My cache drive died (Toshiba Nvme 256 GB, linux error ( nvme nvme0: Device not ready; aborting reset, CSTS=0x3).

 

When setting up Unraid, i read about the (nightly) 'mover' process, and didn't pay any perticular interest into my share settings.  I installed two dozen docker containers and 1 VM since, all docker containers using the /mnt/cache/appdata paths.

 

If i now delete the cache pool with the missing nvme disk and restart the array, will i have all Docker and VM data back from say yesterday because Unradi is then running solely on the array? And if i put in a new Nvme and recreate a cache pool, will the contents of appdata automatically will moved back from the array to the cache?

 

Luckily i had appdata and vm backups running, but just curious if i could delete the cache pool and run on the array for the few days awaiting the new nvme.

Edited February 10, 20251 yr by peewee67

27 minutes ago, peewee67 said:

f i now delete the cache pool with the missing nvme disk and restart the array, will i have all Docker and VM data back

Depends on where they were being stored, if in the pool, they will all be missing.

They were all 'stored' in /appdata in the cache pool, consisting of 1 (now missing) nvme disk. But if that means they now all will be missing, what is the purpose of the mover process to begin with? I'll read up om the mover process in the mean time.

I have been reading up on the mover process (the post below among others)  but i seem to be missing a point.

 

 

If for Appdata the primary storage is set to Cache and secondary to Array, Docker container configs and everything else in /appdata is primarily stored on the cache drive(s) in the cache pool for performance reasons and will be moved to the array later (i understand databases for instance won't get moved while in use) . If then the cache pool dying beyond recovery results in de contents of /appdata being lost, what is the purpose of the mover process in the first place? I mean, it's not much use moving the contents to the array if i can't access it directly from the array in some way?

 

I had databases for Docker containers running on the cache pool also (because of the performance reasons), is that a good or bad practice?

Apddata is typically not moved to the array, and certainly not by default, and note that if the docker service is enabled, it's not possible to move open files, so it would fail.

It is normal to have the appdata share set to be only on the pool, and then use the appdata backup plugin to make a backup copy on the array in case the pool fails.