# =============================================================================
# OpenTelemetry Collector (contrib) — Deployment + Service, ns `monitoring`.
#
# The four apps' OTel SDKs export OTLP/HTTP to this collector (the
# factory-config configmap, infrastructure/kubernetes/configmap.yaml:
#   OTEL_ENABLED: "true" (flipped on in staging)
#   OTEL_EXPORTER_OTLP_ENDPOINT: http://otel-collector.monitoring.svc.cluster.local:4318
# — that FQDN is why this stack must live in namespace `monitoring`; see the
# kustomization.yaml header in this directory for the full rationale).
#
# Pipeline (rendered inline — kept minimal + commented):
#   receivers: otlp (http 4318)
#   processors: batch + memory_limiter
#   traces: otlp → batch → otlp out (endpoint placeholder — Jaeger/Tempo/APM)
#   logs:   otlp → batch → debug (staging only; switch to a real exporter)
#
# METRICS DOUBLE-COUNT GUARD: with OTEL_ENABLED=true the apps run BOTH an
# OTLP metrics push (30 s periodic reader → this collector's /v1/metrics,
# see packages/observability/src/otel.ts) AND a serverless Prometheus reader
# (each app's own /metrics route, scraped by the factory-* jobs in
# prometheus.yml). Feeding BOTH into the same Prometheus would double-count
# every series. Staging default: the direct-scrape jobs (prometheus.yml)
# are the metrics path and this collector's metrics pipeline is DISABLED
# (commented below) — the collector carries TRACES + logs only. To switch
# to collector-scrape: enable the prometheus exporter + add the
# `otel-collector` scrape job, and REMOVE the factory-* scrape jobs.
#
# Traces: OTLP/HTTP in → batch → OTLP out (the endpoint is a placeholder —
# point it at Jaeger/Tempo or a vendor; the apps tolerate an unreachable
# trace pipeline: initOtel logs the export failure and continues).
# Image pinned: otel/opentelemetry-collector-contrib:0.98.0 (0.98.x — the
# line whose OTLP proto matches the SDK 1.30/0.57 exporter train the apps
# pin; see the TASK-0135 report).
# =============================================================================
apiVersion: apps/v1
kind: Deployment
metadata:
  name: otel-collector
  namespace: monitoring
  labels:
    app.kubernetes.io/name: otel-collector
    app.kubernetes.io/part-of: ai-software-factory
spec:
  replicas: 1
  selector:
    matchLabels:
      app.kubernetes.io/name: otel-collector
  strategy:
    type: Recreate
  template:
    metadata:
      labels:
        app.kubernetes.io/name: otel-collector
    spec:
      containers:
        - name: otel-collector
          image: otel/opentelemetry-collector-contrib:0.98.0
          imagePullPolicy: IfNotPresent
          args:
            - --config=/etc/otelcol/config.yaml
          ports:
            - containerPort: 4318
              name: otlp-http
          volumeMounts:
            - name: config
              mountPath: /etc/otelcol
              readOnly: true
          livenessProbe:
            httpGet:
              path: /
              port: 13133
            initialDelaySeconds: 15
            periodSeconds: 15
          readinessProbe:
            httpGet:
              path: /
              port: 13133
            initialDelaySeconds: 5
            periodSeconds: 10
          resources:
            requests:
              cpu: 100m
              memory: 256Mi
            limits:
              cpu: 500m
              memory: 1Gi
          securityContext:
            allowPrivilegeEscalation: false
            runAsNonRoot: true
            # the contrib image runs as `otel` (uid 1000)
            runAsUser: 1000
            runAsGroup: 1000
            seccompProfile:
              type: RuntimeDefault
      volumes:
        - name: config
          configMap:
            name: otel-collector-config
---
apiVersion: v1
kind: ConfigMap
metadata:
  name: otel-collector-config
  namespace: monitoring
  labels:
    app.kubernetes.io/name: otel-collector
    app.kubernetes.io/part-of: ai-software-factory
data:
  config.yaml: |
    receivers:
      otlp:
        protocols:
          http:
            endpoint: 0.0.0.0:4318
      # (a `prometheus` receiver scraping the collector's own :8888 goes here
      #  if the metrics pipeline below is enabled)

    processors:
      memory_limiter:
        check_interval: 1s
        limit_percentage: 80
        spike_limit_percentage: 20
      batch:
        timeout: 5s
        send_batch_size: 1024

    exporters:
      debug:
        verbosity: basic
      otlp/traces:
        endpoint: trace-otlp.monitoring.svc.cluster.local:4317  # PLACEHOLDER: Jaeger/Tempo/your APM
        tls:
          insecure: true
      # prometheus:  # DISABLED by default — the apps are scraped DIRECTLY
        # endpoint: 0.0.0.0:8888  # (jobs factory-* in prometheus.yml). Enable
        # metrics_endpoint: /metrics  # this AND remove the factory-* scrape
        # force_flush_interval: 10s   # jobs if you switch to collector-scrape.

    service:
      pipelines:
        traces:
          receivers: [otlp]
          processors: [memory_limiter, batch]
          exporters: [otlp/traces, debug]
        # metrics:  # see the note on the prometheus exporter above
        #   receivers: [otlp]
        #   processors: [memory_limiter, batch]
        #   exporters: [prometheus]
        logs:
          receivers: [otlp]
          processors: [memory_limiter, batch]
          exporters: [debug]
---
apiVersion: v1
kind: Service
metadata:
  name: otel-collector
  namespace: monitoring
  labels:
    app.kubernetes.io/name: otel-collector
    app.kubernetes.io/part-of: ai-software-factory
spec:
  selector:
    app.kubernetes.io/name: otel-collector
  ports:
    - name: otlp-http
      port: 4318
      targetPort: 4318