JorgeB

It's logged as a disk problem, run an extended SMART test.

Nothing NVMe related is being detected by Linux, this is a hardware issue, but no idea how it's being detected in the BIOS.

Yes, that's why I said to look at the tests, depending on the devices used (and the number of them) it might be the same, and with 6 HDDs it will.

Please post diags.

Same, values on the actual diags are correct, looks like just a GUI issue, and of course the fact that it prevents spindown: [sde] => 0 0 22791858 13954498 [sdc] => 0 0 111481 174 3rd column is total reads, 4th is total writes.

Oct 18 19:19:02 Cube kernel: ata10: SATA link down (SStatus 0 SControl 300) Oct 18 19:19:02 Cube kernel: ata10.00: disabled Disk dropped offline, try replacing cables.

OK, if it's only that and it still keeps happening if might really be a visual bug, but it's very strange, you didn't post new diags now, but in the old ones the actual reads/writes values for both devices are completely different, not what the GUI was showing when the screenshot was taken.

Looks like there never was a valid xfs fileystem on that disk, most likely it was added to the array but never formatted.

Depends on the number and type of devices connected, this will give you a good idea of the performance of both: https://forums.unraid.net/topic/41340-satasas-controllers-tested-real-world-max-throughput-during-parity-check/

The please post new diags and the output of: btrfs fi usage -T /mnt/disk1

Do this: with the array stopped click on disk18 and change the fs from auto to xfs, then start array and post new diags.

This was disk5 in the diags posted: Sep 25 22:56:48 Cube kernel: md: import disk5: (sdh) WDC_WD30EZRX-00AZ6B0_WD-WMC070040204 size: 2930266532

Disk3 is failing to initialize, replace cables and/or try a different controller: Oct 15 15:53:41 Tower45TB kernel: ata2: SATA link up 6.0 Gbps (SStatus 133 SControl 300) Oct 15 15:53:41 Tower45TB kernel: ata2.00: failed to read native max address (err_mask=0x1) Oct 15 15:53:41 Tower45TB kernel: ata2.00: HPA support seems broken, skipping HPA handling Oct 15 15:53:41 Tower45TB kernel: ata2.00: READ LOG DMA EXT failed, trying PIO Oct 15 15:53:41 Tower45TB kernel: ata2.00: failed to get NCQ Send/Recv Log Emask 0x1 Oct 15 15:53:41 Tower45TB kernel: ata2.00: failed to get NCQ Non-Data Log Emask 0x1 Oct 15 15:53:41 Tower45TB kernel: ata2.00: ATA-9: WDC WD102KFBX-68M95N0, VCGAJHLM, 83.00A83, max UDMA/133 Oct 15 15:53:41 Tower45TB kernel: ata2.00: 19532873728 sectors, multi 16: LBA48 NCQ (depth 32), AA Oct 15 15:53:41 Tower45TB kernel: ata2.00: failed to get NCQ Send/Recv Log Emask 0x1 Oct 15 15:53:41 Tower45TB kernel: ata2.00: failed to get NCQ Non-Data Log Emask 0x1 Oct 15 15:53:41 Tower45TB kernel: ata2.00: configured for UDMA/133

If you haven't yet take a look here: https://forums.unraid.net/topic/46802-faq-for-unraid-v6/?do=findComment&comment=819173

Was disk18 ever formatted? According to the log there's no valid filesystem.

Looks more like a power/connection problem, try replacing the cables.

Download pool filesystem is corrupt, best bet is to backup and reformat.

If you still need help with this post current diags.

This suggests a partition problem, not a filesystem problem, please post output of: fdisk -l /dev/sdX

Look for "Power Supply Idle Control" (or similar) in the BIOS and set it to "typical current idle" (or similar).

That model shouldn't need the iommu workaround.

Is the issue only upload or also download? Did you ran iperf the other way?

Most likely, you can try it in a different controller, if issues persist replace or ignore, if there aren't any apparent issues.

If it started crashing without any changes it suggests a hardware problem, start by running memtest.

Screenshot doesn't show everything, you can enable syslog server and post that log after a crash.