Issues with installing Loki in air-gapped env

Grafana Loki
,
1

I am trying to get Loki (SingleBinary deployment) and all its components installed in this on-premise k8s cluster I have, but its hard since the cluster is also air-gapped (I do have access to a proxy if needed, but dont know really where to put that configuration).

I already have kube-prometheus-stack installed and I want to expand the functionality with logs and tracing.

I set global.imageRegistry to the internal container registry thinking it would override everything, like this:

global:
  imageRegistry: <internal container-registry>
loki:
  commonConfig:
...

but I find that I have to give everything an image registry reference anyway, like this:

read:
  image:
    registry: <internal container registry>
    ...
write:
  image:
    registry: <internal container registry>
    ...

Is there something I have overlooked, or am I doing the setup wrong?

2

Your container runtime (docker, containerd, podman, …) should be configured to use HTTP proxy - check doc of used container runtime how to do that

3

I dont seem to have much control over the container runtime (containerd) since the cluster is provisioned by another department. The reason for it is container images must be staged to a local registry.

But I guess there is no other way then. I was just hoping the global.imageRegistry value would just use a specified container registry address for everything and not be overridden by default helm values.