TLDR: Starting array I have internet connection, can ping sites from CLI and have remote access via wireguard. When I start docker I get between 2-20minutes before it looses connection. Tried starting one docker at a time looking for patterns but cant nail it down to anyone thing. 

At a loss as to what to do.


unraid Version: 6.11.5

So everything was fine. But, now things are borked and driving me crazy. The only changes that occurred with my unraid setup is that I removed HomeAssistant OS VM and disabled VMs in settings. As for my network, I added a few IOT devices to my network and moved HAOS to a new piece of hardware that I added to the network. That's the only changes.

 

Now unraid looses Internet access/connection shortly after starting array and dockers start. LAN will continue to work, I can access shares and web GUI locally. But connection to the internet and external access drops. Cant even access with Tailscale or Wireguard. No network configs changed on uraid or router. All I did on unraid was stop the VM and disable VMS in settings. The only change on the router was to set static ip for the new HAOS box.

 

When it happens and I attempt to check for docker updates in the web UI will give me an error about DNS and unable to connect and there's no access to my reverse proxies or cloudflare tunnel since it can't connect to internet.

 

Quote

Community Applications requires your server to have internet access. The most common cause of this failure is a failure to resolve DNS addresses. You can try and reset your modem and router to fix this issue, or set static DNS addresses (Settings - Network Settings) of 208.67.222.222 and 208.67.220.220 and try again.

Not sure what those DNS IPs are. Of course changing DNS in network settings means stopping all dockers restarting them and things work but then it craps out again after 5-20min. Later, CA Fix Common Problems plug end says change DNS to Google DNS 8.8.8.8/8.8.4.4
 

I have 4 docker containers that I use externally using reverse proxies or Cloudflare tunnel:

- Jellyfin (reverse proxy with SWAG) jellyfin.mydomain.com

- Nextcloud (reverse proxy with SWAG) nextcloud.mydomain.com

- FileRun (Cloudflare tunnel) filrun.mydomain.com

- Filebrowser (Cloudflare tunnel) filebrowser.mydomain.com

Also have Tailscale and Wireguard setup.

Looking at the unraid log, there's no errors or warnings but I do see this at times: (TOWER is unraid)

Jul 15 18:09:09 Tower  nmbd[6696]: [2023/07/15 18:09:09.966523,  0] ../../source3/nmbd/nmbd_become_lmb.c:398(become_local_master_stage2)
Jul 15 18:09:09 Tower  nmbd[6696]:   *****
Jul 15 18:09:09 Tower  nmbd[6696]:   
Jul 15 18:09:09 Tower  nmbd[6696]:   Samba name server TOWER is now a local master browser for workgroup WORKGROUP on subnet 172.17.0.1
Jul 15 18:09:09 Tower  nmbd[6696]:   
Jul 15 18:09:09 Tower  nmbd[6696]:   *****
Jul 15 18:09:13 Tower kernel:

Everything has been rebooted/restarted (router, switch, unraid). I moved DHCP over to my pi-hole from the router thinking there was an issue with my router but that didn't help.

 

Is my router crapping out for too many devices? I have no idea what's causing this issue. The unRAID box is the only thing on the network with issues now. Unraid and the HAOS box and a few other hardwired devices all have static IPs. There's no IP conflicts.

 

Network setup:

- Router (tp-link archer A7) + Switch (unmanaged): 192.168.1.1 - router is DHCP server, DNS servers = 1.1.1.1, 8.8.8.8

- Raspberry pi running pihole: 192.168.1.100 (primary DNS server for LAN)

- unraid (static ip) - new HA box (static ip) - 3x computers, ~15 IOT devices (tvs, plugs, chromecasts, etc...)

Also still curious about the message spamming my logs link to that post here

 

tower-diagnostics-20230716-1236.zip

Edited July 17, 20232 yr by in_trauma
Revised info and title, uploaded new diagnostics

UPDATE: After more trial an error with trying to nail down the problem I found that it was definitely docker related. It had to do with Plex having a host network.
 

So I did the following:
Settings > Docker > Host access to custom networks: > set to "Disabled" (was enabled previously)

Stop Plex from auto starting and started all other dockers. DuckDNS and Tailscale still both have host networks. I set the unraid box to ping cloudflare 1.1.1.1, 8hr later and it and all was well.  Then started Plex container. 4hr later all is well.

So, for the time being it seems I've resolved the issue. Now I would just like to learn why this happened. I changed nothing and all of sudden Plex broke things, did a container update change some fundamental setting for communication in the container?

Two posts that helps point me in the right direction: