I have a peculiar problem with my cache drive in one of my servers, a Crucial MX500 1 TB on latest firmware (as recommended):

1. I get CRC-errors on this drive
2. It also shows up as an 'unassigned device'
3. I cannot write to it, 'Fix Common Problems' reports it's read-only
4. I cannot stop the array unless I reboot (and then stop the array before it loads the services)
5. On reboot it starts a parity check
6. This all started after a firmware upgrade that succeeded uneventfully, or so it seemed.

 

Diagnostics attached.

 

I have a spare drive so I would replace it, but I'd rather know what's going on!

citadelle-diagnostics-20230627-0920.zip

Constant ATA errors, try swapping port/cable with a different device.

Changing the cable doesn't have any influence, the CRC errors keep mounting. Since this system (a HP Microserver gen10) doesn't have any spare ports left I'll try testing the combo on a different system to see whether I get CRC errors there as well.

 

Done, no errors... Weird. Next I'll do a full 'surface' scan, but meanwhile I'm going to replace the drive cf. the instructions on-line and see what happens next. That will mean reinstalling the single docker (syncthing) I have running on this particular system, but that is easily done. I don't trust anything on this drive, so -recreating some shares on the cache drive etc. is probably wise anyway. It's a simple backup-server so that is no big hassle (I wouldn't recommend doing this on a more complex system). I've managed to save the backup-files to the array on the 'data' share, so this means I'll only have to recreate the appdata share.

 

Done that. Server (so far) runs without errors on the cache drive (with new cable on same SATA-port). Will make new config's for all backups in syncthing and see what happens. Will update if and when new findings. 🙂

 

After an hour or two no errors on the SSD in use as cache. A surface scan of the old SSD doesn't find any problems. Gremlins?

Edited June 27, 20233 yr by RikStigter
New information