apiVersion: opentelemetry.io/v1beta1
kind: OpenTelemetryCollector
metadata:
  name: otel
spec:
  mode: daemonset
  env:
    - name: CLICKHOUSE_PASSWORD
      valueFrom:
        secretKeyRef:
          name: clickhouse-passwords
          key: collector

  volumes:
    - name: htpasswd
      secret:
        secretName: otel-basicauth
        items:
          - key: .htpasswd
            path: .htpasswd
  volumeMounts:
    - name: htpasswd
      mountPath: .htpasswd
      subPath: .htpasswd

  config:
    receivers:
      hostmetrics:
        scrapers:
          cpu: {}
          disk: {}
          filesystem: {}
          load: {}
          memory: {}
          network: {}
          process: {}
          processes: {}
          paging: {}
      otlp:
        protocols:
          grpc:
            endpoint: 0.0.0.0:4317
          http:
            endpoint: 0.0.0.0:4318
            auth:
              authenticator: basicauth
      kubeletstats:
        collection_interval: 10s
        auth_type: "serviceAccount"
        endpoint: "${env:K8S_NODE_NAME}:10250"
        insecure_skip_verify: true
        metric_groups:
          - node
          - pod
          - container

    processors:
      memory_limiter:
        check_interval: 5s
        limit_mib: 4000
        spike_limit_mib: 500
      batch:
        send_batch_size: 2000
        send_batch_max_size: 2000
        timeout: 5s
      k8sattributes:
        extract:
          metadata:
            - k8s.namespace.name
            - k8s.pod.name
            - k8s.pod.start_time
            - k8s.pod.uid
            - k8s.deployment.name
            - k8s.node.name

    exporters:
      debug:
        verbosity: detailed
      clickhouse:
        endpoint: tcp://clickhouse-otel-cluster.otel.svc.cluster.local:9000?dial_timeout=10s&compress=lz4&async_insert=1
        username: collector
        password: ${env:CLICKHOUSE_PASSWORD}
        ttl: 168h # a week
        create_schema: true
        timeout: 5s
        database: otel
        sending_queue:
          queue_size: 1000
        retry_on_failure:
          enabled: true
          initial_interval: 5s
          max_interval: 30s
          max_elapsed_time: 300s

    extensions:
      health_check:
        endpoint: 0.0.0.0:13133
      pprof:
        endpoint: 0.0.0.0:1777
      zpages:
        endpoint: 0.0.0.0:55679
      basicauth:
        htpasswd:
          file: .htpasswd

    service:
      extensions: [basicauth, health_check, pprof, zpages]
      pipelines:
        traces:
          receivers: [otlp]
          processors: [k8sattributes, memory_limiter, batch]
          exporters: [clickhouse]
        metrics:
          receivers: [otlp, hostmetrics, kubeletstats]
          processors: [k8sattributes, memory_limiter, batch]
          exporters: [clickhouse]
        logs:
          receivers: [otlp]
          processors: [k8sattributes, memory_limiter, batch]
          exporters: [clickhouse]