A few months ago, I installed a Dell H310 HBA flashed to IT mode in my server.  Prior to this, with all disks connected the MB SATA ports, I never saw a single parity error for many years.  Ever since I installed the H310, I ALWAYS get 1 parity error on every parity check.  

 

The parity check is set to non-correcting.  After the first couple of times I saw the error, I ran a correcting check which said it corrected one parity sync error.  Good, or so I thought.  After the error was corrected, the next parity check would always result in 1 error again.  For the last couple of months I have not corrected the error, so. of course, I would expect to see exactly what I see below.  

 

My monthly checks run automatically on the 15th of each month.  Unfortunately, I rebooted the server on June 25, so I only have the July 15 parity check in the logs.  I need to set syslog server up again.

 

Last night I started another parity check and it only produced the 1 error (as expected) since it had not been corrected.

 

Jul 15 05:24:04 MediaNAS kernel: md: recovery thread: P incorrect, sector=4744060216

Jul 26 00:54:44 MediaNAS kernel: md: recovery thread: P incorrect, sector=4744060216

 

Any ideas what could consistently cause 1 parity error with the Dell H310?  Even after being "corrected," 1 error gets detected in subsequent checks.  I probably need more data to see if is always at the same location.  I suppose I could confirm my theory that is has something to do with the H310 by connecting all disks to MB SATA ports for the next check, but so far the evidence is fairly strong in support of that theory.

 

My server has ECC RAM.

 

S.M.A.R.T. reports on all disks look good.

I would be suspicious of the H310 since the problems started after you got it.  Which server is having this problem?  Are you using all eight ports on the H310?   If not, you could either swap the SATA data connectors to use the ones you have not used.  Or, if you are only using one of the two connectors on the H310, move the connector to the second 8087 port.  

Edited July 26, 20196 yr by Frank1940

32 minutes ago, Frank1940 said:

Which server is having this problem?  Are you using all eight ports on the H310?

This is on the "Main" server in my sig.  I currently am using only four of the eight ports on the H310.  I have an 8-bay HDD hot-swap cage in the server and the plan is to move all 8 of these disks (I only have five in it right now) to the H310 and leave the MB ports for SSDs and the Optical drive.  I have the second SAS/SATA cable but I have not connect it. 

 

I think my current configuration is the Parity drive is connected to the MB and the four current data drives are connected to the H310.  I can try moving that cable to the second 8087 port to see if it makes a difference.

I would do the swap to the second 8087 port and run either a non-correcting or a correcting parity operation.  (Depending on whether I had fix a error previously found.  (Begin logging all failure info so if there is a pattern, someone might see it.)  Then immediately run the complimentary operation to see if it was 'fixed' or to fix it.   If port swapping does not fix it, then swap the cable. 

 

You will be logging some real testing hours on the hard drives but the ones that are in that server are NAS-type drives and there is no question that they are designed to be able to do it!