A few weeks ago, my Samsung SSD cache disabled. I moved an already-mounted SSD to act as cache, and pulled this one.

Testing found no issues, so I planned to reformat/restore it, thinking it was a transient issue. I did that yesterday, and overnight it was disabled again. Specifically, per syslog:

• As of 4AM ET, mover ran and no errors reported

• As of 4:40AM ET, Fix Common Problems ran and reported "Error: cache (Samsung_SSD_990_PRO_2TB_S7KHNJ0X312749R) is disabled"

I've run the full SMART test and attached it here, along with the diags from before that full test. That came up clean, as well.

The only other thing I could find is below, in my dmesg:

[66672.535303] XFS (nvme0n1p1): Mounting V5 Filesystem 00ecfc48-3876-4328-95b5-1615bfbef65e
[66672.541689] XFS (nvme0n1p1): Ending clean mount
[152151.312270] block nvme0n1: the capability attribute has been deprecated.

Ideas on what I'm missing here, welcome. Thank you.

morningstar-smart-20260909-0626(1).zip morningstar-diagnostics-20260909-0630.zip

Make sure your Samsung drive is on the latest firmware.

Also there are issues With SSDs dropping due to some power saver settings to fix that you need to append pcie_aspm=off to the boot parameters and reboot.

1 hour ago, raqisasim said:

XFS (nvme0n1p1)

The cache device is nvme1n1, and as far as I can see, it's working normally; it looks like an FCP false alarm.

Update -- another cache disable, different SSD(?!?!) -- first, thank you all for the advice and help!

But here's the new weirdness (new diags, attached). Per my original post, I did as I had before when this 1st occurred: I cleaned off and used a differed SSD, already in my system, for cache.

Then, overnight (again!) -- Fix Common Problems ran at 4:40AM and 7:50AM ET (this a manual run), and both times reported my new cache drive was disabled:

Sep 10 07:50:12 Morningstar root: Fix Common Problems: Error: cache (ORICO_UB202411210915) is disabled

But! Then I went and looked, and the cache appear normal:

So that's even more weird, right? This is a totally different SSD, not even the same manufacturer, and one that's been mounted and in-use on and off as a "scratch pad" for almost a year, now, as well as the prior cache until I re-installed per 1st post.

I do also notice none of my drives are spinning down:

Sep 10 08:11:07 Morningstar root: Total Spundown: 0

Also: MowMdown's pcie_aspm=off is already implemented in my system, and has been since before this fancy new interface:

And I am running CPU in Best Performance mode.

I get it being a false alarm save that the Samsung one really did go disabled on me, so something happened, there.

For now, I'll disable Mover as a starting point, run Mover manually when I can this evening and check things right after.

Other ideas, or things I've missed, are welcome!

morningstar-diagnostics-20260910-0807.zip

Edited September 10Sep 10 by raqisasim

something wierd happened where unraid tried to just mount /dev instead of /dev/nvme2n1p1 to /mnt/cache so it aborted.

Sep  8 06:46:06 Morningstar emhttpd: mounting /mnt/cache
Sep  8 06:46:06 Morningstar emhttpd: shcmd (99708): mkdir -m 0666 -p /mnt/cache
Sep  8 06:46:06 Morningstar emhttpd: cache: btrfs verify devices
Sep  8 06:46:06 Morningstar emhttpd: /sbin/btrfs filesystem show /dev/ 2>&1
Sep  8 06:46:07 Morningstar emhttpd: ERROR: not a valid btrfs filesystem: /dev
Sep  8 06:46:07 Morningstar emhttpd: shcmd (99709): rmdir /mnt/cache
Sep  8 06:46:07 Morningstar emhttpd: cache: mount error: wrong or no file system

And clearly UD plugin mounted it just fine

Sep  8 06:46:13 Morningstar unassigned.devices: Udev: Refresh udev info for /dev/nvme2n1.
Sep  8 06:46:13 Morningstar unassigned.devices: Udev: Refresh udev info for /dev/disk/by-id/nvme-ORICO_UB202411210915-part1.
Sep  8 06:46:13 Morningstar unassigned.devices: Partition found with the following attributes: serial_short='UB202411210915', device='/dev/nvme2n1p1', serial='ORICO_UB202411210915', uuid='1b7f0225-c41d-42c2-bb36-074355493b2a', part='1', disk='/dev/nvme2n', label='themachine', disk_label='themachine', array_disk='', fstype='btrfs', mountpoint='/mnt/disks/themachine', luks='', file_system='btrfs' ...
Sep  8 06:46:13 Morningstar unassigned.devices: Mounting partition 'nvme2n1p1' at mountpoint '/mnt/disks/themachine'...
Sep  8 06:46:13 Morningstar unassigned.devices: Mount cmd: /sbin/mount -t 'btrfs' -o rw,relatime,space_cache=v2 '/dev/nvme2n1p1' '/mnt/disks/themachine'
Sep  8 06:46:13 Morningstar kernel: BTRFS: device label themachine devid 1 transid 605 /dev/nvme2n1p1 (259:4) scanned by mount (2372013)
Sep  8 06:46:13 Morningstar kernel: BTRFS info (device nvme2n1p1): first mount of filesystem 1b7f0225-c41d-42c2-bb36-074355493b2a
Sep  8 06:46:13 Morningstar kernel: BTRFS info (device nvme2n1p1): using crc32c (crc32c-lib) checksum algorithm
Sep  8 06:46:13 Morningstar kernel: BTRFS info (device nvme2n1p1): enabling ssd optimizations
Sep  8 06:46:13 Morningstar kernel: BTRFS info (device nvme2n1p1): turning on async discard
Sep  8 06:46:13 Morningstar kernel: BTRFS info (device nvme2n1p1): enabling free space tree
Sep  8 06:46:13 Morningstar unassigned.devices: Successfully mounted '/dev/nvme2n1p1' on '/mnt/disks/themachine'.

I would just stop the array and start it back up again after unmounting the drive from UD.

edit it appears you mounted it near the end of the logs so I assume it's working?

Sep  9 11:30:50 Morningstar emhttpd: shcmd (99947): mkdir -m 0666 -p /mnt/cache
Sep  9 11:30:50 Morningstar emhttpd: cache: btrfs verify devices
Sep  9 11:30:50 Morningstar emhttpd: /sbin/btrfs filesystem show /dev/nvme2n1p1 2>&1
Sep  9 11:30:52 Morningstar emhttpd: Label: 'themachine'  uuid: 1b7f0225-c41d-42c2-bb36-074355493b2a
Sep  9 11:30:52 Morningstar emhttpd: #011Total devices 1 FS bytes used 144.00KiB
Sep  9 11:30:52 Morningstar emhttpd: #011devid    1 size 953.87GiB used 3.02GiB path /dev/nvme2n1p1
Sep  9 11:30:52 Morningstar emhttpd: mounting: cache devices: 1 slots: 1 missing: 0
Sep  9 11:30:52 Morningstar emhttpd: shcmd (99948): mount -t btrfs -o noatime,nodiscard,space_cache=v2 /dev/nvme2n1p1 /mnt/cache
Sep  9 11:30:52 Morningstar kernel: BTRFS: device label themachine devid 1 transid 608 /dev/nvme2n1p1 (259:4) scanned by mount (2559947)
Sep  9 11:30:52 Morningstar nfsrahead[2559963]: skipping non-NFS device btrfs-6
Sep  9 11:30:52 Morningstar kernel: BTRFS info (device nvme2n1p1): first mount of filesystem 1b7f0225-c41d-42c2-bb36-074355493b2a
Sep  9 11:30:52 Morningstar kernel: BTRFS info (device nvme2n1p1): using crc32c (crc32c-lib) checksum algorithm
Sep  9 11:30:52 Morningstar kernel: BTRFS info (device nvme2n1p1): enabling ssd optimizations
Sep  9 11:30:52 Morningstar kernel: BTRFS info (device nvme2n1p1): enabling free space tree
Sep  9 11:30:52 Morningstar emhttpd: shcmd (99949): mount -o remount,discard=async /mnt/cache
Sep  9 11:30:52 Morningstar emhttpd: update_pool_cfg: 30 cache 0
Sep  9 11:30:52 Morningstar kernel: BTRFS info (device nvme2n1p1 state M): turning on async discard

Edited September 10Sep 10 by MowMdown

On 9/10/2026 at 9:12 AM, MowMdown said:

edit it appears you mounted it near the end of the logs so I assume it's working?

Yes -- the weirdness per my last post was that Fix Common Problems said it was Disabled, but it wasn't. Sorry for the delay getting back; we've had a few other things 'round the house impacting my time to debug this.

So what appears to have happened, was something around the Mover not actually moving anything. There was an additional script to help QBittorrent work with the Mover that actually had stopped working, but wasn't throwing any visible errors.

End result is that I killed that script, and then re-ran Mover and now it's moving, plus Fix Common Problems says everything is OK with the cache, now. So I'm going to tentatively blame this all on this, and remember to keep an eye on fancy helper scripts going forward!