Hello All,

My unRaid server got hung up recently. I was not able to ping, access via the its host name / IP address or SSH into it. I know that it was being stressed out quite a bit and the SSD was pretty hot (around 118F) right before it went unresponsive. I did a hard reset and then the Fix Common Problems plugin alerted me of the Call Traces and the unexpected restart.

 

I had this happen to me before and I believe it was one of my dockers that needed re-installed. I re-installed delugevpn as I thought it had to do with openvpn but I am still getting call traces. I would be grateful if someone could steer me in the right direction.

 

Please let me know if I am missing any important information here. I have attached a diagnostic file here.

 

Thanks for your time,

Tyler

acsfs-diagnostics-20171217-0612.zip

Call traces are from the network bond, disable it if not in use, disable and re-enable if in use to see if it stops.

Okay so I stopped all VMs, dockers and I stopped the array. I then turned off bonding and turned it back on then I rebooted.

 

I am still getting call trace errors. I have attached another diagnostic file. Is there anything else that could be causing issues? The error keeps repeating that openvpn is tainted.

acsfs-diagnostics-20171217-1117.zip

Quick google search

 

What is the output of this command

ethtool -k eth0

This is what I got.

 

ethtool.txt

Shot in the dark based upon a google search.  Try these commands

ethtool -K eth0 lro off
ethtool -K eth0 gro off
ethtool -K eth1 lro off
ethtool -K eth1 gro off

And then monitor your system to see if the traces continue.  If everything seems ok and the traces stop, then add those commands as a user script set to run at first array start only via the user scripts plugin.

 

https://access.redhat.com/solutions/654283

Quote

 It is incorrect to enable LRO when IP forwarding and/or bridging are in use.

 

Edited December 17, 20178 yr by Squid

I yesterday also collided with call traces, I can share observations! 16 hours everything was fine, but recently today, after the evening torrenting, i get a full log of call-traces.

Call-traces start right after the torrent starts! Stopping transmission container stops call-traces. Disabling of the bond did not help.  In the end, I switched the transmission docker container to the host mode and now no call traces while torrenting!

in this log i start torrenting in host mode with no call-traces then switch container to bridge mode and recieve full log of call-traces.....

sorry for my google-translator english =)

 

unraid-diagnostics-20171217-1934.zip

3 minutes ago, vanes said:

Stopping transmission container stops call-traces. Disabling of the bond did not help.  In the end, I switched the transmission docker container to the host mode and now no call traces while torrenting!

Then we're on the right track.  Perhaps @bonienl might have some suggestions as he designed the whole network thing on unRaid.

I am going to include a little more details here as I do not know if it relates to the problem at all.

When I first upgraded to 6.3.5 I was getting some call trace errors, problems pinning CPU threads (see Other Post), but never any freezing. I then upgraded to the pre-release (9f) and this seemed to solve all my problems. Until this pre-release. A little bit after upgrading to this one, I'd say about 3-4 hours after, my system froze.
 

 

Do you think it would help if I just disabled the bonding? Do you think I take advantage of that realistically? I am going to try turning that off.

After looking at your replies, here is my network routing table. Is anything wrong here?

@Tbigs2011 - Is OpenVPN running in Host Mode or in Bridge Mode?  And, is it up to date?

I never set up open VPN on unRaid itself that I know of. Is this something that needs set up? I am using a VPN for a deluge docker and I have a VM that runs an openvpn server on top of ubuntu.

BTW: I ran the commands you suggested but no dice. Still getting the call traces.

Edited December 17, 20178 yr by Tbigs2011

DelugeVPN.  Going from @vanes post, and open VPN was referenced in the traces

 

That is correct.

 

I tried running the docker with a networking mode of host but it will not start afterwards. I get this from the log:

 

2017-12-17 12:16:52.014001 [crit] Network type detected as 'Host', this will cause major issues, please stop the container and switch back to 'Bridge' mode

 

7 minutes ago, Squid said:

DelugeVPN.  Going from @vanes post, and open VPN was referenced in the traces

 

i dont use deluge vpn or open vpn.  I use transmission/linuxservers (without vpn) docker container in my case it was Transmission....

Edited December 17, 20178 yr by vanes

Then maybe we're having a problem with torrents running under 6.4 in a container.  Do the traces stop if you disable DelugeVPN

Yes they do stop if I disable delugevpn . The problem is that I route the traffic from the other dockers to the delugevpn using privoxy that is baked in.

4 minutes ago, Tbigs2011 said:

Yes they do stop if I disable delugevpn

thoughts @binhex?

You would think more people would be having an issue with this though?  I can't be the only one that runs dockers like this, right?

1 minute ago, Tbigs2011 said:

You would think more people would be having an issue with this though?  I can't be the only one that runs dockers like this, right?

IDK.  I like many people stopped using torrents a long time ago and switched to usenet.  But you gotta start somewhere.  As it stands, the problems disappear when stopping an application (or switching the network mode) that uses Torrents.

i am on rc16b

Sorry for reporting the issue in the incorrect spot. Is there a way for me to move the thread to the Pre-Release section?

Anyways, thanks for the excellent help Squid! You're awesome! I'll be shooting you a donation.

 

I changed the network type to BR0 and it started without issues. I no longer have the call traces either.

Edited December 17, 20178 yr by Tbigs2011

I spoke too soon. I still have the trace calls.

acsfs-diagnostics-20171217-1301.zip

What exactly is a trace call? I did a quick google search but I still don't understand. Will it cause a real problem? Besides it freezing the other day I have not had problems. My dockers are connected to the internet and are able to download torrents.

Then as the last thing I can suggest is to run those commands against the bond adapters

ethtool -K bond0 lro off
ethtool -K bond0 gro off
ethtool -K bond1 lro off
ethtool -K bond1 gro off

 

 

2 minutes ago, Tbigs2011 said:

What exactly is a trace call? I did a quick google search but I still don't understand. Will it cause a real problem? Besides it freezing the other day I have not had problems. My dockers are connected to the internet and are able to download torrents.

Its kinda like a recoverable blue screen of death.  (kinda).  Problem here is that as you're downloading, your syslog is going to grow very quickly which will fill the available space allocated for it and cause other problems.  If it was a one time thing I wouldn't care about it.