[Bug]: restart kadalu-csi-nodeplugin daemonset would make current gfs mount unavailable. about kadalu HOT 2 CLOSED

Yes, it is a knowm problem w/ fuse drivers implementing CSI. First, once the mount is established deamonset going down is rare, in extreme cases or user initiated restart we need to scale down & up business pod.

Another case when this is hit is during upgrades, closer solution exists but it's a bit invovled to implement it.

from kadalu.

Yes, it is a known problem w/ fuse drivers implementing CSI. First, once the mount is established deamonset going down is rare, in extreme cases or user initiated restart we need to scale down & up business pod.

Another case when this is hit is during upgrades, closer solution exists but it's a bit invovled to implement it.

Thanks a lot for the explanation. I've noticed these in the meantime. After a cursory look at Azure Blob CSI's complex solution, the most efficient way for our internal usage may be monitoring FailedMount events (mainly at the moment of daemonset upgrading).

from kadalu.