Have a fairly complex Traefik file provider deployment, and I've been working with the allow lists and the like, also the cloudflarewarp plugin, and while trying to get IPv6 to work suddenly external couldn't load but internal could, and in a quite odd realization (went back to Nginx, sites worked again, went back to Traefik and they didn't...) I clobbered the whole thing, made a "basic" version with almost no extra options/middleware and deployed, and it worked.
Now that I had actually deleted/removed the container and I re-deployed my complex setup again, it worked!
As near as I can tell it was the 'defaultAllowList' getting in the way of allowing external loads (I had disabled it towards the end of testing... but had not fully re-deployed, just restarted), but at the moment I have other things to work on. Just wanted to put this out there in case anyone else runs into this and can only use their Traefik locally, but not via the portforward that should allow the external internet to also resolve/use the proxied web pages.
I can't say our problem is the same but we're having similar issues. Traefik works fine, it routes to correct ports, until it doesn't... Restarting doesn't work, we have to delete and rebuild the containers, that's how it begins to work. Just wanted to put it here as well, because it might be a bug.
1 Like
Agreed, the 'watch'/dynamic nature (file or docker) can work, but it does not work all the time for reasons I have not explored.
Full re-deploy and/or full container restart can often be the solution/realization of your change or update.
We only seem to experience some odd behavior with service/container discovery in a test cluster using a Docker socket proxy. Do you use Docker socket directly?
Yes, the snippet is above.
In one container/cloudflareproxy traefik v3, same as @MelihDarcanxyz but in my other container/externalproxy travefik v3 I'm using a MACVLAN that is directly attached to an independent NIC with a subnet specified
If I change/enable/disable a plug-in, or sometimes reconfigure a route/service, I must restart the container to use the re-configuration for either docker+sock/file+macvlan deployment
Just wanted to comment that I've noticed the file provider is reloading automatically now, and I had to move the web servers that the Traefik that was docker provider setup to connect to, to another host, so it is now running file provider too.
It would likely be an impossibly jump to 'watch the docker on another host' but, if you ever explore it, I'm a use case that would at least be interested in trying it.
That said IPv6 is harshly dynamic (you can set up a MACVLAN and on each Stack/Container configure its Networks MAC/IPv6/IPv4 and STILL your container when restarted will suddenly have an extra MAC and a 2nd or 3rd IPv6 that you never assigned) so good luck.