Bug Report

What did you do?

Using operator with metrics. Implementation of Metrics (io.javaoperatorsdk.operator.monitoring.micrometer.MicrometerMetrics) is using
io.micrometer.prometheusmetrics.PrometheusMeterRegistry.

This produces a lot of warnings in logs:

Prometheus requires that all meters with the same name have the same set of tag keys. There is already an existing meter named 'operator_sdk_events_received' containing tag keys [action, event, group, kind, name, namespace, scope, version]. The meter you are attempting to register has keys [event, group, kind, name, namespace, scope, version].

It happens in the following method:

Mentioned implementation allows registration of the same metric but with different set of tags. This is not allowed by PrometheusMeterRegistry - see as prometheus treats this as separate time series see

What did you expect to see?

I expect to see all metrics without plenty of warnings in logs (Splunk).

Possible Solution

As operator uses 2 types of events - resource related event with action and non-resource related - I would expect to have the same set of tags, but in case of non-resource related maybe something general.

  public void receivedEvent(Event event, Map<String, Object> metadata) {
    if (event instanceof ResourceEvent) {
      incrementCounter(
          event.getRelatedCustomResourceID(),
          EVENTS_RECEIVED,
          metadata,
          Tag.of(EVENT, event.getClass().getSimpleName()),
          Tag.of(ACTION, ((ResourceEvent) event).getAction().toString()));
    } else {
      incrementCounter(
          event.getRelatedCustomResourceID(),
          EVENTS_RECEIVED,
          metadata,
          Tag.of(EVENT, event.getClass().getSimpleName()));
          Tag.of(ACTION, "NA");
    }
  }