Gateway ServiceMonitor scrapes nginx port 8080 by default but nginx has no /metrics — can we add option to disable or provide nginx metrics?

[bmailhe]

Hi team — quick question / request about the gateway in the Mimir chart.

What I see (using chart 6.0.2)

• The chart deploys an nginx‑based gateway (listening on 8080) and the chart also creates a ServiceMonitor for the gateway by default.
• The ServiceMonitor points to port 8080 expecting a /metrics endpoint, but the nginx gateway does not expose Prometheus metrics on /metrics out of the box, so the target is DOWN and produces noise in Prometheus.

Questions / request

1. Did I miss a config option to point the ServiceMonitor at a real metrics endpoint (or to configure the metrics path/port)?
2. Can you add an option to disable the gateway ServiceMonitor (eg. gateway.metrics.enabled=false or gateway.serviceMonitor.enabled=false)? That would let users who don’t run an exporter avoid the bad target.
3. Alternatively, could the chart ship a lightweight nginx exporter (or enable nginx’s stub_status / prometheus module) so the gateway exports useful metrics on /metrics by default?

Why this helps

• Right now the default results in a ServiceMonitor scraping a non‑existent endpoint → false alerts / noisy targets in Prometheus.
• Either disabling the ServiceMonitor by default (or making it configurable) or providing a real /metrics endpoint would remove noise for users who keep default settings.

Suggested minimal API (examples)

• Disable ServiceMonitor:

gateway:
  serviceMonitor:
    enabled: false

• Or configure path/port the ServiceMonitor scrapes:
• Or include an optional exporter sidecar (toggle):

Thanks