6.10.3 - unraid crashing since update and setup of Frigate w/ TPU passthrough

[daemian]

Hi all,

 

Hoping someone smarter than I can help me figure out whats going on. I have been running unraid for many years. I have been on the current hardware for over a year and its been very stable. Recently, I began experiencing crashes pretty regularly (uptime always less than 24 hours). This started after 2 changes:

 

1. I began to really use Frigate (docker). I had run a test setup w/o a TPU for a few weeks with just one camera for a few weeks without issue. However, once my Coral arrived I passed it through to my docker and added my 2 other cameras.

2. I did also upgrade to 6.10.3 at that same time.

 

Now I am experiencing the crashes. It's almost certain the Frigate setup or update is to blame as that is what changed when I started having issues. Hoping someone can look at the diagnostics and help me figure out what the specific issue is.

 

Thanks you in advanced

ur01-diagnostics-20220722-1359.zip

Edited July 22, 2022 by daemian

[JorgeB]

Enable the syslog server and post that after a crash.

[daemian]

ok I have enabled the syslog server. Thanks for coaching me along here. I will report back when i have data after a crash.

[daemian]

ok so the server has crashed a few times since the last post. I did try downgrading back to 6.9.2 - but the problem persisted. Attached is the syslog file from the latest crash time frame (Aug 2 around 11:30am).

 

TIA!

syslog-192.168.166.2 (2).log

[JorgeB]

Aug  2 07:49:22 ur01 kernel: macvlan_broadcast+0x10e/0x13c [macvlan]
Aug  2 07:49:22 ur01 kernel: macvlan_process_broadcast+0xf8/0x143 [macvlan]

 

Macvlan call traces are usually the result of having dockers with a custom IP address, upgrading to v6.10 and switching to ipvlan should fix it (Settings -> Docker Settings -> Docker custom network type -> ipvlan (advanced view must be enable, top right))

[daemian]

just to provide an update. I made the suggested change and so far its been almost 7 days without issue. I will post back if it occurs again - otherwise future readers can assume that did it. Thanks for your help Jorge!