JorgeB

Start here: https://forums.unraid.net/topic/46802-faq-for-unraid-v6/?do=findComment&comment=819173

Please post the diagnostics: Tools -> Diagnostics

I have some N300 6TB, as well as the X300, good performance and no failures so far, I do find them a bit noisy though, rotation noise, not seek, if that's a problem.

Stop array, unassign parity2, start array.

Yep. Correct.

If this is available it will fit any controller.

SMART report for that disk will show the estimated time for an extended test, e.g: Extended self-test routine recommended polling time: ( 983) minutes.

You can do that, still need to change the UUID of the old disk to be able to have both mounted simultaneously after. Also note that if you haven't yet you'll need to start Unraid with both parity and disk1 unassigned to make Unraid "forget" them and be able to re-assign them.

You may have a ghost in the machine... There are signs of the problem earlier: Feb 12 00:01:28 Tower kernel: sd 10:0:2:0: attempting task abort! scmd(0000000003923bdb) Feb 12 00:01:28 Tower kernel: sd 10:0:2:0: [sdk] tag#6275 CDB: opcode=0x85 85 09 0e 00 00 00 02 00 07 00 00 00 00 00 2f 00 Feb 12 00:01:28 Tower kernel: scsi target10:0:2: handle(0x000b), sas_address(0x4433221104000000), phy(4) Feb 12 00:01:28 Tower kernel: scsi target10:0:2: enclosure logical id(0x500605b00991da10), slot(7) Feb 12 00:01:32 Tower kernel: sd 10:0:2:0: task abort: SUCCESS scmd(0000000003923bdb) Feb 12 00:02:08 Tower kernel: sd 10:0:2:0: device_block, handle(0x000b) Feb 12 00:02:10 Tower kernel: sd 10:0:2:0: device_unblock and setting to running, handle(0x000b) Feb 12 00:02:10 Tower kernel: sd 10:0:2:0: [sdk] Synchronizing SCSI cache Feb 12 00:02:10 Tower kernel: sd 10:0:2:0: [sdk] Synchronize Cache(10) failed: Result: hostbyte=0x01 driverbyte=0x00 Feb 12 00:02:10 Tower kernel: mpt2sas_cm0: removing handle(0x000b), sas_addr(0x4433221104000000) Feb 12 00:02:10 Tower kernel: mpt2sas_cm0: enclosure logical id(0x500605b00991da10), slot(7) I would swap that disk with another from the onboard SATA controller to see if it changes anything.

Do you mean you're using SATA for the vdisk Bus? That shouldn't be a problem, but you can try a different setting..

If you assign a drive to a previously unused slot Unraid will clear the disk in the background then add it to the array maintaining parity.

https://www.phoronix.com/scan.php?page=news_item&px=AMD-Hiring-More-Linux-2021 This will hopefully end up improving the experience for Unraid users with AMD hardware, like for example the common controller and IOMMU issues with some AMD hardware and Unraid/Linux, as well as overall stability.

If you have a spare rebuild to the spare then check the data on the old disk, you can do that with UD but need to change the UUID to be possibly to mount with the array started, then you can compared data from both disks to see if anything is missing.

After the preclear is done run the diskspeed docker.

Can't see the disk errors due to log spam.

This was fixed for v6.9, at least used and free space, exception is a raid1 pool with an odd number of devices, but that is a btrfs issue.

Ahh, missed that, you can try this:

There's a warning when you format a drive that all data will be lost and parity can't help.

That is a setting: Settings -> Disk Settings -> Enable auto start What is a while? Are those diags downloaded when the server is inaccessible?

Probably not.

Yes, that or use an ad-don controller, but note that one of the NVMe devices is also having issues, so probably best to try a different board if that's an option.

Some AMD boards have issues with the SATA controller and Linux, sometimes BIOS update or upgrading to v6.9 helps, since you already did both not much more I can suggest.

Yep, it's what I do. Correct.

You need to remove -n or nothing will be done.

Interesting, but it mentions 12 M.2 SATA devices, not NVMe