I ran a parity check today and received one error.  I ran from unMenu with the No Correction option.  The error is detected immediately after starting the check.  Of note I also have a disk that was showing a Multi_Zone_Error_Rate of 1 (has since dropped to 0.)  Would it be likely that the disk that had the error was at fault and not the parity drive?  Any suggestions on how to proceed?

 

Edit to add:

 

The line from the syslog showing the error.

Dec 4 21:45:14 DarkTower kernel: md: parity incorrect: 1728

 

This was detected the same 4 different times (I didn't run 4 full checks, after the 1st full check with the 1 parity error I started and stopped 3 more times to verify the error.)

Did you have a system crash / hard reboot?  That is the most common cause of these symptoms.

No crash or hard reboot since the last parity check.

No crash or hard reboot since the last parity check.

It sounds like a real error otherwise it would not be on every check in the same place.

 

An error at that low an address is probably is probably in the housekeeping area of the file-system.   You'll need to do a "check" that does the update if you want to fix it, but there is probably no way to know which disk has the bit that is wrong.

 

In other words you'll be updating parity to match what it reads from the data disks even if it is the data disk with the flipped bit.

Ok, will just let the parity check correct the error and move on then.  Thanks for the help.

This was a parity sync error, right, not a disk error (in the error column).

 

This is not typical behavior.  Sounds like maybe a memory error. Have you run an overnight memory test?  What is the motherboard - is it one that lots of people are using?

This was a parity sync error, right, not a disk error (in the error column).

 

This is not typical behavior.  Sounds like maybe a memory error. Have you run an overnight memory test?  What is the motherboard - is it one that lots of people are using?

 

No disk error (reported by unRAID, but there was the disk with the SMART error reported.)  However the same error occurred at the same spot 4 different times during parity checks.  This is easy to verify as the error occurs almost immediately after starting the parity check.  I swapped out the power supply last week and ran memtest for 10 hours straight with no errors reported.