Help! Not sure where to start troubleshooting BTRFS error - XFS (md2): Metadata corruption detected at xfs_dinode

[civic95man]

On 8/2/2020 at 3:11 PM, mc_866 said:

This docker is running with a defined static IP

From my experience, this is the cause of the call trace. 

On 8/2/2020 at 3:11 PM, mc_866 said:

Should I not run this with a static IP?

My understanding is that if you don't have a static IP for your unifi-controller, you're going to have a hard time.  I had issues setting it and my network up until I gave the controller a static IP.  Same can be said for Plex.

On 8/2/2020 at 4:42 PM, mc_866 said:

Nothing to be concerned with for those irq traces then?

Some say that those call traces are harmless, others have had issues when they occur, such as system instability or lockups.  Best to not risk it and just get those fixed.  My solution to this was to isolate all of my dockers to a separate vlan.  I *believe* those call traces are caused by broadcast packets which macvlan doesn't handle well but unifi doesn't seem to route them between vlans. This allowed me to keep the static IPs.

 

As for the btrfs corruption, there may be a correlation between it and the call traces, or it could be something else unrelated.

 

Good luck!

[mc_866]

18 minutes ago, civic95man said:

From my experience, this is the cause of the call trace. 

My understanding is that if you don't have a static IP for your unifi-controller, you're going to have a hard time.  I had issues setting it and my network up until I gave the controller a static IP.  Same can be said for Plex.

Some say that those call traces are harmless, others have had issues when they occur, such as system instability or lockups.  Best to not risk it and just get those fixed.  My solution to this was to isolate all of my dockers to a separate vlan.  I *believe* those call traces are caused by broadcast packets which macvlan doesn't handle well but unifi doesn't seem to route them between vlans. This allowed me to keep the static IPs.

 

As for the btrfs corruption, there may be a correlation between it and the call traces, or it could be something else unrelated.

 

Good luck!

Thank you!

 

So right now I have a static IP defined for my Unraid server which was defined in my PFSense setup.

 

The unifi-controller has a separate static IP also defined from PFSense

 

Is your thought to keep the static IP for unifi-controller, or abandon that and see if the trouble continues?
 

Edited August 5, 2020 by mc_866

[civic95man]

7 minutes ago, mc_866 said:

Is your thought to keep the static IP for unifi-controller, or abandon that and see if the trouble continues?

If you don't feel like messing with vlans right now and just want to test things out, you can try to disable the static IPs for just the dockers.  It's best for unraid itself to keep a static IP and shouldn't be related to your call traces.

 

is your pfsense setup on a vm?

[mc_866]

No, separate discrete PFSense box running connectivity and DHCP for me.

 

Unraid install separate running storage and dockers.

[civic95man]

Alright, well keep unraid on it's own static IP, use automatic ip assignment on dockers (all), and retest the stability to see what happens. 

 

Do you have log mirroring enabled or at least a monitor attached?

[mc_866]

I setup syslog -> mirror to flash

 

I'm still working to see if I can grab the most recent log. Which folder would that be in?

 

I removed the static IP and restarted the docker, now it's handing out an IP outside of DHCP that's tied to one of my AP's so that won't work.

 

Is there a way to force a different IP without defining a static?

EDIT, changed network to bridge, think that's it.

 

Edited August 5, 2020 by mc_866

[civic95man]

31 minutes ago, mc_866 said:

I'm still working to see if I can grab the most recent log. Which folder would that be in?

they should be under "logs" on your flash drive

 

41 minutes ago, mc_866 said:

EDIT, changed network to bridge, think that's it.

yes, that should work. Do note that docker assigns IP addresses itself to the container, not from your network DHCP server. So you may want to setup a valid reserve pool on your DHCP/PFSense box and assign that to the network bridge in docker.  That way you won't have address conflicts

[mc_866]

Found the logs but nothing listed since 7-29

 

[civic95man]

verify "Settings" --> "Syslog Server" --> "Mirror syslog to flash" is set to "yes" so we can hopefully capture what happens if the system locks up again

[mc_866]

28 minutes ago, civic95man said:

verify "Settings" --> "Syslog Server" --> "Mirror syslog to flash" is set to "yes" so we can hopefully capture what happens if the system locks up again

Confirmed.

 

But no logs since 7-30. That was in /boot/logs using MC. Is there another location I should be looking?

[civic95man]

Thats the correct place to look. Strange. You could try to delete both rsyslog.cfg and rsyslog.conf in the config directory on the flash drive and set up the "mirror syslog to flash" again.  That will clear the syslog server settings and allow you to start over fresh.

 

I guess you could also try to setup a remote syslog server on another computer and have unraid send it there. kiwi syslog server seems to be a popular choice. 

[mc_866]

3 hours ago, civic95man said:

Thats the correct place to look. Strange. You could try to delete both rsyslog.cfg and rsyslog.conf in the config directory on the flash drive and set up the "mirror syslog to flash" again.  That will clear the syslog server settings and allow you to start over fresh.

 

 

I deleted those two files manually and set mirror syslog to flash again.

 

I'll have to look into the remote option. Haven't had time to do that just yet.

[mc_866]

Stability seems to be back in play since getting rid of the static IP on my Unfi container.
 

Hopefully the stability will continue! Thanks for the help!!!!