I've been getting 5 sync errors during parity check, and it always seems to be 5, and the same sectors.  if I run a correcting check, they'll be gone on the following check, but show up again on subsequent checks:

 

Quote

Mar 16 02:39:03 Jarvis kernel: md: recovery thread: P corrected, sector=1574961048
Mar 16 02:39:03 Jarvis kernel: md: recovery thread: P corrected, sector=1574961056
Mar 16 02:39:03 Jarvis kernel: md: recovery thread: P corrected, sector=1574961064
Mar 16 02:39:03 Jarvis kernel: md: recovery thread: P corrected, sector=1574961072
Mar 16 02:39:03 Jarvis kernel: md: recovery thread: P corrected, sector=1574961080

 

As far as I know, all shutdowns have been clean.  I do have a couple drives that I have been trying to clear off to retire from the array, but now I'm hesitant.  What else can I do to diagnose and resolve this?

 

 

jarvis-syslog-20240309-0231.zip jarvis-syslog-20240316-1717.zip jarvis-diagnostics-20240316-1316.zip

12 hours ago, adent042 said:

always seems to be 5, and the same sectors.

This is usually a controller issue, or controller/disk combination, do you see the issue if you do two consecutive checks without rebooting? IIRC this usually only happens after a reboot.

I don't think so. Or at least if the first is a correcting check. Is this something to worry about?  My concern would be it masking other issues.

 

Or is there anything that can be done to fix, workaround, or at least further troubleshoot?

 

Thanks,

Try using a different add-on controller if possible, the Asmedia controller would be the most likely culprit.

Ok, I'll have to source a add-on controller. The asmedia one is, as I'm sure you've guessed, is the one on the motherboard.

1 hour ago, adent042 said:

The asmedia one is, as I'm sure you've guessed, is the one on the motherboard.

Asmedia should be an add-on controller, very unlikely that the onboard Intel SATA controller is the problem.

I didn't install an add-on controller when I rebuilt the server.  But I'll have to check the specs on the mobo.

Ah, I think I know what it is. I’ll see if I can find a different add on controller. Or just retire the drives I was going to anyway, then move the remaining array drives to the intel controller ports. 

@adent042 Did you ever determine which disk this was from? I have the exact same issue, but have had it with two different HBAs. And it's the exact same count/sectors for me.

kernel: md: recovery thread: PQ corrected, sector=1574961048
kernel: md: recovery thread: PQ corrected, sector=1574961056
kernel: md: recovery thread: PQ corrected, sector=1574961064
kernel: md: recovery thread: PQ corrected, sector=1574961072
kernel: md: recovery thread: PQ corrected, sector=1574961080