[6.11.0-rc5] Host to custom docker macvlan network communication no longer works.

Only thing that comes to mind is the ARP table within your switch needed a flush.  

 

May want to label your issue solved.

The original issue is not fixed tho, DHCP/IP problems were just a 'side quest'. After getting that sorted, I'm still not able to ping my home assitant in a custom docker macvlan network, even tho 'Host access to custom networks' is enabled.

 

After confirming everything DHCP/IP related works and home assistant is still not reachable on rc5, I've gone back to rc4 and home assistant docker is reachable from host, so it must be something to do with the upgrade.

I have confirmed the same issue on my second Unraid server that runs rc5. It went unnoticed there, because only secondary DNS server has its own macvlan IP, so it's not as impacting as home-assistant.

I'm having almost the exact same problem with this build. Looking for a permanent solution too.

It also happened to me with rc4 but I was able to get things working again by uninstalling and reinstalling the NPM docker, but I also restarted both the router and switch, not sure I've done that this time... 

Any resolution welcomed.. 

### UPDATE ###

Ok, so it was obvious once it finally clicked (after rolling back and restarting the router and switches did nothing) 
I shut down docker.  Turned the host access to NO, and applied, then set to YES and re-applied.... started the services, and all is well. 
I TOTALLY forgot that sometimes settings get out of whack with the GUI... it's happened with other network settings to me... it was just a while ago and I fortunately don't have these kinds of issues often... 
I would love to go back to RC5 again and see if all is well, but I can't afford the time today, and there will prolly be a RC6 soon enough...  Seems like somehow this gets hosed for some people on the RC5 update..  and I am super curious if toggling that setting is the fix there too, but I am operational again and have to get back to other stuff...  

Thanks to everyone (here and in the Unraid Community Discord) 

Edited September 22, 2022 by Garbonzo

Having same problem, I use pi-hole and pipe my other dockers through it. After RC5 none of my dockers can use DNS hosted by pihole. Removing the custom DNS config from the container restores connectivity. Workaround above to toggle the setting and restart Docker did not help.

 

update: on 6.11 stable problem still there. Toggling the host setting wasn't working however with docker off, I changed macvlan to ipvlan, saved and then change ipvlan back to macvlan and saved again then started docker. It seems to be working again.

 

Definitely some weird config issue introduced with rc5

Edited September 23, 2022 by nickp85

I'm seeing this issue as well. If tried toggling off and back on, starting docker between changes and it still doesn't work. Toggling worked for me prior to this release when the setting stopped working after hard reboots/crashes.

Yea, toggling worked for me before too, but doesn't anymore.

I still had luck, but had to toggle again after a reboot (this is after rolling back to rc4) 
I am hesitant to go back to rc5 again after seeing others saying toggling it doesn't fix for them. 
I thought I had read 6.11 was out of beta, but I may have misread that... anyway, I am probably going to sit revisiting rc5 out this time, and try again if there is a rc6+ prior to release. 

1 hour ago, Garbonzo said:

I thought I had read 6.11 was out of beta, but I may have misread that... anyway, I am probably going to sit revisiting rc5 out this time, and try again if there is a rc6+ prior to release. 

https://unraid.net/blog/6-11-stable

Anyone wants to be a guinea pig to test that out?

Okay, I've installed 6.11 stable on my secondary server and did the 'turn off docker -> change setting -> change it back -> turn on docker' dance. Pi-hole docker is able to ping server IP and server is able to ping pi-hole, so it looks like it's fixed in stable 6.11. I'm not going to do an update on my main today, might try tomorrow.

1 hour ago, Xenor said:

Okay, I've installed 6.11 stable on my secondary server and did the 'turn off docker -> change setting -> change it back -> turn on docker' dance. Pi-hole docker is able to ping server IP and server is able to ping pi-hole, so it looks like it's fixed in stable 6.11. I'm not going to do an update on my main today, might try tomorrow.

I am planning on doing the stable tomorrow as well, I have to power down for other things anyway... I was still in the next release, thats why I wasn't seeing the stable... I was "almost" sure I was mistaken about the stable being out...

 

+1

 

To this issue. I am on 6.11.5 and I used to do the Toggle "Host access to custom networks" to disabled, start docker, stop docker, and then set "Host access to custom networks" to enabled and start docker again. This used to fix it Pre- 6.11.5

 

Now going forward on the latest release I can't get connectivity between my host and the docker containers on the custom br0 network.

 

Any help would be appreciated, I've tried a full restart, resetting router, and multiple iterations of turning docker on/off.