- Urgent
So I just updated to Unraid 7 from 6.12.11 and it seems to have broken my subnet routing with tailscale. It seems I can still ping the devices across the tailscale subnet route, but can't access any services there like I was in 6.12.
2 things I have noticed that cause this problem. If I disable the docker service and/or disable "Host access to custom networks" the subnets are still accessible but then host access is lost to dockers (which is a problem for me). Secondarily, If i leave everything as is with docker, and then start ANY VM with networking, the problem is resolved? For example. All VM's stopped, and docker service running (even with all docker containers stopped), results in no access to other subnets except for pings. If I start up a Windows VM, then everything functions as it did in 6.12.11. If I stop the VM, the issue re-occurs immediately.
Nothing changed settings wise for me since upgrading, this issue happened right away, and rebooting and changing settings doesn't seem to help. I have checked IP routes and they are the exact same as they were before. I have another unraid server that is more critical still on 6.12.11 and the configuration there, being the same on this server, works just fine, so the issue is linked to some breaking changes in 7.0.0. I am unable to recreate this bug in 6.12.11. I would like to note I am making no use of the new tailscale docker integrations and have been using the Tailscale plugin. I have checked that all plugins are up to date. I have also tried changing route metrics, and while changing the metric of shim-br0 can bring it back up, that consequently brings back down subnet access.
Tailscale logs show no error, just reconfiguration when a new interface is brought up (either with docker coming up, or the vnet adapter with VM's). I have also tried enabling and disabling settings in the tailscale config such as ip forwarding, use tailscale subnets, use tailscale DNS, and none of those seem to make any difference.
I am using docker custom network br0 for my containers to assign them IP's in my subnet. These static IP's are reserved for these containers so no IP overlaps are occurring. I am also not using the docker DHCP service. The subnet CIDR I am using is 10.2.0.0/24.
If anyone has experienced this or has any insight to what is happening/diagnostics to try, please let me know. I will upload a diagnostics file if necessary. My setup is fairly simple and is just a tailscale link between my primary server in a different location on 10.0.0.0/24 and my subnet here at 10.2.0.0/24. No otherwise advanced configuration. Thanks!
-
1
