[Feat] add support for creating ServiceMonitor in Helm chart

helm/templates/servicemonitor.yaml

@@ -0,0 +1,25 @@
+{{- if .Values.serviceMonitor.enabled }}
+apiVersion: monitoring.coreos.com/v1
+kind: ServiceMonitor
+metadata:
+  name: {{ .Release.Name }}
+  namespace: {{ default .Release.Namespace .Values.serviceMonitor.namespace }}
+  labels:
+    app.kubernetes.io/managed-by: Helm
+    {{- include "chart.engineLabels" . | nindent 4 }}
+spec:
+  selector:
+    matchLabels:
+      {{- include "chart.engineLabels" . | nindent 6 }}
+  namespaceSelector:
+    matchNames:
+      - {{ .Release.Namespace }}
+  endpoints:
+    - port: "service-port"
+      path: {{ .Values.serviceMonitor.path }}
+      interval: {{ .Values.serviceMonitor.interval }}
+      {{- if .Values.serviceMonitor.relabelings }}
+      relabelings:
+        {{- toYaml .Values.serviceMonitor.relabelings | nindent 8 }}
+      {{- end }}
+{{- end }}

helm/values.schema.json

@@ -400,6 +400,30 @@
           }
         }
       }
+    },
+    "serviceMonitor": {
+      "type": "object",
+      "properties": {
+        "enabled": {
+          "type": "boolean"
+        },
+        "namespace": {
+          "type": "string"
+        },
+        "path": {
+          "type": "string"
+        },
+        "interval": {
+          "type": "string"
+        },
+        "relabelings": {
+          "type": "array",
+          "items": {
+            "type": "object",
+            "additionalProperties": true
+          }
+        }
+      }
     }
   }
 }

helm/values.yaml

@@ -268,3 +268,23 @@ routerSpec:
   #  httpGet:
   #    # -- Path to access on the HTTP server
   #
+
+# -- ServiceMonitor configuration for Prometheus Operator
+serviceMonitor:
+  # -- Enable ServiceMonitor
+  enabled: false
+  # -- Namespace where ServiceMonitor resource should be created
+  namespace: ""
+  # -- Path to scrape metrics from
+  path: "/metrics"
+  # -- Interval at which metrics should be scraped
+  interval: "10s"
+  # -- RelabelConfigs to apply to samples before scraping
+  # -- More info: https://github.com/prometheus-operator/prometheus-operator/blob/main/Documentation/api-reference/api.md#monitoring.coreos.com/v1.RelabelConfig
+  relabelings: []
+  # - sourceLabels: [__meta_kubernetes_pod_node_name]
+  #   separator: ;
+  #   regex: ^(.*)$
+  #   targetLabel: nodename
+  #   replacement: $1
+  #   action: replace

observability/README.md

@@ -42,6 +42,18 @@ Open the webpage at `http://<IP of your node>:3000` to access the Grafana web pa
 
 Import the dashboard using the `vllm-dashboard.json` in this folder.
 
+## Enable metrics collection with ServiceMonitor
+
+The helm chart includes built-in support for metrics collection using the ServiceMonitor resource for Prometheus. To enable this feature, set `serviceMonitor.enabled` to `true` in your `values.yaml` file:
+
+```yaml
+serviceMonitor:
+  enabled: true
+
+```
+
+This will automatically create a ServiceMonitor resource that configures Prometheus to scrape metrics from your vLLM deployments. The ServiceMonitor is discovered by the Prometheus Operator deployed by kube-prometheus-stack.
+
 ## Use Prometheus Adapter to export vLLM metrics
 
 The vLLM router can export metrics to Prometheus using the [Prometheus Adapter](https://github.com/prometheus-community/helm-charts/tree/main/charts/prometheus-adapter).

observability/kube-prom-stack.yaml

@@ -90,17 +90,3 @@ prometheusOperator:
 ##
 prometheus:
   enabled: true
-
-  # Monitor vLLM pods using ServiceMonitor
-  additionalServiceMonitors:
-    - name: "vllm-monitor"
-      selector:
-        matchLabels:
-          app.kubernetes.io/managed-by: Helm
-          environment: test
-          release: test
-      namespaceSelector:
-        matchNames:
-          - default
-      endpoints:
-        - port: "service-port"