2.0.1 install post render error "prometheus.io/scrape" already defined

[dat-intellis]

The chart 2.0.1 release yields the following error when attempting to install with a simple configuration:

Solution: Reverting back to 2.0.0

Error:

k8s-monitoring	2.0.1	False	False	Helm install failed for release observability/k8s-monitoring with chart [email protected]: error while running post render on files: map[string]interface {}(nil): yaml: unmarshal errors: line 20: mapping key "prometheus.io/scrape" already defined at line 19

Values:

cluster:
  name: my-cluster

destinations:
  - name: hostedMetrics
    type: prometheus
    url: <REDACTED>.grafana.net/api/prom/push
    auth:
      type: basic
      username: <REDACTED>
      password: <REDACTED>

  - name: hostedLogs
    type: loki
    url: <REDACTED>.grafana.net/loki/api/v1/push
    auth:
      type: basic
      username: <REDACTED>
      password: <REDACTED>

  - name: otlpGateway
    type: otlp
    url: <REDACTED>.grafana.net:443
    auth:
      type: basic
      username: <REDACTED>
      password: <REDACTED>
    traces:  { enabled: true }

clusterEvents:
  enabled: true

clusterMetrics:
  enabled: true

alloy-metrics:
  enabled: true
  
alloy-singleton:
  enabled: true

[petewall]

There's an underlying problem with windows exporter.
Unless that's important, try adding this:

clusterMetrics:
  windows-exporter:
    enabled: false
    deploy: false

[petewall]

I created an issue upstream to fix this: prometheus-community/helm-charts#5129

[petewall]

Upstream PR: prometheus-community/helm-charts#5132
In the meantime, i'll remove the problem code. it will result in duplicate metrics, if:

1. clusterMetrics w/ Windows exporter is enabled
2. Windows nodes are present
3. annotationAutodiscovery is enabled
4. The annotations are modified to match on prometheus.io/scrape

That's a lot of conditions, but not unheard of.