Hi

 

I am getting sync errors every time I run a parity check. I have done a memtest for 24 hours and the RAM is solid, passed all the tests.

 

At my wits end here - parity checks take 2 days!

 

 

bitpartnas-diagnostics-20240517-0920.zip

7 minutes ago, shabos said:

every time I run a parity check

 

Are these correcting or non-correcting parity checks?

 

IS the number of errors always the same?

 

Do you ever run a correcting parity check?  With this box checked:

 

Edited May 17, 20242 yr by Frank1940

Sync errors are different every time, that box is indeed checked.

This says that the one in the syslog is non-correcting.

 

May 17 18:18:04 BitpartNas kernel: mdcmd (36): check nocorrect
May 17 18:18:04 BitpartNas kernel: md: recovery thread: check P ...

 

You actually have to manually start the Parity check from the MAIN page (with the box checked) to get a correcting parity check.   (You have to jump through hoops to get an self-starting correcting parity check!)

 

EDIT:  don't be a big hurry to do a correcting check.  Let a few more people have a chance to look at things. 

Edited May 17, 20242 yr by Frank1940

Yes. I did a non-correcting check for a few minutes after two back to back parity correcting checks just to see if the errors still occured.

 

I have definitely done correcting checks.

Since memtest is only definitive if it finds an error, I would run a parity check again with just a pair of RAM sticks, if the same try the other pair, that will basically rule the RAM, next suspect would be a disk or the board/CPU.

 

Note that when testing you should always run two passes, since the first pass can find errors caused by the original issue.

 

 

Had a little time and looked at the SMART reports.  Summary below:

 

Disk7 has no report.  (is there a Disk7 installed?)

Disk1 has errors-- recently...

Disk 9 has errors-- old.

Disk10 has errors-- old.

 

I am no expert on interpreting SMART errors, I defer that to @JorgeB,  but I would looking to run the short SMART tests on these disks as starter. 

 

 

Edited May 17, 20242 yr by Frank1940

I did individual RAM tests on each pair and they passed.

 

I think faulty disks is most likely the problem - I'm replacing disk 1, 5, 9 and 10. (Disk 7 is empty)

 

I will let you know how it goes, will be a 3 day process at least.

 

Thanks for the help.

Edited May 17, 20242 yr by shabos

Disk errors should cause ready errors, not sync errors, and when a disk is the problem usually SMART looks fine, though the sync errors could always be cause by a disk that also has SMART issues.

Replaced all disks with SMART errors, parity works now.