[6.12.1] macvlan call traces found

[ds123]

Hi,

 

After upgrading from 6.11.5 to 6.12.1 I received the following error from the Fix Common Problem plugin -

Quote

macvlan call traces found - 

For the most stable system, you should switch the network driver in Settings - Docker (With the service stopped) from macvlan to instead be ipvlan

 

I never had this error before 6.12.x. For now, the server is working properly so there is currently no impact.

Can someone please help in understanding this error? Should I change to ipvlan? What are the implications of doing this?

In terms of my docker network types - I created a custom network for most of my containers. In addition to that, I have one container that uses br0 (pi hole) with fixed ip address and another containers that use host network (plex and duckdns). Wanted to make sure nothing breaks if I change the docker custom network type to ipvlan.

 

Thanks,

Edited June 23, 2023 by ds123

[JorgeB]

1 minute ago, ds123 said:

Should I change to ipvlan?

Yes.

2 minutes ago, ds123 said:

What are the implications doing that?

Usually none.

[ds123]

8 minutes ago, JorgeB said:

Yes.

Usually none.

Thanks for the quick reponse.

Will I be able to keep the same network settings in each of my docker containers? including my user-defined network, br0 and host? Or should something change?

[JorgeB]

13 minutes ago, ds123 said:

Will I be able to keep the same network settings in each of my docker containers?

Should be.

[SShadow]

I had a similar call trace issue with 6.11.5 causing system lockups every 1 to 2 weeks setup as macvlan.  Switched to ipvlan with no other changes and everything worked the same for me.  I have custom networks with set ip address setup on all my docker containers.  Switching to ipvlan fixed the call trace issue and made my server stable again to run for an entire month until its scheduled reboot. 

[ds123]

Thanks. 

Changed to ipvlan, so far no issues.

[TheOgre]

Just wanting to tag onto this thread, even after switching to ipvlan, the plugin is still detecting macvlan traces. Any way to diagnose why or anything else I need to do?

[bonienl]

28 minutes ago, TheOgre said:

Just wanting to tag onto this thread, even after switching to ipvlan, the plugin is still detecting macvlan traces. Any way to diagnose why or anything else I need to do?

 

The plugin checks the content of the syslog, you would need to reboot your server to clear the reported occurrences.

 

[page3]

After upgrading to 6.12 I also started getting these issues, plus the gui becoming unavailable. Then the server crashing - first time in 9 years. 
 

Switched to ipvlan which made the server less responsive and started causing network issues.

 

Tried moving dockers to a second network card, but issues persisted. 
 

Rolled back to 6.11.5 and all is good again. I’ll wait a while before thinking of upgrading again. 

[3dee]

On 6/23/2023 at 10:41 PM, page3 said:

Tried moving dockers to a second network card, but issues persisted. 
 

Rolled back to 6.11.5 and all is good again. I’ll wait a while before thinking of upgrading again. 

 

Same for me, moving containers from br0 to br1 did not solve macvlan call traces. I'm back on 6.11.5 again.

 

Let me know when it's safe to upgrade to 6.12.x

[vw-kombi]

From what I have read, the ubiquiti unifi controller does not like all the machine switching caused by ipvlan due to the shared mac address for all the containers.  I have 17 containers, and the unify software shows me the real time upload/download of each, but as it is by mac address, this is not possible if I change to ipvlan - Personally, I have no issues with macvlan - never had a kernel panic or a failure to date, but I worry about upgrading from 6.11.5 and introducting these issue, forcing me to change and losing the below :

 

 

[netfox]

How can I Rolled back to 6.11.5?

 

And which is the best version to rollback to avoid this issue?

Edited July 12, 2023 by netfox

[Knutowskie]

I am on 6.11.5 and had the "macvlan traces found" in my fix common things plugin. So far nothing unstable. Machine rock solid.

So i decided to switch to ipvlan to resolve the "issue" and the journey began. With ipvlan my "proxynet" network (containing swag, nextcloud and collabora) stopped responding. No idea how and why. Just didnt work. I read something that it could be related to AVM Fritzboxes. So i am using a Fiber Firtzbox... nice.

After switching back to macvlan everything started acting wonky. docker service won't come up. So i tired to go for a safe reboot, but /mnt/cache was busy. hooray. its the docker.img that didnt let go. so i unmounted loop2 and then the machine restarted. 

If you are using a fritzbox you are fucked. Thats it. I am living in fear now as the macvlan traces occur. Someday they may become a problem. 

Everything worked fine. Why they do break things up and FOR MONTHS no fix? Any idea  how to work around this? No i can't buy another router.

Edited December 31, 2023 by Knutowskie

[JorgeB]

6 hours ago, Knutowskie said:

If you are using a fritzbox you are fucked.

With v6.12.4 you can use macvlan without the call traces, see the release notes.

[Knutowskie]

Okay i upgraded and made the changes recommended. We will see.