Parity Check errors after PSU failure

[taalas]

Hi,

I recently had my PSU fail in my unRAID server. I installed a new PSU and the array came up fine. Did a (non-correcting) parity check and it shows 3547 parity errors.

 

The turn of events:

- The server went down due to the PSU failing

- I tried to find out whether it really was the PSU by using a different one without putting it into the case, the server booted fine, started a parity check which I aborted (doe to wanting to putting the PSU in the case).

- I installed the new PSU

- I started the server and started a new (non-correcting) parity check (which resulted in 3547 parity errors, but no read errors on the drives)

 

The Main page shows no read errors on the drives (which happened the last time i had parity problems).

I had recently invoked an unBALANCE moving process to empty a drive. This was running while the PSU died. Some of the files that were moved during this process now show as being on 2 disks at once (disk 3 and disk 6 in my case, they were to be moved from 3 to 6).

 

Any advice on how to proceed in my case? Should I proceed deleting the duplicates on drive 6 (or drive 3)?

 

What are the next steps?

 

Thanks!

Edited March 29, 2020 by taalas
clarification of events

[JorgeB]

A few sync errors are normal, even expected, after an unclean shutdown, just run a correcting check (and you still need to delete the duplicate files).

[taalas]

Thanks for your quick reply johnnie.black.

 

I edited my initial post to better reflect what happened (I had started the server once between installing the new PSU).

 

So, I should not worry about the parity errors (even if there seem to be alot of them), run a correcting parity check and then delete the duplicate files after? Do I delete them directly from /mnt/disk[n]/... or will this be a problem with parity?

 

I do have drives that show read errors in their SMART logs (afaik the number did not change during this though), but no uncorrectable ones.

 

Thanks for your advice

[JorgeB]

52 minutes ago, taalas said:

So, I should not worry about the parity errors (even if there seem to be alot of them), run a correcting parity check and then delete the duplicate files after?

Number of errors is pretty normal, especially if it went down during writes.

 

52 minutes ago, taalas said:

Do I delete them directly from /mnt/disk[n]/... or will this be a problem with parity?

Yes and no.