Hi all,

I just ran a non-correcting parity check on my Unraid system and got 26 sync errors. Here's the context:

• I recently replaced the CPU in the system.

• There have been a few hard reboots (unclean shutdowns) in the last few weeks.

• I hadn't run a parity check for several months before this.

• This check was non-correcting, so no changes were made.

• All drives have passed SMART tests with no reallocated or pending sectors.

• I did notice a few drives have ICRC errors in the SMART logs, which I understand may point to transient or cable-related issues.

So far, the system seems stable, and no data loss is evident to my knowledge. I'm trying to determine the best path forward.

My questions:

1. Is it reasonable to assume these sync errors were caused by the hard reboots or the CPU replacement?

2. Would you recommend running a correcting parity check now?

3. Should I be concerned about the ICRC errors, or is this something to just monitor unless it increases?

Any advice would be appreciated!

---

If you had unclean shutdowns then a few parity errors are to be expected when running a check.

The way forward would normally be to run a correcting check. This should show the same number of errors as each one is corrected. A subsequent check should show 0 errors.

A small number of CRC errors is not normally something to worry about other than them slightly affecting performance as they trigger retries, but if the count keeps increasing then you should be checking the power/SATA cabling. Note that this number resets so all you can do is stop it increasing.

Sorry, I am a bit paranoid about getting corrupt data and I am new to Unraid.

Should I finish the non-correcting parity check? Or just go straight for the correcting one?

How do I know if my data is corrupted?

Just now, Bob-omb said:

Should I finish the non-correcting parity check? Or just go straight for the correcting one?

Really up to you - probably worth going straight to a correcting check.

1 minute ago, Bob-omb said:

How do I know if my data is corrupted?

The only way is to be certain is to have checksums on the files or to compare it with your backups. This is built in if using the BTRFS or XFS formats, but if using XFS you have to installed something extra such as the File Integrity plugin to get checksums. However if you have not been having disk errors then corruption of files is very rare.

4 minutes ago, itimpi said:

but if using XFS you have to installed something extra such as the File Integrity plugin to get checksums

They are in XFS. I tried that plugin once but I could not find any documentation on how to use it so I gave up. Would love to know how I could use it though.

4 minutes ago, itimpi said:

However if you have not been having disk errors then corruption of files is very rare.

CRC errors doesn't count?

1 hour ago, Bob-omb said:

CRC errors doesn't count?

Nope, they're just retries.

If the new hardware isn't trusted a 24h+ memtest would be a good thing to do. Then check filesystem on all drives/pools, then a correcting parity check.

Edited July 6, 2025Jul 6 by Kilrah