Bug: Helm value api.prometheus_http_prefix for grafana-distributed not working as documented

[Jeansen]

What is the bug?

If I set the following in the values file:

mimir:
  structuredConfig:
    api:
      prometheus_http_prefix: ''

Then, the endpoint in ConfigMap grafana-mimir-gateway-nginx is still the default prometheus:

  # Rest of /prometheus goes to the query frontend
    location /prometheus {
      set $query_frontend grafana-mimir-query-frontend.default.svc.cluster.local.;
      proxy_pass      http://$query_frontend:8080$request_uri;
    }

How to reproduce it?

Deploy Mimir an make sure this excerpt is added to values.yaml:

mimir:
  structuredConfig:
    api:
      prometheus_http_prefix: ''

When Deployed, check ConfigMap grafana-mimir-gateway-nginx and look for the part containing

  # Rest of /prometheus goes to the query frontend
    location /prometheus {
    ...
    }

What did you think would happen?

I would expect that the ConfigMap grafana-mimir-gateway-nginx to have the following entry:

    location / {
      set $query_frontend grafana-mimir-query-frontend.default.svc.cluster.local.;
      proxy_pass      http://$query_frontend:8080$request_uri;
    }

But instead, it is still has the default prefix:

  # Rest of /prometheus goes to the query frontend
    location /prometheus {
      set $query_frontend grafana-mimir-query-frontend.default.svc.cluster.local.;
      proxy_pass      http://$query_frontend:8080$request_uri;
    }

I followed https://github.com/grafana/mimir/blob/main/docs/sources/mimir/configure/configure-the-query-frontend-work-with-prometheus.md

What was your environment?

Kubernetes 1.32.0,
Helm: v3.17.1+g980d8ac
Mimir Helm release: grafana/mimir-distributed 5.6.0 2.15.0

Any additional context to share?

If prometheus_http_prefix: "/" is set, then it works as expected.

So, either this is a bug or documentation needs an update/review.

[narqo]

Let's figure out if this is an issue in our docs or something we need to fix in the chart. cc @56quarters