JorgeB

You should update the LSI firmware, replace cables on disk7 and post new diags.

For this you just need to backup the cache pool, though you should already have backups of anything important.

Not really, if the first time it was wrongly correctly due to a RAM bit flip, second time it would be corrected again to return to original state.

Please post new diags.

Yes

New firmware wans't going to make the disks appear, it was just in case they dropped because of that, try power cycling the server (not just rebooting), if they don't come back online replace/swap cables on those disks.

Better to re-post there, there's some danger with this forum when attempting to merge threads.

No, they just show data corruption, bad RAM is the #1 reason for that, but there could be other reasons, or the problem is not being detetced by memtest, it's not always

If it happens again grab diags before rebooting, for now you can swap cables/slot with another disk, to rule that out if it happens again to the same disk.

You can monitor btrfs filesystem for corruption (and other) errors, if data corruption is found a scrub will list the affected files in the syslog. It would be possible if you knew corruption happened on the data device and parity wasn't corrected, like after actual bit rot, but most corruptions are caused by other factors, and from those not easy or possible to recover from parity, and since each array drive is a single device filesystem and doesn't have redundancy it also can't be fixed by btrfs, that's just one many reasons why backups of anything important are still needed.

This, usually it's plug and play as far as the array is concerned, unless using RAID controllers, VMs might require some changes.

Unraid only disables a drive after a write error, posting the diagnostics, if you didn't reboot yet, might give more clues.

You need a cooler with narrow ILM LGA 2011 support, e.g.: https://noctua.at/en/nh-u12dx-i4 Or a cheaper but still good cooler: https://store.supermicro.com/4u-active-cpu-cooler-snk-p0050ap4.html

Those type of errors are usually bad power/connection, what have you replaced so far?

They are logged as disk errors, though they can be intermittent, like mentioned extended SMART test is what you should do.

Boot menu of Unraid, there's a memtest option, only works for CSM/legacy boot, not UEFI boot.

You're still having multiple disk errors: Sep 2 07:38:55 tower kernel: md: disk10 read error, sector=10128 Sep 2 07:38:55 tower kernel: md: disk8 read error, sector=12800 The disabled disks can't be correctly emulated with errors on additional disks.

Data corruption was detected on the pool, you should run memtest, then backup and reformat the pool.

It's a bug, fixed for rc2, but if you remove the empty pool it should appear.

You can try, but the other device is out of sync.

Then you only option would be using a file recovery util, like UFS Explorer.

Mover will only move to the array if use cache=yes for that share.

Most times it doesn't, see here: https://forums.unraid.net/topic/46802-faq-for-unraid-v6/?do=findComment&comment=700582

What type of pool? raid1? Diags only show what I assume is the formatted cache device, where's the other one?

Not that I know of, but this is very easy to do with the CLI, just need to specify the source/dest paths.