Unsolved
10 Posts
0
2123
csi-unity crashloop due to "fc_host: no such file or directory"
I'm trying to configure OpenShift 4 to work with a Dell Unity which is available via iSCSI and NFS. I've deployed the Dell CSI Unity operator using the OperatorHub and following these instructions: https://dell.github.io/storage-plugin-docs/docs/partners/redhat/
The node daemonset always crashes:
[jgazeley@hephaestos okd4_cluster_builder]$ oc get po
NAME READY STATUS RESTARTS AGE
dell-csi-operator-controller-manager-945c46857-xrwzp 1/1 Running 6 13d
unity-controller-5b695b6467-7pjgc 5/5 Running 16 3d1h
unity-controller-5b695b6467-wrrrr 5/5 Running 21 3d1h
unity-node-hcz8c 1/2 CrashLoopBackOff 857 3d1h
unity-node-hndhv 1/2 CrashLoopBackOff 856 3d1h
unity-node-jd6bp 1/2 CrashLoopBackOff 856 3d1h
unity-node-ktkz8 1/2 CrashLoopBackOff 857 3d1h
unity-node-kz49x 1/2 CrashLoopBackOff 856 3d1h
The crash is due to:
time="2021-04-21T08:39:09Z" level=error arrayid=ckm00201700161 runid=node-2231 msg="Cannot read directory: /sys/class/fc_host Error: open /sys/class/fc_host: no such file or directory" func="github.com/dell/csi-unity/service/utils.GetFCInitiators()" file="/go/src/csi-unity/service/utils/emcutils.go:90"
Even though I have disable FibreChannel. Can anyone help? Thanks
Flo_csI
2 Intern
2 Intern
•
166 Posts
0
April 27th, 2021 02:00
Hi @jgazeley,
This is a "normal" error, I requested to reclassify it as a warning.
Do you have more logs on the crashing container ?
Can you make sure you have the csiDebug set to true : https://github.com/dell/csi-unity/blob/master/helm/csi-unity/values.yaml#L16
Thanks.
jgazeley
10 Posts
0
April 27th, 2021 03:00
Hi @Flo_csI
Thanks for your reply. I set X_CSI_UNITY_DEBUG to true in OperatorHub and redeployed. This is the config I'm running now, which is 99% standard except I removed the storageClass for FibreChannel:
So my node pods all got redeployed and immediately started crashlooping:
Here are the logs from one of the pods:
Thanks
jgazeley
10 Posts
0
April 27th, 2021 06:00
bmcfeeters
72 Posts
0
April 27th, 2021 12:00
Hi @jgazeley , yes, your post is now visible
One thing that I noticed in the logs was this line...
Error: [rpc error: code = Internal desc = runid= Adding iSCSI initiator error: Initiator found (iqn.1994-05.com.redhat:bfbbd4d9d631), and attached to someother host (Host_7) instead of host: Host_12]"
I hit a similar issue once when I cloned a Linux VM and had to reset the iSCSI IQN to a new value before the driver would be able to register it properly with the array.
Good luck
Bryan
jgazeley
10 Posts
0
April 28th, 2021 02:00
Thanks Bryan, I'll try this. Does each node need a unique IQN, or is it one IQN per cluster?
jgazeley
10 Posts
0
April 28th, 2021 04:00
jgazeley
10 Posts
0
April 28th, 2021 06:00