FreeNAS locked up

[tallen116]

I was wondering if anyone has experienced FreeNAS locking up running 11.2 RELEASE?

When it happens, the NFS shares stop working and the GUI becomes unresponsive. I can still access SSH and the ZFS pool status is healthy. These have been in production for about 2 years and no issues until upgraded to 11.2 RELEASE.

 

[l@e]

What hardware?
Can you check services from the ssh if they still running?
Inspect also log files if you can catch any error there.

 

[tallen116]

The hardware is a Supermicro SSG-2028R-ACR24L. I am running FreeNAS on ESXi with PCI passthrough for the LSI 3008 controllers. The FreeNAS VM has 8 cores and 192 GB of RAM.

I checked some log files such as /var/log/messages and dmesg. There wasn't any logs stating there was an issue. The nfsd service was running. I end up restarting the server to bring it back to being functional. It usually has to be cold cycled because it won't fully reboot properly.

It has happened 3 times in the past 60 days since updating to 11.2 RELEASE. Is there any other logs that would produce debug or more verbosity?

 

[Meyers]

I was having similar pool lockups on my FreeNAS mini system. No logs whatsoever. I think it was due to overheating but I can't be sure. All I know is I cranked the fan up to full on and haven't had a problem since.

Can you correlate the lockups with scrubs or anything like that?

I've had problems with a couple systems since upgrading to 11.2. I had to revert to 11.1 on one system because it's unusable otherwise. I would probably revert to 11.1 if I were you.

 

[tallen116]

I don't think it is related to heat. I have another array that is the same exact hardware in the same rack that is completely fine. The only difference is the problem array has more workloads writing data to it via NFS. I hasn't happened again, but I added some NFS Tuneables to the system to see if it makes any difference.

 

[Meyers]

I don't think it is related to heat. I have another array that is the same exact hardware in the same rack that is completely fine. The only difference is the problem array has more workloads writing data to it via NFS. I hasn't happened again, but I added some NFS Tuneables to the system to see if it makes any difference.

Yeah I figured it probably wasn't the same thing outside of the fact that my pool would also just completely freeze up with no error logs anywhere. I'm skeptical about it being heat, but so far I haven't had the pool freeze up.

Out of curiosity, what NFS tunables did you add?