6.9.2 - Errors after parity check (3 checks)

[Marco L.]

Hi,

1-2 months ago, power went down. After the automatic parity check Unraid did post-reboot, the server had 34 errors. Since then, I've done 2 more checks (3 in total), and all of the 3 gave 34 errors, so it seems that the errors can't be removed by the parity check. The checkbox for correcting is checked.

 

What can I do ? I've already done a memtest with one successful pass (took already fairly long, a few hours). I don't think it is cable related, as I moved my server beginning of 2021 to the Silverstone NAS case and had to buy brand new special cables for that case (SFF-8643) that worked well for a while before this incident.

 

I've started migrating my array from 4TB to 12TB disks, maybe one of the old 4TB that shows its age ?

 

 

mediaserver-diagnostics-20211114-1607.zip

Edited November 14, 2021 by Marco L.
add unraid version

[itimpi]

There is only one parity check showing in the diagnostics and that is set to be a non-correcting check.

[Marco L.]

Don't know then... That's what I'm seeing

 

 

 

I'm launching a fourth one then !

Edited November 14, 2021 by Marco L.

[itimpi]

57 minutes ago, Marco L. said:

Don't know then... That's what I'm seeing

 

 

 

I'm launching a fourth one then !

It might be worth installing the Parity Check History plugin even if you do not intend to use its capability to split checks up into increments as one of the other features is that parity history entries will have the type of check that was run added as additional information.

[Marco L.]

Ok, after the next planned check, it has found 0 errors ! The one I launched manually did the trick it seems.

 

It seems then that the auto checks done after a crash don't repair the errors, which is very weird, because you need twice the amount of time... Which becomes a lot on 10+TB drives.

[JonathanM]

1 minute ago, Marco L. said:

 

It seems then that the auto checks done after a crash don't repair the errors, which is very weird, because you need twice the amount of time..

There are very good, if a little complex, reasons NOT to automatically repair after a crash. You can always stop a non-correcting check and start a correcting one at any time if you are sure all your hardware is healthy and the only reason for the errors is an improper shutdown.

[Marco L.]

31 minutes ago, JonathanM said:

There are very good, if a little complex, reasons NOT to automatically repair after a crash. You can always stop a non-correcting check and start a correcting one at any time if you are sure all your hardware is healthy and the only reason for the errors is an improper shutdown.

Do you have a link to an explanation topic ? Otherwise I will search it later