Dockers can't talk anymore

[tucansam]

Since I can't put this in the docker container forum, I'll try here.

 

I have been running a half dozen or so docker containers for years, and haven't touched any configs in at least 18 months, other than the containers updating when I see new versions available.

 

All dockers were running in "br1" mode, with an individual IP address.  Sonarr, Radarr, and Headphones could all talk to Sab, and I could use their individual IP addresses:port to manage each piece.

 

Last week I tried to log into the web GUI of Sonarr, and it hung, its log showing DNS resolution errors and "no route to host" errors.  After a week of messing around, I installed a new container of Sonarr written by a different author, with the intent of getting it working and blowing the old one up completely.

 

That docker container cannot talk to Sab.  

 

The original (hung) Sonarr now works, but only if I run it in bridge mode, and use <unraid IP>:8989.  However, in this configuration, it cannot talk to Sab.

 

In br1 configuration, it hangs, and cannot talk to Sab.

 

With Sab in bridge mode, I cannot bring up the GUI.  It cannot talk to Sab.  Using br1, I can log into Sab's GUI.

 

Radarr and Headphones can both talk to Sab when it is run br1.

 

I have looked at my network closely, and there are no duplicate IP addresses that could be interfering.  All dockers' IPs are statically assigned, outside of the range of my DHCP server.  No new hosts have appeared on my network (neighbors, etc).

 

This is 100% a Sonarr problem and is now affecting two different containers by two different authors, one of which is a brand new vanilla config.

 

I am banging my head against the wall trying to figure out what is going on here.  The last TV show that was downloaded was 3/25 (sonarr), but Headphones downloaded an album on 4/1..... So something somewhere is working.  

 

Totally out of ideas here.  Any suggestions?  This literally worked flawlessly for years, and has been 100% hands off until last week.

 

 

[tucansam]

I did some more digging, and after upgrading from unraid 5.x to 6.x ages ago, one of the first things I noticed is that an interface called "br1" was created, and I could no longer work with eth0 or eth1.  Here is the current network config, and I don't remember changing or adding any of what is listed, only that I went from using one of two available interfaces on 5.x (eth0) to wanting to be able to do the same on 6.x, but being unable to do so:

 

 

 

Here is the docker network config with the docker stopped, and i know for a fact I made zero modifications here because I didn't even this this page exited until  a forum search ten minutes ago:

 

 

 

I don't know when or where along the way this bridged interface got created, but I don't like it, because it is making thing unnecessarily complicated, I think.  I want to simply use eth1 for everything under the sun with this server, and I'm not sure how to make this happen.  Eth0 is onboard (Realtek) and Eth1 is an Intel PCIe card.

 

Assuming its relatively pain-free to take the entire system, dockers and all, back to a super basic simple network config, then I wonder about the ramifications to the individual dockers, ie, can I continue to have them each have their own IP address so I can manage them as individual machines on my network?  

 

And this still doesn't explain the sonarr/sab problem presently plaguing me.

 

Edited April 7, 2020 by tucansam