JorgeB

Not a good sign, still run another one so we can compare the affected blocks.

Looking at the diags this might be an xfs issue/bug: Aug 13 15:07:16 Tower kernel: XFS (md3): Internal error !uuid_equal(&mp->m_sb.sb_uuid, &head->h_fs_uuid) at line 278 of file fs/xfs/xfs_log_recover.c. Caller xlog_header_check_mount+0x4d/0x96 [xfs] You can try installing latest xfs_progs (5.7.0), it might help, download from here then put the txz file on a folder called "extra" on your flash drive, reboot, after boot check that it loaded (xfs_repair -V) and run it again on disk3. Whether it helps or not don't forget to then remove the extra folder on the flash after it's done.

Doesn't look like a disk issue, maybe power/cables, if you have spare cables and/or PSU try that. You can run a check on disk2, but firt you need to run xfs_repair on it.

I assume that was a non correcting check?

That's a strange error, did you just change the case? Controller is the same?

The difference between those models is that the EL2 has dual expanders, but that's not for speed, it's for redundancy, your current backplane already supports dual link, that doubles available bandwidth, for real world usable bandwidth examples see below: https://forums.unraid.net/topic/41340-satasas-controllers-tested-real-world-max-throughput-during-parity-check/?do=findComment&comment=406521

Start the array in maintenance mode then: xfs_repair -v /dev/md3

Any unclean shutdown before this check? If not run another check and post new diags if more errors.

Most likely just one that doesn't work well with an LSI, and IIRC I've seen it before with that model, if you can't use the Intel ports try a different model SSD if possible.

Best way is to use SSH Key-Based Authentication so no password is needed.

SATA3_A1 and A2 are the Asmedia ports.

It doesn't, it looks more like a connection issue, replace both cables.

If you continue to read the quoted post you see why Unraid doesn't support that.

That's fine, you could also replace the existing one first (see FAQ notes about single devive replacement) then add the other one.

Yes.

Could be a bad PSU, or just a cable/splitter if they share one, could also be the miniSAS to SATA breakout cable for example.

Again looks like a power/connection problem.

diags?

Most likely, unfortunately nothing of the rebuild is on the diags, you need to check the data, that's why it's good to have checksums of all files. Just the everything looks normal so far.

Unfortunately only thing visible in the syslog is filesystem corruption on disk2, that caused the syslog to be spammed and there are many missing hours, and no disk errors visible. Multiple disk errors in apparently healthy disks suggest a controller/power/connection problem, but difficult to guess without the syslog showing the issues. I would reboot to clear the log and re-sync parity, if more error post new diags.

I'm sorry, don't have any ideas other than you need to rule out one thing at at time.

Dual parity could be used to identify which disk is corrupt, the problem is that if there's corruption on multiple devices it might corrupt the data even further, more info here: https://forums.unraid.net/topic/46170-unraid-server-release-620-beta20-available/page/10/?tab=comments#comment-456693

You can do that but note that LSI HBAs can only trim SSDs with read zeros after trim support.

Yep.

Any non rotationtional device, in the array or a pool.