ServiceMonitor manifests are not getting created

[gouravsw]

I am able to deploy the Harbor in K8S cluster. Harbor is up and running without any issues. I am using the latest release

https://github.com/goharbor/harbor-helm/releases/tag/v1.8.1

Harbor OSS version: v2.4.1

For monitoring purpose via Prometheus, I am planning to enable the metrics and it's subfield servicemonitor to true. So that servicemonitor manifests will be created. But somehow these ServiceMonitor manifests are not getting created. While checking the templates for metrcis. It is passing all the checks

{{- if and ( .Capabilities.APIVersions.Has "monitoring.coreos.com/v1" ) .Values.metrics.enabled .Values.metrics.serviceMonitor.enabled }}

As per expression metrics.enabled is true as well as metrics.serviceMonitor.enabled is true. And this check (.Capabilities.APIVersions.Has "monitoring.coreos.com/v1") has the requested api-version

k api-versions  |grep coreos
monitoring.coreos.com/v1
monitoring.coreos.com/v1alpha1

Could someone provide the insight why the ServiceMonitor manifests are not generating? What mistake I am making in values.yaml file?

metrics:
  enabled: true
  core:
    path: /metrics
    port: 8001
  registry:
    path: /metrics
    port: 8001
  jobservice:
    path: /metrics
    port: 8001
  exporter:
    path: /metrics
    port: 8001
  serviceMonitor:
    enabled: true
    additionalLabels: {}
    interval: "15"
    metricRelabelings: 
      # - action: keep
      #   regex: 'kube_(daemonset|deployment|pod|namespace|node|statefulset).+'
      #   sourceLabels: [__name__]
    relabelings: 
      # - sourceLabels: [__meta_kubernetes_pod_node_name]
      #   separator: ;
      #   regex: ^(.*)$
      #   targetLabel: nodename
      #   replacement: $1
      #   action: replace

[zyyw]

As this comment mentioned

https://github.com/goharbor/harbor-helm/blob/a57ee9e7672e03f3c605661109c5291c3819b511/values.yaml#L835

you should probably install prometheus-operator first.

[gouravsw]

@zyyw Prometheus operator was deployed prior to harbor.

[timbrown5]

I'm seeing this behaviour too:

kubectl api-versions | grep monitoring.coreos.com/v1
monitoring.coreos.com/v1
monitoring.coreos.com/v1alpha1

values.yaml:

metrics:
  enabled: true
  serviceMonitor:
    enabled: true

In my case I am using customize to deploy the helm charts:

helmCharts:
  - name: harbor
    version: 1.9.2
    repo: https://helm.goharbor.io
    releaseName: deployment-harbor
    namespace: harbor
    valuesFile: values.yaml

helmCharts:
  - name: kube-prometheus-stack              
    version: 36.6.2
    repo: https://prometheus-community.github.io/helm-charts
    releaseName: deployment-monitoring
    namespace: monitoring
    includeCRDs: false # Run install-crds.sh instead 
    valuesFile: values.yaml

I deployed the Prometheus helm chart after harbor, but the template should be yaml should get regenerated and reapplied each time and I get no Harbor ServiceMonitor listed in the latest yaml that kustomize generates.

[timbrown5]

Ok, in my case I think it's that kustomize isn't telling Harbor that that API exists. I will store a local copy and update the chart to remove that check (storing a local copy and manually rebasing seems to be a best practice anyway).

[divick]

I too am facing same issue with kustomize. Not sure if creating ServiceMonitor separately is a good idea.

[NyCodeGHG]

when kustomize executes helm it probably does not get the capabilites correctly. maybe it's better to remove that check like most helm charts do?

[sudermanjr]

Any use of helm template will also have this issue. I think removing the Capabilities check entirely would be best.