feature/k8s-237: clean scraped metrics

helm-chart-sources/addons-operator/Chart.yaml

@@ -2,6 +2,6 @@ apiVersion: v2
 name: addon-operator
 description: A Helm chart to distribute the project addon-operator
 type: application
-version: 0.1.6
+version: 0.1.7
 appVersion: "v0.0.8-dd63afe5"
 icon: "https://example.com/icon.png"

helm-chart-sources/addons-operator/templates/monitoring/controller-manager-claim-metrics.yaml

@@ -13,13 +13,17 @@ metadata:
   namespace: {{ .Release.Namespace }}
 spec:
   endpoints:
-  - bearerTokenFile: /var/run/secrets/kubernetes.io/serviceaccount/token
-    path: /metrics
-    port: https
-    scheme: HTTPS
-    tlsConfig:
-      insecureSkipVerify: true
-      serverName: localhost
+    - bearerTokenFile: /var/run/secrets/kubernetes.io/serviceaccount/token
+      path: /metrics
+      port: https
+      scheme: HTTPS
+      tlsConfig:
+        insecureSkipVerify: true
+        serverName: localhost
+      {{- with .Values.monitoring.addonClaimOperator.serviceScrapeConfig.metricRelabelConfigs }}
+      metricRelabelConfigs:
+        {{- toYaml . | nindent 8 }}
+      {{- end }}
   jobLabel: addons-operator-claim
   selector:
     matchLabels:

helm-chart-sources/addons-operator/templates/monitoring/controller-manager-metrics.yaml

@@ -13,13 +13,17 @@ metadata:
   namespace: {{ .Release.Namespace }}
 spec:
   endpoints:
-  - bearerTokenFile: /var/run/secrets/kubernetes.io/serviceaccount/token
-    path: /metrics
-    port: https
-    scheme: HTTPS
-    tlsConfig:
-      insecureSkipVerify: true
-      serverName: localhost
+    - bearerTokenFile: /var/run/secrets/kubernetes.io/serviceaccount/token
+      path: /metrics
+      port: https
+      scheme: HTTPS
+      tlsConfig:
+        insecureSkipVerify: true
+        serverName: localhost
+      {{- with .Values.monitoring.addonOperator.serviceScrapeConfig.metricRelabelConfigs }}
+      metricRelabelConfigs:
+        {{- toYaml . | nindent 8 }}
+      {{- end }}
   jobLabel: addons-operator
   selector:
     matchLabels:

helm-chart-sources/addons-operator/values.yaml

@@ -13,106 +13,95 @@ managerClaim:
   enable: false
   image:
     repository: controller-claim
-
 manager:
   replicas: 1
-
   image:
     repository: controller
     pullPolicy: IfNotPresent
-
   ## Arguments
   ##
   args:
     - --leader-elect
-
   ## Environment variables
   ##
   env: []
-
   ## Image pull secrets
   ##
   imagePullSecrets: []
-
   ## Pod-level security settings
   ##
   podSecurityContext:
     runAsNonRoot: true
     seccompProfile:
-        type: RuntimeDefault
-
+      type: RuntimeDefault
   ## Container-level security settings
   ##
   securityContext:
     allowPrivilegeEscalation: false
     capabilities:
-        drop:
-            - ALL
+      drop:
+        - ALL
     readOnlyRootFilesystem: true
-
   ## Resource limits and requests
   ##
   resources:
     limits:
-        cpu: 500m
-        memory: 512Mi
+      cpu: 500m
+      memory: 512Mi
     requests:
-        cpu: 10m
-        memory: 64Mi
-
+      cpu: 10m
+      memory: 64Mi
   ## Manager pod's affinity
   ##
   affinity: {}
-
   ## Manager pod's node selector
   ##
   nodeSelector: {}
-
   ## Manager pod's tolerations
   ##
   tolerations: []
-
 ## Helper RBAC roles for managing custom resources
 ##
 rbacHelpers:
   # Install convenience admin/editor/viewer roles for CRDs
   enable: false
-
 ## Custom Resource Definitions
 ##
 crd:
   # Install CRDs with the chart
   enable: true
   # Keep CRDs when uninstalling
   keep: true
-
 ## Controller metrics endpoint.
 ## Enable to expose /metrics endpoint with RBAC protection.
 ##
 metrics:
   enable: false
   # Metrics server port
   port: 8443
-
 ## Cert-manager integration for TLS certificates.
 ## Required for webhook certificates and metrics endpoint certificates.
 ##
 certManager:
   enable: false
-
 ## Webhook server configuration
 ##
 webhook:
   enable: true
   # Webhook server port
   port: 9443
-
 ## Prometheus ServiceMonitor for metrics scraping.
 ## Requires prometheus-operator to be installed in the cluster.
 ##
 prometheus:
   enable: false
-
 networkPolicies:
   enabled: false
-  systemEnabled: false
+  systemEnabled: false
+monitoring:
+  addonOperator:
+    serviceScrapeConfig:
+      metricRelabelConfigs: []
+  addonClaimOperator:
+    serviceScrapeConfig:
+      metricRelabelConfigs: []

helm-chart-sources/addonset/files/addons/addons-operator/values/infra.yaml.gotmpl

@@ -0,0 +1,47 @@
+monitoring:
+  addonOperator:
+    serviceScrapeConfig:
+      metricRelabelConfigs:
+        - sourceLabels: [__name__]
+          action: keep
+          regex: "\
+            controller_runtime_active_workers|\
+            controller_runtime_max_concurrent_reconciles|\
+            controller_runtime_reconcile_errors_total|\
+            controller_runtime_reconcile_panics_total|\
+            controller_runtime_reconcile_time_seconds_bucket|\
+            controller_runtime_reconcile_total|\
+            controller_runtime_terminal_reconcile_errors_total|\
+            go_gc_duration_seconds_count|\
+            go_gc_duration_seconds_sum|\
+            go_goroutines|\
+            go_memstats_alloc_bytes_total|\
+            go_memstats_gc_sys_bytes|\
+            go_memstats_heap_idle_bytes|\
+            go_memstats_heap_inuse_bytes|\
+            go_memstats_heap_released_bytes|\
+            go_memstats_mcache_sys_bytes|\
+            go_memstats_mspan_sys_bytes|\
+            go_memstats_other_sys_bytes|\
+            go_memstats_stack_inuse_bytes|\
+            go_memstats_sys_bytes|\
+            leader_election_master_status|\
+            process_cpu_seconds_total|\
+            process_max_fds|\
+            process_open_fds|\
+            process_resident_memory_bytes|\
+            process_start_time_seconds|\
+            process_virtual_memory_bytes|\
+            rest_client_requests_total|\
+            scrape_series_added|\
+            up|\
+            workqueue_adds_total|\
+            workqueue_depth|\
+            workqueue_longest_running_processor_seconds|\
+            workqueue_queue_duration_seconds_count|\
+            workqueue_queue_duration_seconds_sum|\
+            workqueue_retries_total|\
+            workqueue_unfinished_work_seconds|\
+            workqueue_work_duration_seconds_count|\
+            workqueue_work_duration_seconds_sum\
+            "

helm-chart-sources/addonset/files/addons/addons-operator/values/system.yaml.gotmpl

@@ -5,3 +5,51 @@ manager:
     limits:
       cpu: 500m
       memory: 512Mi
+
+monitoring:
+  addonClaimOperator:
+    serviceScrapeConfig:
+      metricRelabelConfigs:
+        - sourceLabels: [__name__]
+          action: keep
+          regex: "\
+            controller_runtime_active_workers|\
+            controller_runtime_max_concurrent_reconciles|\
+            controller_runtime_reconcile_errors_total|\
+            controller_runtime_reconcile_panics_total|\
+            controller_runtime_reconcile_time_seconds_bucket|\
+            controller_runtime_reconcile_total|\
+            controller_runtime_terminal_reconcile_errors_total|\
+            go_gc_duration_seconds_count|\
+            go_gc_duration_seconds_sum|\
+            go_goroutines|\
+            go_memstats_alloc_bytes_total|\
+            go_memstats_gc_sys_bytes|\
+            go_memstats_heap_idle_bytes|\
+            go_memstats_heap_inuse_bytes|\
+            go_memstats_heap_released_bytes|\
+            go_memstats_mcache_sys_bytes|\
+            go_memstats_mspan_sys_bytes|\
+            go_memstats_other_sys_bytes|\
+            go_memstats_stack_inuse_bytes|\
+            go_memstats_sys_bytes|\
+            leader_election_master_status|\
+            process_cpu_seconds_total|\
+            process_max_fds|\
+            process_open_fds|\
+            process_resident_memory_bytes|\
+            process_start_time_seconds|\
+            process_virtual_memory_bytes|\
+            rest_client_requests_total|\
+            scrape_series_added|\
+            up|\
+            workqueue_adds_total|\
+            workqueue_depth|\
+            workqueue_longest_running_processor_seconds|\
+            workqueue_queue_duration_seconds_count|\
+            workqueue_queue_duration_seconds_sum|\
+            workqueue_retries_total|\
+            workqueue_unfinished_work_seconds|\
+            workqueue_work_duration_seconds_count|\
+            workqueue_work_duration_seconds_sum\
+            "