$ oc describe pod -n argos prometheus-argos-1 | tail -5
Events:
  Type     Reason              Age   From                     Message
  ----     ------              ----  ----                     -------
  Normal   Scheduled           85m   default-scheduler        Successfully assigned argos/prometheus-argos-1 to xx-xx-xx-w6n7r-worker-hb27h
  Warning  FailedAttachVolume  85m   attachdetach-controller  Multi-Attach error for volume "pvc-230bbeda-df2e-48d4-8b99-dc9d9bc9e3c1" Volume is already exclusively attached to one node and can't be attached to another

however
pvc-230bbeda-df2e-48d4-8b99-dc9d9bc9e3c1
is unattached.

We find that there are 2 volume attachment resources pointing at the same node, same lun:
$ oc get volumeattachment csi-9826d6cbe313fe82f2a8bc00be1bb25aa3167319eded23802cce46f6c42e5d06 csi-8d9cc02b65e4800f6973f514f6c0cad32f893697c07173331fb0c7da62e9094c -o yaml
apiVersion: v1
items:
- apiVersion: storage.k8s.io/v1
  kind: VolumeAttachment
  metadata:
    annotations:
      csi.alpha.kubernetes.io/node-id: xx-xx-xx-worker-e16as-westeurope1-qqnfm
    creationTimestamp: "2025-11-05T23:30:10Z"
    finalizers:
    - external-attacher/disk-csi-azure-com
    name: csi-9826d6cbe313fe82f2a8bc00be1bb25aa3167319eded23802cce46f6c42e5d06
    resourceVersion: "354371758"
    uid: 11e93a17-8873-4b68-9168-83ffa25dbe29
  spec:
    attacher: disk.csi.azure.com
    nodeName: xx-xx-xx-worker-e16as-westeurope1-qqnfm
    source:
      persistentVolumeName: pvc-94aa38d1-f83e-48f2-80e8-6c03839c756c
  status:
    attached: true
    attachmentMetadata:
      LUN: "0"
- apiVersion: storage.k8s.io/v1
  kind: VolumeAttachment
  metadata:
    annotations:
      csi.alpha.kubernetes.io/node-id: xx-xx-xx-worker-e16as-westeurope1-qqnfm
    creationTimestamp: "2025-11-05T23:02:38Z"
    finalizers:
    - external-attacher/disk-csi-azure-com
    name: csi-8d9cc02b65e4800f6973f514f6c0cad32f893697c07173331fb0c7da62e9094c
    resourceVersion: "354327328"
    uid: 4b1ce05a-2a39-4502-9385-e305530533ad
  spec:
    attacher: disk.csi.azure.com
    nodeName: tst-we-int08a-w6n7r-worker-e16as-westeurope1-qqnfm
    source:
      persistentVolumeName: pvc-230bbeda-df2e-48d4-8b99-dc9d9bc9e3c1
  status:
    attached: true
    attachmentMetadata:
      LUN: "0"
kind: List
metadata:
  resourceVersion: ""
    

OCP 4.18.24
    

Reccurence but not predictable
    

    1. Install OCP 4.18 IPI cluster on Azure with azuredisk CSI driver
    2. Create a pod with persistent volume claim using azuredisk
    3. Reboot the node where the pod is scheduled
    4. Try to start the pod again
    

    Pod nay fail to start with Multi-Attach error, even though the volume is not attached to any node.
    

    Pod should start successfully after node reboot.