I made some changes to my hardware. Added a few disks, changed HBA, etc. Ulitmateltely nothing major though. Still same Mobo/CPU/Ram/GPU and also still the same flash disk. So not sure why this should have an impact.

A few things were weird after starting my Unraid server:

• SSH was disabled. I am certain I didn't change this setting. Had to re-enable it again. Not a big deal

• My cache pool wasn't assigned anymore. So, I just manually added the two mirroed ZFS disks back. All worked after adding them back. I wasn't sure which of the two ZFS NVME disks was the main one and which one the mirrored one and don't whether it matters. All data still there and all good.

• One file on the cache pool is outdated. No clue why and I just edited it back. Just a list of folders for mover exclusion, but weird that it changed it back

• Time on my Windows was out of time. I had to manually set it back

Net net, nothing major, but very weird... Any idea what could drive this change and anything to watch out for? And also whether my practice to randomly assign the two zfs disks back to a pool is something that can be done this way?

All those settings are stored on the flash drive, so it suggests an issue with it or its filesystem

Mmh... That's weird. Is there a way to check the integrity of the flash disk? Some form of chkdsk or whatever else?

Also, how to go ahead about the zfs pool thing? Does it matter, which drive of the mirror I assign first or can I just swap main with mirror as they are anyways the same?

Lastly, I also noticed a SMART test failed on one of the disks. Not sure this was like this before or whether that's new. The short self-test completes without errors, but the disk shows "failed" in the dashboard. Also when I select "identity", it says that the smart test passed. That's a traditional 18TB HD disk that is part of the array. Not sure anything to worry about and the disk is really failing or just a false negative in the dashboard? As said, the self-test completes without error and the identity tab also says "passed".

6 minutes ago, steve1977 said:

Is there a way to check the integrity of the flash disk? Some form of chkdsk or whatever else?

You can run chkdsk on Windows.

6 minutes ago, steve1977 said:

Does it matter, which drive of the mirror I assign first or can I just swap main with mirror as they are anyways the same?

Nope, create a new pool, assign all the relevant devices in any order, leave the fs set to auto, start array to reimport the pool.

Thanks. Any thoughts on my SMART question?

1 minute ago, steve1977 said:

Thanks. Any thoughts on my SMART question?

You are likely to get better informed feedback if you attach your system’s diagnostics (with everything in the one zip file) to your next post in this thread. It is always a good idea when asking questions to supply your diagnostics so we can see details of your system, how you have things configured, and the current syslog.

Thanks both for your help. Attaching a diagnostic log:

Array disks 5, 6 and 14 show SMART health errors in the dashboard, but not when running the short smart / identity.

Cache 2 is the opposite (NVME disk). Dashboard shows the disk is healthy, but the smart check shows error. You helped me in another threat that I can safely ignore this as this is related to how long I've used the disk, but it does not show any hardware issues (yet).

Unassigned disk 1 is a brand new unformatted disk that I plan to use as future parity disk

Unassigned disk 2 is a HD that I used to have some issues with and trying to move over all files before it dies. It shows though that all files are working.

You can tell that I have some disk issues. I have a 9300-8i HBA and one of my concerns is that it is overheating, which leads to some issues? So, I am considering to upgrade to 9400-8i or 9500-8i, which may have better heat.

Would love to hear your thoughts specifically on disks 5, 6 and 14. Ideally also on unassigned disk 2.

tower-diagnostics-20250928-0919.zip

7 hours ago, steve1977 said:

disks 5, 6 and 14

UDMA CRC errors are usually a bad SATA cable, you can acknowledge the current errors, and if more appear, replace the cables.

7 hours ago, steve1977 said:

unassigned disk 2.

Loosk fine to me.

Thank you. I’ll need to upgrade 3 harddisks to larger size (going to 28TB). Shall I replace the ones with the udma crc errors? Or just the smallest size ones? Or any way to prioritize? Shall I send another diagnostic log with all my current disks to help me prioritize?

42 minutes ago, steve1977 said:

Shall I replace the ones with the udma crc errors?

Like mentioned, that's not a disk problem; I would not take that as motivation to replace those first.

How would you determine which one to replace first?

I didn't check the other disk's SMART reports, but if they are all good, I would replace the older/smaller disks first.