simstudioai--sim/helm/sim/templates/telemetry.yaml

{{- if .Values.telemetry.enabled }}
---
# OpenTelemetry Collector Configuration
apiVersion: v1
kind: ConfigMap
metadata:
  name: {{ include "sim.fullname" . }}-otel-config
  namespace: {{ .Release.Namespace }}
  labels:
    {{- include "sim.labels" . | nindent 4 }}
    app.kubernetes.io/component: telemetry
data:
  otel-config.yaml: |
    receivers:
      otlp:
        protocols:
          grpc:
            endpoint: 0.0.0.0:4317
          http:
            endpoint: 0.0.0.0:4318
      prometheus:
        config:
          scrape_configs:
            - job_name: 'sim-app'
              static_configs:
                - targets: ['{{ include "sim.fullname" . }}-app:{{ .Values.app.service.port }}']
            - job_name: 'sim-realtime'
              static_configs:
                - targets: ['{{ include "sim.fullname" . }}-realtime:{{ .Values.realtime.service.port }}']
      
    processors:
      batch:
        timeout: 1s
        send_batch_size: 1024
      memory_limiter:
        limit_mib: 512
      
    exporters:
      {{- if .Values.telemetry.jaeger.enabled }}
      jaeger:
        endpoint: {{ .Values.telemetry.jaeger.endpoint }}
        tls:
          insecure: {{ not .Values.telemetry.jaeger.tls.enabled }}
      {{- end }}
      {{- if .Values.telemetry.prometheus.enabled }}
      prometheusremotewrite:
        endpoint: {{ .Values.telemetry.prometheus.endpoint }}
        headers:
          Authorization: {{ .Values.telemetry.prometheus.auth | quote }}
      {{- end }}
      {{- if .Values.telemetry.otlp.enabled }}
      otlp:
        endpoint: {{ .Values.telemetry.otlp.endpoint }}
        tls:
          insecure: {{ not .Values.telemetry.otlp.tls.enabled }}
      {{- end }}
      logging:
        loglevel: info
    
    extensions:
      health_check:
        endpoint: 0.0.0.0:13133
      pprof:
        endpoint: 0.0.0.0:1777
      zpages:
        endpoint: 0.0.0.0:55679
    
    service:
      extensions: [health_check, pprof, zpages]
      pipelines:
        traces:
          receivers: [otlp]
          processors: [memory_limiter, batch]
          exporters:
            - logging
            {{- if .Values.telemetry.jaeger.enabled }}
            - jaeger
            {{- end }}
            {{- if .Values.telemetry.otlp.enabled }}
            - otlp
            {{- end }}
        metrics:
          receivers: [otlp, prometheus]
          processors: [memory_limiter, batch]
          exporters:
            - logging
            {{- if .Values.telemetry.prometheus.enabled }}
            - prometheusremotewrite
            {{- end }}
            {{- if .Values.telemetry.otlp.enabled }}
            - otlp
            {{- end }}
        logs:
          receivers: [otlp]
          processors: [memory_limiter, batch]
          exporters:
            - logging
            {{- if .Values.telemetry.otlp.enabled }}
            - otlp
            {{- end }}
---
# OpenTelemetry Collector Deployment
apiVersion: apps/v1
kind: Deployment
metadata:
  name: {{ include "sim.fullname" . }}-otel-collector
  namespace: {{ .Release.Namespace }}
  labels:
    {{- include "sim.labels" . | nindent 4 }}
    app.kubernetes.io/component: telemetry
spec:
  replicas: {{ .Values.telemetry.replicaCount }}
  selector:
    matchLabels:
      {{- include "sim.selectorLabels" . | nindent 6 }}
      app.kubernetes.io/component: telemetry
  template:
    metadata:
      labels:
        {{- include "sim.selectorLabels" . | nindent 8 }}
        app.kubernetes.io/component: telemetry
    spec:
      {{- with .Values.global.imagePullSecrets }}
      imagePullSecrets:
        {{- toYaml . | nindent 8 }}
      {{- end }}
      serviceAccountName: {{ include "sim.serviceAccountName" . }}
      automountServiceAccountToken: false
      {{- include "sim.podSecurityContext" .Values.telemetry | nindent 6 }}
      containers:
        - name: otel-collector
          image: {{ include "sim.image" (dict "imageRoot" .Values.telemetry.image "global" .Values.global "chartAppVersion" .Chart.AppVersion) }}
          imagePullPolicy: {{ .Values.telemetry.image.pullPolicy }}
          command:
            - /otelcol-contrib
            - --config=/etc/otel-collector-config/otel-config.yaml
          ports:
            - name: otlp-grpc
              containerPort: 4317
              protocol: TCP
            - name: otlp-http
              containerPort: 4318
              protocol: TCP
            - name: health
              containerPort: 13133
              protocol: TCP
            - name: pprof
              containerPort: 1777
              protocol: TCP
            - name: zpages
              containerPort: 55679
              protocol: TCP
          env:
            - name: GOGC
              value: "80"
          volumeMounts:
            - name: otel-config
              mountPath: /etc/otel-collector-config
              readOnly: true
          livenessProbe:
            httpGet:
              path: /
              port: health
            initialDelaySeconds: 10
            periodSeconds: 30
            timeoutSeconds: 5
            failureThreshold: 3
          readinessProbe:
            httpGet:
              path: /
              port: health
            initialDelaySeconds: 5
            periodSeconds: 10
            timeoutSeconds: 5
            failureThreshold: 3
          resources:
            {{- toYaml .Values.telemetry.resources | nindent 12 }}
          {{- include "sim.containerSecurityContext" .Values.telemetry | nindent 10 }}
      volumes:
        - name: otel-config
          configMap:
            name: {{ include "sim.fullname" . }}-otel-config
      {{- with .Values.telemetry.nodeSelector }}
      nodeSelector:
        {{- toYaml . | nindent 8 }}
      {{- end }}
      {{- with .Values.telemetry.tolerations }}
      tolerations:
        {{- toYaml . | nindent 8 }}
      {{- end }}
      {{- with .Values.telemetry.affinity }}
      affinity:
        {{- toYaml . | nindent 8 }}
      {{- end }}
---
# OpenTelemetry Collector Service
apiVersion: v1
kind: Service
metadata:
  name: {{ include "sim.fullname" . }}-otel-collector
  namespace: {{ .Release.Namespace }}
  labels:
    {{- include "sim.labels" . | nindent 4 }}
    app.kubernetes.io/component: telemetry
spec:
  type: {{ .Values.telemetry.service.type }}
  ports:
    - name: otlp-grpc
      port: 4317
      targetPort: otlp-grpc
      protocol: TCP
    - name: otlp-http
      port: 4318
      targetPort: otlp-http
      protocol: TCP
    - name: health
      port: 13133
      targetPort: health
      protocol: TCP
  selector:
    {{- include "sim.selectorLabels" . | nindent 4 }}
    app.kubernetes.io/component: telemetry
{{- end }}