35.680894       1 csi_handler.go:86] Error processing "csi-4ee97f129eafa27066c9e111ea25bd0d40dfacbd126e4f20e4eaf400d91b256f": failed to detach: node "compassionate-hertz-e8z" not found
I1025 00:51:35.680919       1 controller.go:167] Started VA processing "csi-4ee97f129eafa27066c9e111ea25bd0d40dfacbd126e4f20e4eaf400d91b256f"
I1025 00:51:35.680924       1 csi_handler.go:76] CSIHandler: processing VA "csi-4ee97f129eafa27066c9e111ea25bd0d40dfacbd126e4f20e4eaf400d91b256f"
I1025 00:51:35.680928       1 csi_handler.go:127] Starting detach operation for "csi-4ee97f129eafa27066c9e111ea25bd0d40dfacbd126e4f20e4eaf400d91b256f"
I1025 00:51:35.680956       1 csi_handler.go:134] Detaching "csi-4ee97f129eafa27066c9e111ea25bd0d40dfacbd126e4f20e4eaf400d91b256f"
I1025 00:51:35.680964       1 csi_handler.go:335] Saving detach error to "csi-4ee97f129eafa27066c9e111ea25bd0d40dfacbd126e4f20e4eaf400d91b256f"
I1025 00:51:35.684097       1 csi_handler.go:345] Saved detach error to "csi-4ee97f129eafa27066c9e111ea25bd0d40dfacbd126e4f20e4eaf400d91b256f"
I1025 00:51:35.684113       1 csi_handler.go:86] Error processing "csi-4ee97f129eafa27066c9e111ea25bd0d40dfacbd126e4f20e4eaf400d91b256f": failed to detach: node "compassionate-hertz-e8z" not found
I1025 00:51:36.111862       1 reflector.go:286] github.com/kubernetes-csi/external-attacher/vendor/k8s.io/client-go/informers/factory.go:87: forcing resync
I1025 00:51:36.111932       1 controller.go:197] Started PV processing "pvc-37f8f87ed59b11e8"
I1025 00:51:36.111941       1 csi_handler.go:350] CSIHandler: processing PV "pvc-37f8f87ed59b11e8"
I1025 00:51:36.111977       1 csi_handler.go:386] CSIHandler: processing PV "pvc-37f8f87ed59b11e8": VA "csi-4ee97f129eafa27066c9e111ea25bd0d40dfacbd126e4f20e4eaf400d91b256f" found
I1025 00:51:37.012480       1 controller.go:197] Started PV processing "pvc-37f8f87ed59b11e8"
I1025 00:51:37.012497       1 csi_handler.go:350] CSIHandler: processing PV "pvc-37f8f87ed59b11e8"
I1025 00:51:37.012532       1 csi_handler.go:386] CSIHandler: processing PV "pvc-37f8f87ed59b11e8": VA "csi-4ee97f129eafa27066c9e111ea25bd0d40dfacbd126e4f20e4eaf400d91b256f" found
I1025 00:51:37.391372       1 reflector.go:428] github.com/kubernetes-csi/external-attacher/vendor/k8s.io/client-go/informers/factory.go:87: Watch close - *v1beta1.VolumeAttachment total 39 items received
I1025 00:51:45.671543       1 reflector.go:286] github.com/kubernetes-csi/external-attacher/vendor/k8s.io/client-go/informers/factory.go:87: forcing resync
I1025 00:51:45.671632       1 controller.go:167] Started VA processing "csi-4ee97f129eafa27066c9e111ea25bd0d40dfacbd126e4f20e4eaf400d91b256f"
I1025 00:51:45.671646       1 csi_handler.go:76] CSIHandler: processing VA "csi-4ee97f129eafa27066c9e111ea25bd0d40dfacbd126e4f20e4eaf400d91b256f"
I1025 00:51:45.671654       1 csi_handler.go:127] Starting detach operation for "csi-4ee97f129eafa27066c9e111ea25bd0d40dfacbd126e4f20e4eaf400d91b256f"
I1025 00:51:45.671715       1 csi_handler.go:134] Detaching "csi-4ee97f129eafa27066c9e111ea25bd0d40dfacbd126e4f20e4eaf400d91b256f"
I1025 00:51:45.671731       1 csi_handler.go:335] Saving detach error to "csi-4ee97f129eafa27066c9e111ea25bd0d40dfacbd126e4f20e4eaf400d91b256f"
I1025 00:51:45.679013       1 csi_handler.go:345] Saved detach error to "csi-4ee97f129eafa27066c9e111ea25bd0d40dfacbd126e4f20e4eaf400d91b256f"
I1025 00:51:45.679038       1 csi_handler.go:86] Error processing "csi-4ee97f129eafa27066c9e111ea25bd0d40dfacbd126e4f20e4eaf400d91b256f": failed to detach: node "compassionate-hertz-e8z" not found
I1025 00:51:45.679076       1 controller.go:167] Started VA processing "csi-4ee97f129eafa27066c9e111ea25bd0d40dfacbd126e4f20e4eaf400d91b256f"
I1025 00:51:45.679086       1 csi_handler.go:76] CSIHandler: processing VA "csi-4ee97f129eafa27066c9e111ea25bd0d40dfacbd126e4f20e4eaf400d91b256f"
I1025 00:51:45.679094       1 csi_handler.go:127] Starting detach operation for "csi-4ee97f129eafa27066c9e111ea25bd0d40dfacbd126e4f20e4eaf400d91b256f"
I1025 00:51:45.679133       1 csi_handler.go:134] Detaching "csi-4ee97f129eafa27066c9e111ea25bd0d40dfacbd126e4f20e4eaf400d91b256f"
I1025 00:51:45.679146       1 csi_handler.go:335] Saving detach error to "csi-4ee97f129eafa27066c9e111ea25bd0d40dfacbd126e4f20e4eaf400d91b256f"
I1025 00:51:45.772287       1 csi_handler.go:345] Saved detach error to "csi-4ee97f129eafa27066c9e111ea25bd0d40dfacbd126e4f20e4eaf400d91b256f"
I1025 00:51:45.772335       1 csi_handler.go:86] Error processing "csi-4ee97f129eafa27066c9e111ea25bd0d40dfacbd126e4f20e4eaf400d91b256f": failed to detach: node "compassionate-hertz-e8z" not found
I1025 00:51:46.112052       1 reflector.go:286] github.com/kubernetes-csi/external-attacher/vendor/k8s.io/client-go/informers/factory.go:87: forcing resync
I1025 00:51:46.112117       1 controller.go:197] Started PV processing "pvc-37f8f87ed59b11e8"
I1025 00:51:46.112125       1 csi_handler.go:350] CSIHandler: processing PV "pvc-37f8f87ed59b11e8"
I1025 00:51:46.112164       1 csi_handler.go:386] CSIHandler: processing PV "pvc-37f8f87ed59b11e8": VA "csi-4ee97f129eafa27066c9e111ea25bd0d40dfacbd126e4f20e4eaf400d91b256f" found
I1025 00:51:49.393802       1 reflector.go:428] github.com/kubernetes-csi/external-attacher/vendor/k8s.io/client-go/informers/factory.go:87: Watch close - *v1.PersistentVolume total 37 items received

compassionate-hertz-e8z was a node from the old node pool that was deleted which may or may not be related. Strangely pvc-37f8f87ed59b11e8 isn't even listed in my account under volumes, the two that were successfully deleted using kubectl delete are.

The only thing I haven't tried yet is deleting the cluster, I'm just using it for testing/learning purposes so that won't be a problem if I need to do that.

1 comment
  • Having the same problem, and apparently manually patching the finalizers solves it.
    However, that's kind of a manual solution and not very scalable.
    Sometimes, this kind of errors just prevent a statefulset from launching pods and I have to manually fix for ever case.

    What is the root cause of the problem? and is there any permanent solution for it?

2 Answers

Who is having this issue:
Patch the finalizers, e.g

kubectl patch pv <pvc-ID> -p '{"metadata":{"finalizers":null}}'
kubectl patch pvc <pv-ID> -p '{"metadata":{"finalizers":null}}'

after that it's possible to delete them, the same workflow is applied for the pods.

I have the same issue. If I run
kubectl get pvc
I have a list of PV having Reclaim set to Delete and Status to Released.

kubectl delete pvc <ID>
It says it is deleted, but the command is not terminated.

Thanks.

Have another answer? Share your knowledge.