This is a mini 4 bay M.2 computer with an Intel Twin Lake Duel core  Celeron Processor, 12 Gig LPDDR5 Memory, 2x 2.5 Gig Network, etc.

I have 4 – 2TB M.2 drives installed in it.

I purchased a 2^nd license to run Unraid on it.

The 1^st USB I set up installed the OS on the G9 but started dropping the M.2’s and crashing so I think I setup a bad USB.

Created a new USB and setup a trial license on then booted it up. Boot and install went fine, made the new cache pool, named it “cachepool”. Assigned the drives, they showed up and on starting the pool, set it to format, after a few minutes it stopped with a “unmountable or wrong file system”, and dropped 3 drives. They were just gone. Rebooted and they showed up again, reassigned them to the pool, tried to format the pool again and no dice. Downloaded diagnostic then restarted the PC. Reassigned them to the pool again, set it to format and still failed but drives were still there as started a format again, it formatted for about 15 minutes and failed again. It shows the pool as btrfs, no size, and temp for the 1^st drive but no temp for the other drives.

Shut down the server, rebooted into Win 11 and deleted the volumes on the 4 M.2 drives. Rebooted in to Unraid and the drives are still shown assigned but no file system. Started the pool and then the format. After a few minutes it becomes unresponsive then crashes, or reboots or back to Unmountable or wrong file system.

Not sure how to proceed from here.

tower-diagnostics-20250820-1132.zip tower-diagnostics-20250820-0842.zip

So, it will not let me format a pool with the M.2 Drives, but it allowed me to make an array with the 4 drives. 1 set as Parity and 3 as data. Took about 1.5 hours to make parity and seems to be working so far.

Have made some shares, installed a docker and am copying file to a share from OneDrive.

We will see how it goes.

At this time I can triple boot this Mini PC, Unraid, Win on USB, and Ubuntu.

Here is the latest diagnostics.

tower-diagnostics-20250820-2047.zip

Devices are dropping offline after the blkdiscard commands:

Aug 20 11:05:17 Tower emhttpd: shcmd (196): /sbin/blkdiscard /dev/nvme3n1p1

Aug 20 11:05:47 Tower kernel: nvme nvme3: I/O tag 385 (2181) opcode 0x9 (I/O Cmd) QID 2 timeout, aborting req_op:DISCARD(3) size:4294966784

Aug 20 11:05:47 Tower kernel: nvme nvme3: I/O tag 386 (1182) opcode 0x9 (I/O Cmd) QID 2 timeout, aborting req_op:DISCARD(3) size:4294966784

Aug 20 11:05:47 Tower kernel: nvme nvme3: I/O tag 387 (1183) opcode 0x9 (I/O Cmd) QID 2 timeout, aborting req_op:DISCARD(3) size:4294966784

Aug 20 11:11:34 Tower emhttpd: shcmd (201): /sbin/blkdiscard /dev/nvme1n1p1

Aug 20 11:12:05 Tower kernel: nvme nvme1: I/O tag 320 (c140) opcode 0x9 (I/O Cmd) QID 3 timeout, aborting req_op:DISCARD(3) size:4294966784

Aug 20 11:12:05 Tower kernel: nvme nvme1: I/O tag 321 (e141) opcode 0x9 (I/O Cmd) QID 3 timeout, aborting req_op:DISCARD(3) size:4294966784

Aug 20 11:12:05 Tower kernel: nvme nvme1: I/O tag 322 (8142) opcode 0x9 (I/O Cmd) QID 3 timeout, aborting req_op:DISCARD(3) size:4294966784

Aug 20 11:17:51 Tower emhttpd: shcmd (206): /sbin/blkdiscard /dev/nvme2n1p1

Aug 20 11:18:21 Tower kernel: nvme nvme2: I/O tag 962 (63c2) opcode 0x9 (I/O Cmd) QID 3 timeout, aborting req_op:DISCARD(3) size:4294966784

Aug 20 11:18:21 Tower kernel: nvme nvme2: I/O tag 963 (23c3) opcode 0x9 (I/O Cmd) QID 3 timeout, aborting req_op:DISCARD(3) size:4294966784

Aug 20 11:18:21 Tower kernel: nvme nvme2: I/O tag 964 (03c4) opcode 0x9 (I/O Cmd) QID 3 timeout, aborting req_op:DISCARD(3) size:4294966784

This suggests a device firmware issue; they are not correctly handling a full device TRIM. Array devices are not trimmed, hence the difference.

Thank you JorgeB. That makes sense. Looks like nvme0 worked fine but 1, 2, & 3 have the issue. They are the newest and least expensive m.2's I bought to populate the box.

When I am flush again I might buy 4 Samsung 990's, that is what I have in my Qnap box, they work fine.

Oh and you helped me on my other server that was a power issue, have updated it to the beta release and has been running fine for 2 days so far.