Hey everyone,

 

Yesterday my Unraid had disabled a disk because of read/write errors, I guess.

 

I wasn't able to perform a SMART self-test afterwards either until after a reboot.

 

The SMART tests I performed on the disks turned out fine.

 

I'm now rebuilding the disk onto itself.

 

I think the disk is fine but I'm not sure what went wrong yesterday.

 

I've attached the syslog from the 'event'.

 

Perhaps someone here can shed some light on a possible cause.

 

The system has been running stable since I built it in June.

 

Thanks for looking.

 

J.

syslog.txt

It would help to have the full diagnostics.  I've had similar problems, though, where a disk is dropped from my Marvell based controller (SAS2LP in my case).  If you have good Smart data following the reboot then a rebuild is appropriate.  It may also be necessary to check cables and evaluate your power supply if problems continue.

Have to echo what tdallen said, and say it's hard to do good troubleshooting with only an excerpt from the syslog.  The diagnostics would help identify the drive much better, and we'd be sure we had all the info about the issue.

 

If I assume the drive in question is associated with ata14, and with the SCSI channel 11:0:7:0 (which means it's on the 8th port of a SAS card using mvsas driver), and it's Disk 9, *and* if we assume this is the very first message related to this problem, then it looks like either the 8th SATA port crashed somehow, or Disk 9 became partially disconnected, then later wholly disconnected, causing somehow a mvsas driver crash, and causing the kernel to drop the drive as inaccessible.  You can ignore all of the read and write errors, as they happened after the drive was lost, so there is probably nothing at all wrong with the drive.  Once you reconnect and reboot, it may appear to be fine again, but since writes were lost, the drive still has to be rebuilt, to sync with the emulated version of Disk 9, regain the missing writes.

 

In addition to tdallen's suggestions, I'd look for a firmware update for the SAS card.

Hi,

 

I've attached the full diagnostics zip.

 

After it had crashed I did disconnect all the sata cables from the Supermicro 5-in-3 enclosure and reconnected them.

They're all locking sata cables though and they all seemed to still be connected properly.

 

In the mean time the disk's been rebuilt successfully.

 

J.

tower-diagnostics-20161211-1233.zip

This looks like the diagnostics from after the failure, but before the reboot (which is good set of diagnostics to have).  Disk 9 is still showing a bad smart report with "bad response to IEC mode page".

 

Do you have a fresh diagnostics from after the reboot?

This looks like the diagnostics from after the failure, but before the reboot (which is good set of diagnostics to have).  Disk 9 is still showing a bad smart report with "bad response to IEC mode page".

 

Do you have a fresh diagnostics from after the reboot?

 

Hi tdallen,

 

That was indeed the diagnostics from before the reboot.

 

Here's one I generated just now.

 

J.

tower-diagnostics-20161216-1711.zip

Funny thing, this problem.

 

Disk 9 got disabled again this past night. Wasn't just a one-off then.

 

I'm now rebuilding onto a new disk, although the disk doesn't seem to be the problem judging from the SMART reports.

Figured it can't hurt to try though.

 

The only things that have changed on my system recently were the addition of disk 9 and an upgrade of Unraid from version 6.2.3 to version 6.2.4.

 

I've attached the diagnostics again in case anyone feels like checking it.

 

Let's see what happens with this new disk.

tower-diagnostics-20161218-0730.zip

OK, same thing happened again this past night.

 

So, its' definitely not the disk (unless there's some incompatibility with this type as they're the same).

 

I've now hooked up one of these disks to the same sata cable but outside the Supermicro 5-in-3 unit, trying to exclude possible problems one at a time.

 

J.

 

OK, same thing happened again this past night.

 

So, its' definitely not the disk (unless there's some incompatibility with this type as they're the same).

 

I've now hooked up one of these disks to the same sata cable but outside the Supermicro 5-in-3 unit, trying to exclude possible problems one at a time.

 

J.

 

You mentioned a locking cable earlier.  I should point out that one of the manufacturers (WD ,as I recall) changed the design of the SATA data connector on his hard drives so that most locking connectors not longer provide a secure connection at the drive end.  (Should not be an issue with the drives in the cages but could be with a direct cable connection.)

 

For reason, see here:    https://community.wd.com/t/why-would-wd-design-this/17650

 

 

I haven't had any issues anymore since I hooked up the disk outside of the Supermicro 1-in-3  unit so it seems there's something wrong there.

A bad sata connector or something with the backplane.

 

Anyway, the problem is resolved.

 

Happy new year everyone!