BlobCSI Driver Sync Fails Post Server/VM Reboot | Requires PV/PVC Recreation with Unique Volume_Handle

[Hidayathullashaik]

Hi Team,

I am referencing the issue logged earlier: Blobfuse CSI Configurations Issue #1620.

[Problem Statement]
I have ensured that the volume_handle name is unique across Kubernetes clusters. However, after a planned or sudden reboot of a node/server, the Blobfuse CSI driver stops syncing to the Azure Storage Account. This forces me to delete the existing Persistent Volume (PV) and Persistent Volume Claim (PVC) and recreate them with a new volume_handle in the Kubernetes PV YAML file.

[Expected Behavior]
The Blobfuse CSI driver should automatically remount to the Azure Storage Account using the Blob configurations and storage account secrets whenever the server reboots.

Please advise on resolving this issue or if there are additional configurations needed to ensure the remount process works seamlessly post-reboot.

[andyzhangx]

@Hidayathullashaik before you reboot the VM, you should reschedule the pods with blobfuse volume to other nodes, otherwise the mounted blobfuse volume path could turn to be a local disk path, that's a known k8s kubelet bug, not sure whether they have solved in latest k8s version.