With the tracking in place from issue #2, node drivers fetch existing mounts from controller and re-mount on startup.
When the CSI driver is updated or upgraded, all existing mounts fail (mfsmount binary is in the node driver container).
With tracked volumes, the node driver can get the mounted volume records from the controller and re-mount them on startup. Depending on the error handing in the running containers, the services could be recovered without restart.
With the tracking in place from issue #2, node drivers fetch existing mounts from controller and re-mount on startup.
When the CSI driver is updated or upgraded, all existing mounts fail (mfsmount binary is in the node driver container).
With tracked volumes, the node driver can get the mounted volume records from the controller and re-mount them on startup. Depending on the error handing in the running containers, the services could be recovered without restart.