[BUG] File system freeze impossible about longhorn HOT 3 OPEN

1. It's about vm snapshots, for backup.
2. It's running in my personal lab, on a hyper-v host. Since azure is built on the same tech, that kernel enables the communication with the hypervisor.

Unfortunately, we don't have too much experience with hyper-V instances.

Jan 13 18:54:36 docker1 Hyper-V VSS: FREEZE of /var/lib/kubelet/plugins/kubernetes.io/csi/driver.longhorn.io/c975b852d71be89749b5433946a5956e799e8eeb795f1f22c51908c65d44a506/globalmount failed; error:5 Input/output error

The mount point might be corrupted, it's hard to tell what happened. Have you tried to restart the pod to trigger another volume attachment? Or has Longhorn gone through the process if the setting Automatically Delete Workload Pod when The Volume Is Detached Unexpectedly setting is enabled? Under normal circumstances, if the global mount point is corrupted, Longhorn should perform a clean up and remount.

Could you try accessing the UI and provide the support-bundle? The support-bundle might be able to provide some insights.

from longhorn.

Hi, to investigate the cause, would you help clarify the below issues?

1. Which snapshot do you mean? Longhorn volume snapshot or VM snapshot? Or other kinds of snapshot like ETCD/cluster snapshot
2. Notice that you are using azure-kernel Are the Windows servers Azure machines? Or they are in the personal lab?

from longhorn.

Hello @shuo-wu!

1. It's about vm snapshots, for backup.
2. It's running in my personal lab, on a hyper-v host. Since azure is built on the same tech, that kernel enables the communication with the hypervisor.

from longhorn.