Cache drive problems

[Idolwild]

Got a bit of a mess here. Added a NVMe (one of two I was going to add) to an existing SSD cache, and things seems to have gone haywire - cache claims it's mounted R/O, getting "Drive mounted read-only or completely full. Begin Investigation Here" in FCP, getting a "Cache pool BTRFS too many profiles" warning, and balancing doesn't seem to be doing anything. I've since taken the old SSD drive (256GB) out of the pool (it had 5x CRC errors due to "old age") and put in the second NVMe, but the disk won't mount in anything but RO, and I seem stuck. I have /appdata/ backups from the last month (hopefully good) so I don't lose all my Docker configs.

 

Can someone guide me through resuscitating the current cache pool?

or

Starting over - how do I format the NVMe's and begin an entirely new cache pool?

(whichever is best advised)

 

I've searched the forums - a lot of people having almost similar problems, but I don't see a clear solution after banging my head against it. Any help welcome.

 

 

Edited August 4, 2020 by Idolwild
spelling/clarification

[Idolwild]

Reporting back in case someone has the same problem. I got impatient with waiting for some hand-holding, so chose a path and forged it. I don't have the expertise to troubleshoot the problem and fix it, so I chose the "scorched Earth approach" - I knew I had good daily backups from the "Backup/Restore Appdata" plugin, so I was hoping that I could get away with zero-data-loss.

 

Steps I took:

 

1. Stopped the array
2. Deprovision both malfunctioning cache drives
3. Use the "Unassigned Devices/Plus" plugins to remove/format the NVMe drives - remove the current partition, format the drive, pre-clear if you want to (requires "Destructive Mode" enabled)
4. I decided to only use one of the NVMe drives for cache going forward - I'm going to reserve it for my /appdata/ and incoming file transfers, but I'm going to move media downloads off to the second NVMe drive hosted as an "Unassigned Device" that is auto-mounted - I want to keep the risk of filling up the drive with downloads OFF the cache pool - and I didn't want to waste an entire TB on a cache that I could easily restore from backup while moving the file system volatility off to something else equally fast
5. Brought the array back online, confirming that I wanted to acknowledge the missing second disk and the blank first disk
6. Used the Backup/Restore plugin to re-populate the new cache drive with my appdata
7. Deleted and resized (downward) the Docker image
8. Rebooted and performed some chants
9. When the array came back up, re-installed my important Docker images - all /appdata/ config was saved and worked, everything remembered it's settings and such
10. Success!

Hope this helps someone - always do your backups!!!

Edited August 5, 2020 by Idolwild