search

getmeli / meli

Platform for deploying static sites and frontend applications easily. Automatic SSL, deploy previews, reverse proxy, and more.
Other
2.4k stars 97 forks source link

Configuration Issues #242

Closed mtiller closed 2 years ago

mtiller commented 2 years ago

I attempted to create a helm chart wrapping Mongo and Meli. The Pods come up just fine and the ingresses and services all seem properly configured but I don't understand how things should map. I don't need SSL, so I'm only targeting ports 80, 2019 and 3001. But what I don't understand is if I intend to run this on sites.example.org, how should this all be routed? What should sites.example.org route to? (which port, do I need to rewrite the HOST header?)

What should <something>.sites.example.org route to? (again, which port, etc)

I looked in deployment/k8s at the Kubernetes manifest there. But there is no Ingress definition there, which leaves me guessing.

My guess is that sites.examples.org should route to port 80 and keep the Host header of sites.examples.org and that my-site.sites.examples.org should route to port 80 and keep the Host header of my-site.sites.examples.org. But what about ports 2019 and 3001? Do they need to be exposed via the Ingress? Does the client assume those ports or does it assume they get remapped somehow? Or are ports 2019 and 3001 only used internally with no need for an Ingress?

Thanks

gempain commented 2 years ago

@mtiller hi ! Your first guess is the right one: you should be using port 80, which is the only one exposed by our Docker image. Port 2019 is for Caddy's API and 3000 (not sure it's 3001) is for the Nodejs backend, but those are internal to the container.

mtiller commented 2 years ago

Everything seems to be working with my helm configuration. Thanks.