Open maxirus opened 2 years ago
It's necessary because it reads from some shared volumes. It might be possible to co-locate the front and gateway in one pod (or even container), or use a more sophisticated ingress not to not proxy the front in the gateway, but this won't solve serving the backend-files (media and static data). These have to be shared between the async and the back part anyway, so a RWX-Volume is necessary in most cases anyway.
I do agree that double-proxying when using an ingress
should be avoided.
Typically bad practice to combine services into a single container. But I wonder why taiga-back can't serve this content (static & media)?
All of the other routes can surely be handled by Ingress tho. A double-proxy typically leads to issues but perhaps this isn't a fault of the Helm Chart but rather the design of Taiga.
I'm pretty sure that taiga-back
could serve static files, but it's certainly better to use a software that is better at serving static files, like nginx. So It would at least need a pod
with the taiga-back
and nginx
containers.
Is the Taiga Gateway needed when utilizing Kubernetes (NGINX) Ingress? Should these paths be handled natively with the NGINX Ingress Controller (if available)?