Add optional ServiceMonitor in flyte-binary

charts/flyte-binary/README.md

@@ -172,4 +172,8 @@ Chart for basic single Flyte executable deployment
 | serviceAccount.imagePullSecrets | list | `[]` |  |
 | serviceAccount.labels | object | `{}` |  |
 | serviceAccount.name | string | `""` |  |
+| serviceMonitor.enabled | bool | `false` |  |
+| serviceMonitor.interval | string | `"60s"` |  |
+| serviceMonitor.labels | object | `{}` |  |
+| serviceMonitor.scrapeTimeout | string | `"30s"` |  |
 

charts/flyte-binary/templates/service-monitor.yaml

@@ -0,0 +1,25 @@
+{{- if .Values.serviceMonitor.enabled }}
+apiVersion: monitoring.coreos.com/v1
+kind: ServiceMonitor
+metadata:
+  name: {{ include "flyte-binary.fullname" . }}
+  namespace: {{ .Release.Namespace | quote }}
+  labels: {{- include "flyte-binary.labels" . | nindent 4 }}
+    {{- if .Values.commonLabels }}
+    {{- tpl ( .Values.commonLabels | toYaml ) . | nindent 4 }}
+    {{- end }}
+    {{- if .Values.serviceMonitor.labels }}
+    {{- tpl ( .Values.serviceMonitor.labels | toYaml ) . | nindent 4 }}
+    {{- end }}
+  {{- if .Values.commonAnnotations }}
+  annotations: {{- tpl ( .Values.commonAnnotations | toYaml ) . | nindent 4 }}
+  {{- end }}
+spec:
+  endpoints:
+    - port: http-metrics
+      path: /metrics
+      interval: {{ .Values.serviceMonitor.interval }}
+      scrapeTimeout: {{ .Values.serviceMonitor.scrapeTimeout }}
+  selector:
+    matchLabels: {{- include "flyte-binary.baseLabels" . | nindent 6 }}
+{{- end }}

charts/flyte-binary/values.yaml

@@ -343,6 +343,18 @@ service:
   # extraPorts Additional ports to add to Flyte service
   extraPorts: []
 
+# serviceMonitor Configure a Prometheus Operator ServiceMonitor for Flyte
+# Requires the ServiceMonitor CRD and an `http-metrics` port on the http service (see `service.extraPorts`)
+serviceMonitor:
+  # enabled Create a ServiceMonitor for Flyte
+  enabled: false
+  # interval Interval at which Prometheus scrapes metrics
+  interval: 60s
+  # scrapeTimeout Timeout for each scrape request
+  scrapeTimeout: 30s
+  # labels Add labels to the ServiceMonitor, e.g. those required by your Prometheus to discover it
+  labels: {}
+
 # ingress Configure ingress for Flyte
 ingress:
   # create Create ingress resources

docs/deployment/configuration/monitoring.rst

@@ -156,8 +156,11 @@ Once the prerequisites are in place, follow the instructions in this section to
            - name: http-metrics
              protocol: TCP
              port: 10254
+         serviceMonitor:
+           enabled: true
 
       The above configuration enables the ``serviceMonitor`` that Prometheus can then use to automatically discover services and scrape metrics from them.
+      If your Prometheus only selects ``ServiceMonitor`` objects with specific labels, add them under ``serviceMonitor.labels``.
 
 .. note::