configmap.yaml

---
# TFO-Agent ConfigMap (Node DaemonSet)
# Per-node OS metrics: node_exporter ON, kubernetes collector OFF.
# Auth and endpoint are injected via env vars (TELEMETRYFLOW_API_KEY_ID/SECRET/ENDPOINT).

apiVersion: v1
kind: ConfigMap
metadata:
  name: tfo-agent-config
  namespace: telemetryflow
  labels:
    app.kubernetes.io/name: tfo-agent
    app.kubernetes.io/component: monitoring
data:
  tfo-agent.yaml: |
    agent:
      description: "TFO Agent - ${NODE_NAME}"
      tags:
        environment: "${ENVIRONMENT}"
        cluster: "${CLUSTER_NAME}"

    heartbeat:
      interval: 60s
      timeout: 10s
      include_system_info: true

    collectors:
      node_exporter:
        enabled: true
        interval: 15s
        cpu: true
        memory: true
        disk_io: true
        filesystem: true
        network: true
        load_avg: true
        thermal: false
        textfile: false
        conntrack: false
        psi: false
        vmstat: false
        sockstat: false

      # Native log collector (file tailing + journald)
      logs:
        enabled: false
        paths: []

      # Fluent Bit log collector (production-grade, bundled in image)
      fluent_bit:
        enabled: true
        config_dir: /tmp/tfo-agent-fluentbit
        flush_interval: 5
        log_level: info
        storage_enabled: true
        health_check: true
        restart_on_crash: true
        tail:
          enabled: true
          paths: []
          multiline_parser: "docker,cri"
        systemd:
          enabled: true
          units: [kubelet, docker, containerd]
        kubernetes:
          enabled: false

      # kubernetes is handled by the separate K8s Collector Deployment (tfo-agent-k8s)
      kubernetes:
        enabled: false

      ebpf:
        enabled: false

    prometheus_server:
      enabled: true
      port: 8888
      path: /metrics

    exporter:
      otlp:
        enabled: true
        batch_size: 100
        flush_interval: 10s
        compression: gzip

    buffer:
      enabled: true
      max_size_mb: 100
      path: /var/lib/tfo-agent/buffer
      flush_interval: 30s

    logging:
      level: info
      format: json

---
# TFO-Agent K8s Collector ConfigMap
# Single-replica Deployment: kubernetes collector ON, node_exporter OFF.
# Connects to TFO Platform backend REST API (not the OTLP Collector port).
# Auth and endpoint are injected via env vars.

apiVersion: v1
kind: ConfigMap
metadata:
  name: tfo-agent-k8s-config
  namespace: telemetryflow
  labels:
    app.kubernetes.io/name: tfo-agent
    app.kubernetes.io/component: k8s-collector
data:
  tfo-agent.yaml: |
    agent:
      description: "TFO K8s Collector - ${CLUSTER_NAME}"
      tags:
        environment: "${ENVIRONMENT}"
        cluster: "${CLUSTER_NAME}"

    heartbeat:
      interval: 60s
      timeout: 10s

    collectors:
      node_exporter:
        enabled: false

      kubernetes:
        enabled: true
        interval: 30s
        kubeconfig: ""   # empty = in-cluster ServiceAccount auto-detection
        context: ""
        namespaces: []
        exclude_namespaces:
          - kube-system
          - kube-public
          - kube-node-lease
        nodes: true
        pods: true
        deployments: true
        namespaces_collect: true
        storage: true
        services: true
        workloads: true
        events: true
        resource_counts: true
        network: true    # Kubelet /stats/summary (requires nodes/proxy RBAC)
        metrics_api: true  # CPU/Memory usage from metrics-server (set false if not installed)
        hpa: true
        pdb: true
        pod_logs: true
        pod_logs_tail_lines: 100
        node_logs: true
        node_logs_tail_lines: 200
        node_log_sources: [kubelet, kube-proxy, containerd]
        # Extended metrics — replaces Prometheus + kube-state-metrics + cAdvisor
        apiserver_metrics: true
        coredns_metrics: true
        coredns_service: "" # Auto-discover via pod labels (supports vanilla k8s, EKS, GKE, AKS, RKE2, k3s)
        container_extended_metrics: true  # cpu_throttled, memory_working_set, oom
        pv_io_stats: true                 # PV usage + IOPS from Kubelet volume stats
        sync_to_backend: true
        sync_interval: 60s
        cluster_name: ""     # auto-detected from CLUSTER_NAME env or hostname
        cluster_provider: "" # auto-detected from env/filesystem heuristics
        # cluster_id is auto-registered on startup (find-or-create)
        # Set TELEMETRYFLOW_K8S_CLUSTER_ID env var to skip auto-registration

      ebpf:
        enabled: false

    prometheus_server:
      enabled: true
      port: 8888
      path: /metrics

    exporter:
      otlp:
        enabled: false  # K8s state syncs directly to backend REST API, not via OTLP

    buffer:
      enabled: true
      max_size_mb: 50
      path: /var/lib/tfo-agent/buffer
      flush_interval: 30s

    logging:
      level: info
      format: json