Monthly parity check always returns 2 errors

[nicosemp]

Hello, so I have read through tens of posts from people having errors but it was always their problem only, found by looking at their diagnostics.

 

My server has three 4TB disks of which one is parity.

 

I have a non-correcting parity check scheduled every first of the month, it lasts about 8 hours and (almost) always finishes with 2 errors (see attached screenshot).

Mind you it's exactly 2! Either no errors (rare) or 2.

As I read on the forum, what I usually do in case of 2 errors is run a correcting parity check, which finds those 2 errors and corrects them.

Then I run another non-correcting parity check and I get no errors (they were just corrected).

The problem is that the 2 errors come back the next month. If I'm lucky I get one error-free month but I have never had two months without errors, even if they're just 2 errors.

 

I should mention that I never had any problems with files or anything.

I also ran a SMART short self-test on all three drives, which finished with no errors.

 

I attached my diagnostics, downloaded right after a parity check finished with 2 errors.

I'll be keeping the server on for additional information if needed.

Any help is greatly appreciated!

It would be great to understand why this happens, even if there is no solution.

 

 

unraid-home-diagnostics-20220701-1037.zip

Edited July 1, 2022 by nicosemp

[JorgeB]

Did you notice if the 2 errors are on the same sectors? They are very close together, if you didn't correct parity and post new diags when it finds more sync errors.

[nicosemp]

Thanks for the reply. It would make sense to find them in the same sectors again.

So diagnostics show the sectors where the errors are found? I didn't know this, where should I look inside the zip file? but I just found them.

 

I will correct parity right now and report back next month, when I expect the 2 errors to come back.

 

Thank you!

 

EDIT: found the sectors near the end of ./logs/syslog.txt

Edited July 2, 2022 by nicosemp

[nicosemp]

Hi @JorgeB, here's the new Diagnostics as promised.

It seems like the 2 errors are on the same sectors.

 

July 1st:

Jul  1 05:50:44 unraid-home kernel: md: recovery thread: P incorrect, sector=3519069768
Jul  1 05:50:44 unraid-home kernel: md: recovery thread: P incorrect, sector=3519069800

 

August 1st:

Aug  1 05:49:49 unraid-home kernel: md: recovery thread: P incorrect, sector=3519069768
Aug  1 05:49:49 unraid-home kernel: md: recovery thread: P incorrect, sector=3519069800

 

Am I looking at the right data?

What does this mean and how can I solve this?

 

Thank you!

unraid-home-diagnostics-20220801-1120.zip

[JorgeB]

Both diags posted don't show a correcting check, assuming you have really done that and the errors return it suggests a problem with a controllers or a disk, controller would be my first guess, very unlikely the Intel controller would cause issues, so I would start with replacing that Asmedia controller with a different model.

[nicosemp]

I ran a correcting parity check right after the July 1st one, which was non-correcting, to fix those 2 errors. Today's check was also a non-correcting one: it's scheduled for the 1st of the month, and the errors came back.

 

I will try a different controller asap and report back, thank you!

[nicosemp]

I replaced the Asmedia controller with a Startech 8P6G-PCIE-SATA-CARD and the September parity check gave 0 errors!

I've had 0 errors with the Asmedia a few times too, so I'll wait for 3 consecutive months with no errors to call it a victory.

Thank you so much!

[kizer]

I was going to say is it a Marvel controller? Lol

Every Parity check I did I always had 2 errors. 

[nicosemp]

0 errors again in October and November: it was the Asmedia controller indeed.

Thank you @JorgeB!