Support Secondary Reverse Proxy

[oschwartz10612]

[miloschwartz]

Point a Pangolin resource to another reverse proxy running on the same network as Newt

[the-potato-church]

Any ideas why this isn't currently possible? 🙏🏼

Edit: Is it becasue Newt doesn't currently support DNS lookup? (at least I am having trouble pointing it to my local DNS server (I have verified that /bin/sh nslookup <internal domain> from within the container returns the correct IP)

[miloschwartz]

@the-potato-church This is on the back burner for now. We need to find a way to fit it into the dev cycle. If anyone has experience and wants to tackle this in a PR, we'd be very grateful.

[the-potato-church]

@miloschwartz Do you have any pointers as to the root cause? Happy to attempt working on it myself. 🙏🏼

[oschwartz10612]

@the-potato-church I started looking into it on this branch which is outdated now. I think the issue was something Traefik related but I honestly do not remember what it was. Feel free to give it a shot if you would like!

[dguihal]

Weird .... In fact it does .... I mean I managed to do this

Traefik -> Gerbil -> Newt (hosted on kube) -> Traefik (hosted on kube) -> Service (hosted on kube)

The point is that I had to create a dedicated ingress on the traefik on kube as the initial hostname is preserved all along, The hostname override should solve this, but for some reason I am unable to make it work.

I only had to point to the traefik service name as target for pangolin service.

[ryanehamil]

I had run into a similar issue when layering two traefiks. There may be multiple solutions but mine involved using proxypass. This might get more difficult if there are two points of tls decryption.

Mine was just using the first traefik as an SNI relay with no TLS termination. Then the second layer was the TLS endpoint. I'll take a peek at that branch and see if it might be related.