The lustre-csi-driver should be able to recognize when a mount has already completed, and should return without an error.
Here is the 'kubectl describe' event:
Events:
Type Reason Age From Message
---- ------ ---- ---- -------
Warning FailedMount 3m9s (x44 over 90m) kubelet MountVolume.SetUp failed for volume "lustre4-nnf-dm-system-readwritemany-pv" : rpc error: code = Internal desc = NodePublishVolume - Mount Failed: Error mount failed: exit status 17
Mounting command: mount
Mounting arguments: -t lustre 2056@kfi4:2120@kfi4:2184@kfi4:2248@kfi4:/lustre4 /var/lib/kubelet/pods/c1ab40dc-3864-4938-b66a-7490abd6524c/volumes/kubernetes.io~csi/lustre4-nnf-dm-system-readwritemany-pv/mount
Output: mount.lustre: according to /etc/mtab 2056@kfi4:2120@kfi4:2184@kfi4:2248@kfi4:/lustre4 is already mounted on /var/lib/kubelet/pods/c1ab40dc-3864-4938-b66a-7490abd6524c/volumes/kubernetes.io~csi/lustre4-nnf-dm-system-readwritemany-pv/mount
And the node's mount output:
# mount | grep lustre
2056@kfi4:2120@kfi4:2184@kfi4:2248@kfi4:/lustre4 on /var/lib/kubelet/pods/c1ab40dc-3864-4938-b66a-7490abd6524c/volumes/kubernetes.io~csi/lustre4-nnf-dm-system-readwritemany-pv/mount type lustre (rw,checksum,flock,nouser_xattr,lruresize,lazystatfs,nouser_fid2path,verbose,encrypt)
Here's the pod and node summaries:
$ kubectl get pod -n nnf-dm-system nnf-dm-worker-9bhxv -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
nnf-dm-worker-9bhxv 0/2 ContainerCreating 0 103m <none> elcap317 <none> <none>
$ kubectl get node elcap317
NAME STATUS ROLES AGE VERSION
elcap317 Ready <none> 102m v1.29.3
The lustre-csi-driver should be able to recognize when a mount has already completed, and should return without an error.
Here is the 'kubectl describe' event:
And the node's mount output:
Here's the pod and node summaries:
Here's the CSI driver:
The CSI driver log shows one successful mount followed by many more mount attempts: