Multi-Attach error
The pod needs a ReadWriteOnce volume that’s still attached to a different node. Block storage can only be attached to one node at a time.
That is the wording in Kubernetes v1.37. Older releases started the same event with Multi-Attach error for volume "pvc-…", which is the name most people search for. If no pod is using the volume but it is still attached, the message ends Volume is already exclusively attached to one node, waiting on detach before it can be attached to another node instead.
Diagnose in order
Section titled “Diagnose in order”- Where is it attached?
kubectl get volumeattachment -o wide | grep <pv-name>shows the node. - Is the old pod still there? A rolling update keeps the old pod running until the new one is ready, which deadlocks with a single RWO volume.
- Is the old node healthy? After a node crash, the attachment stays until Kubernetes is sure the node is gone.
Usual causes and fixes
Section titled “Usual causes and fixes”| Situation | Fix |
|---|---|
| Deployment rolling update with one RWO volume | Use strategy: type: Recreate, or a StatefulSet |
Old node crashed and is NotReady |
Confirm the node is really down (powered off or fenced), then delete the Node object or apply the node.kubernetes.io/out-of-service taint so Kubernetes force-detaches |
Old pod stuck in Terminating |
See Stuck in Terminating |
Force-detaching a disk from a node that’s actually still running and writing to it can corrupt data. Fence first, then detach.
Read next: Storage and CSI.