This message was deleted.
# longhorn-storagea
adamant-kite-43734
08/07/2023, 5:06 PMThis message was deleted.
q
quiet-area-89381
08/07/2023, 6:04 PMthe application is Z2JH (Zero to Jupyter Hub). Kubernetes is a microk8s distribution. CSI is longhorn.
quiet-area-89381
08/07/2023, 6:06 PMThe error is for a PV backed by longhorn used by the Jupyter single user pod. PV/PVC are assigned to each user and kept when pods are not running. Those pods are automatically terminated after 8h of not being used. The volume should then get detached properly. It is reattached the next time the same user restarts his/her Jupyter server.
quiet-area-89381
08/07/2023, 6:32 PMin the past, this error was related to multipath not being configured properly on the longhorn nodes. But this is fixed.
quiet-area-89381
08/07/2023, 6:33 PMthe crazy part is that it happens a lot with one particular volume of a specific user. She has the multi attachment error at least once a week and often more.
quiet-area-89381
08/07/2023, 6:33 PMIt's being driving me crazy, not finding much in the logs to help me
b
billowy-painting-56466
08/08/2023, 1:05 AMThe error you've encountered originates from a propagation message triggered by this line during an attempt to save the detach error to a VA that no longer exists. The actual detachment error should look like this.
I am not sure if this can be fixed by Longhorn, as it appears to be related to the csi-attacher. I suggest you can create an issue with relevant context, reproducing step and support-bundle so we can analyse the issue further.
q
quiet-area-89381
08/08/2023, 1:10 AMthanks for the answer and the pointers to the source code! I'll dig even more in the logs.
16 Views