5.10.21 21.02.3 becomes unresponsive after approximately a week

[SymbiosisSystems]

I'm running a Helios4 NAS with OMV 5.6.6-1 and ZFS 2.0.3. After running for approximately a week it becomes unresponsive - OMV login shows a blue screen but doesn't get to login page, SSH timeout and serial login just keeps going back to the login prompt without asking for the password. Repowering the unit (physically disconnecting the power because there is no way to shut it down gracefully) brings it back to life and it runs again for a week or so before the same thing happens again. I'm not quite sure where to start troubleshooting this. Any ideas ?

 

[stronthoop]

Hi SymbiosisSystem,

 

i've had a similar situation and reinstalled everything just to find out it was the browser type I was using.

 

Don't use Internet explorer, but Brave browser / google chrome. That should solve your issue.

 

The SSH timeout is weird.. Maybe you check dmesg log for what happens at the last line before you rebooted.

 

Just my 2 cents.

 

Kr,
Stront

[Mangix]

```

~> uptime
 17:09:17 up 19 days, 23:36,  1 user,  load average: 0.10, 0.10, 0.14

```

 

I use btrfs instead of OpenZFS.

[SymbiosisSystems]

Stront, the behaviour suggests high cpu usage or possibly lack of free memory; I can't login via ssh or local serial console because it times out. Unfortunately dmesg doesn't go back beyond the reboot.

 

Mangix, do you suspect it's related to OpenZFS ?

[Mangix]

On 5/13/2021 at 2:25 AM, SymbiosisSystems said:

Stront, the behaviour suggests high cpu usage or possibly lack of free memory; I can't login via ssh or local serial console because it times out. Unfortunately dmesg doesn't go back beyond the reboot.

 

Mangix, do you suspect it's related to OpenZFS ?

Wouldn't doubt it. OpenZFS is poorly supported with 32-bit kernels.

[SymbiosisSystems]

To post a positive update on this I've upgraded Armbian to 5.10.23 21.05.2 with OMV 5.6.8-1 and ZFS 2.0.3-1~bpo10+1 and it's now been running for 25 days without becoming unresponsive, memory usage continues to sit at around 33% and system load is low.