Unraid Lock Up If Parity Check Is Run - Unraid 6.9.2

[Lime1028]

Hello, I have been having this issue recently and could use some help. I'm able to start up the machine and login through the web console to unraid, I was able to start the array and everything was working fine, but when I started a parity check it ran normally for a few seconds, got a couple gigs into the check and then stopped. The web console started running really slow, some tabs in it just wouldn't load or show anything, like the dashboard and docker tabs.

 

Pulling the syslog, or what I could get of it, I notice the following errors:

 

Jul 19 22:24:54  kernel: mdcmd (42): check 
Jul 19 22:24:54  kernel: md: recovery thread: check P Q ...
Jul 19 22:25:16  kernel: ------------[ cut here ]------------
Jul 19 22:25:16  kernel: kernel BUG at drivers/md/unraid.c:345!
Jul 19 22:25:16  kernel: invalid opcode: 0000 [#1] SMP NOPTI
Jul 19 22:25:16  kernel: CPU: 11 PID: 2308 Comm: mdrecoveryd Not tainted 5.10.28-Unraid #1

 

It looks like 20 seconds after the check started it crashed. I've attached the entire syslog file, crash happens near the end.

 

No idea what these errors mean or how to fix them. I tried updating my bios and unfortunately that hasn't helped.

tower-syslog-20210720-0530.zip

[trurl]

Go to Tools-Diagnostics and attach the complete Diagnostics ZIP file to your NEXT post in this thread 

[JorgeB]

Like mentioned you should post the full diags, also was it working before and it started doing this without any change, or did it start after some hardware change or Unraid update?

 

There have been other similar reports, with very different hardware, and since it's the Unraid driver crashing it might b a good idea to open a bug report.

[Lime1028]

8 minutes ago, trurl said:

Go to Tools-Diagnostics and attach the complete Diagnostics ZIP file to your NEXT post in this thread 

My apologies. It completely locked up when I tried to get the diagnostics. I restarted it and got them, though I haven't tried to run a parity check again, as it will probably crash it again.

tower-diagnostics-20210720-0952.zip

[Lime1028]

1 minute ago, JorgeB said:

Like mentioned you should post the full diags, also was it working before and it started doing this without any change, or did it start after some hardware change or Unraid update?

 

There have been other similar reports, with very different hardware, and since it's the Unraid driver crashing it might b a good idea to open a bug report.

I see two possible initiating factors. This started recently after an unfortunate event where I was transferring files to it from another computer over the network. That computer died (graphics card blew a shut resistor) while transferring files. The Unraid server was then taken offline to move it to gain access to some wires, and in the process I swapped the graphics card (I removed some ancient AMD card and swapped in a new Nvidia GT 710, because it's passive and the fan on the old thing was starting to make a ton of noise). Ever since then there has been errors. It's only ever on the parity check though. I don't know if it's because of the card (it's working fine for video output) or because of damage cause when that computer crashed during data transfer.

[JorgeB]

Different video card should not be an issue, could be hardware going bad.

[Lime1028]

15 minutes ago, JorgeB said:

Different video card should not be an issue, could be hardware going bad.

All the hardware was working fine before I last shut it down. I also see no signs of any issues when running health checks on the drives. Everything within the system is about a year and a half old. All the drives are proper NAS drives that should have years of life. If someone knows what they're looking for in the diagnostics maybe they can spot something, but to my untrained eye the hardware seems fine.

[JorgeB]

I don't see any other reason for the Unraid driver to start crashing with hardware that was working before, but if you want you can create a bug report, LT might be able to tell you more based on the error.