version: '3'
x-common: &common
  networks:
    - signoz-net
  deploy:
    restart_policy:
      condition: on-failure
  logging:
    options:
      max-size: 50m
      max-file: '3'
x-clickhouse-defaults: &clickhouse-defaults
  !!merge <<: *common
  image: clickhouse/clickhouse-server:24.1.2-alpine
  tty: true
  user: "1000:1000"
  deploy:
    placement:
      constraints: [node.hostname == manager-node-3]
    labels:
      signoz.io/scrape: 'true'
      signoz.io/port: '9363'
      signoz.io/path: '/metrics'
  depends_on:
    - init-clickhouse
    - zookeeper-1
  healthcheck:
    test:
      - CMD
      - wget
      - --spider
      - -q
      - 0.0.0.0:8123/ping
    interval: 30s
    timeout: 5s
    retries: 3
  ulimits:
    nproc: 65535
    nofile:
      soft: 262144
      hard: 262144
x-zookeeper-defaults: &zookeeper-defaults
  !!merge <<: *common
  image: bitnami/zookeeper:3.7.1
  user: root
  deploy:
    placement:
      constraints: [node.hostname == manager-node-1]
    labels:
      signoz.io/scrape: 'true'
      signoz.io/port: '9141'
      signoz.io/path: '/metrics'
  healthcheck:
    test:
      - CMD-SHELL
      - curl -s -m 2 http://localhost:8080/commands/ruok | grep error | grep null
    interval: 30s
    timeout: 5s
    retries: 3
x-db-depend: &db-depend
  !!merge <<: *common
  depends_on:
    - clickhouse
    - schema-migrator
services:
  init-clickhouse:
    !!merge <<: *common
    image: clickhouse/clickhouse-server:24.1.2-alpine
    command:
      - bash
      - -c
      - |
        version="v0.0.1"
        node_os=$$(uname -s | tr '[:upper:]' '[:lower:]')
        node_arch=$$(uname -m | sed s/aarch64/arm64/ | sed s/x86_64/amd64/)
        echo "Fetching histogram-binary for $${node_os}/$${node_arch}"
        cd /tmp
        wget -O histogram-quantile.tar.gz "https://github.com/SigNoz/signoz/releases/download/histogram-quantile%2F$${version}/histogram-quantile_$${node_os}_$${node_arch}.tar.gz"
        tar -xvzf histogram-quantile.tar.gz
        mv histogram-quantile /var/lib/clickhouse/user_scripts/histogramQuantile
    deploy:
      restart_policy:
        condition: on-failure
    volumes:
      - /mnt/cephfs/signoz/data/clickhouse/user_scripts:/var/lib/clickhouse/user_scripts/
  zookeeper-1:
    !!merge <<: *zookeeper-defaults
    # ports:
    #   - "2181:2181"
    #   - "2888:2888"
    #   - "3888:3888"
    volumes:
      - /mnt/cephfs/signoz/data/zookeeper-1:/bitnami/zookeeper
    environment:
      - ZOO_SERVER_ID=1
      - ALLOW_ANONYMOUS_LOGIN=yes
      - ZOO_AUTOPURGE_INTERVAL=1
      - ZOO_ENABLE_PROMETHEUS_METRICS=yes
      - ZOO_PROMETHEUS_METRICS_PORT_NUMBER=9141
  clickhouse:
    !!merge <<: *clickhouse-defaults
    # TODO: needed for clickhouse TCP connectio
    hostname: clickhouse
    # ports:
    #   - "9000:9000"
    #   - "8123:8123"
    #   - "9181:9181"
    configs:
      - source: clickhouse-config-v1
        target: /etc/clickhouse-server/config.xml
      - source: clickhouse-users-v1
        target: /etc/clickhouse-server/users.xml
      - source: clickhouse-custom-function-v1
        target: /etc/clickhouse-server/custom-function.xml
      - source: clickhouse-cluster-v1
        target: /etc/clickhouse-server/config.d/cluster.xml
    volumes:
      - /mnt/cephfs/signoz/data/clickhouse/data/user_scripts:/var/lib/clickhouse/user_scripts/
      - /mnt/cephfs/signoz/data/clickhouse/data:/var/lib/clickhouse/
      # - ../common/clickhouse/storage.xml:/etc/clickhouse-server/config.d/storage.xml
  signoz:
    !!merge <<: *db-depend
    image: signoz/signoz:v0.86.1
    command:
      - --config=/root/config/prometheus.yml
    # ports:
    # - "8080:8080" # signoz port
    #   - "6060:6060"     # pprof port
    configs:
      - source: signoz-prometheus-config-v1
        target: /root/config/prometheus.yml
    volumes:
      - /mnt/cephfs/signoz/data/dashboards:/root/config/dashboards
      - /mnt/cephfs/signoz/data/sqlite:/var/lib/signoz/
    environment:
      - SIGNOZ_ALERTMANAGER_PROVIDER=signoz
      - SIGNOZ_TELEMETRYSTORE_CLICKHOUSE_DSN=tcp://clickhouse:9000
      - SIGNOZ_SQLSTORE_SQLITE_PATH=/var/lib/signoz/signoz.db
      - DASHBOARDS_PATH=/root/config/dashboards
      - STORAGE=clickhouse
      - GODEBUG=netdns=go
      - TELEMETRY_ENABLED=true
      - DEPLOYMENT_TYPE=docker-swarm
    healthcheck:
      test:
        - CMD
        - wget
        - --spider
        - -q
        - localhost:8080/api/v1/health
      interval: 30s
      timeout: 5s
      retries: 3
    networks:
      - signoz-net
      - traefik_public
    deploy:
      labels:
        - 'traefik.enable=true'
        # --- Router für Signoz UI ---
        - 'traefik.http.routers.signoz.rule=Host(`signoz.genius.ceo`)'
        - 'traefik.http.routers.signoz.entrypoints=https'
        - 'traefik.http.routers.signoz.tls.certresolver=main'
        # --- Service für Signoz UI ---
        - 'traefik.http.services.signoz.loadbalancer.server.port=8080'
        # --- Netzwerk für Traefik ---
        - 'traefik.swarm.network=traefik_public'
  otel-collector:
    !!merge <<: *db-depend
    image: signoz/signoz-otel-collector:v0.111.42
    user: root
    command:
      - --config=/etc/otel-collector-config.yaml
      - --manager-config=/etc/manager-config.yaml
      - --copy-path=/var/tmp/collector-config.yaml
      - --feature-gates=-pkg.translator.prometheus.NormalizeName
    configs:
      - source: otel-collector-config-v4
        target: /etc/otel-collector-config.yaml
      - source: otel-collector-manager-config-v1
        target: /etc/manager-config.yaml
    volumes:
      - /var/run/docker.sock:/var/run/docker.sock:ro
    environment:
      - OTEL_RESOURCE_ATTRIBUTES=host.name={{.Node.Hostname}},os.type={{.Node.Platform.OS}}
      - LOW_CARDINAL_EXCEPTION_GROUPING=false
    ports:
      # - "1777:1777"     # pprof extension
      - '4317:4317' # OTLP gRPC receiver
      - '4318:4318' # OTLP HTTP receiver
    deploy:
      replicas: 3
      labels:
        - 'traefik.enable=true'
        # --- Router für Signoz Collector UI ---
        - 'traefik.http.routers.signoz-collector.rule=Host(`collector.genius.ceo`)'
        - 'traefik.http.routers.signoz-collector.entrypoints=https'
        - 'traefik.http.routers.signoz-collector.tls.certresolver=main'
        # --- Service für Signoz Collector UI ---
        - 'traefik.http.services.signoz-collector.loadbalancer.server.port=4318'
        # --- Netzwerk für Traefik ---
        - 'traefik.swarm.network=traefik_public'
    depends_on:
      - clickhouse
      - schema-migrator
      - signoz
    networks:
      - signoz-net
      - traefik_public
  schema-migrator:
    !!merge <<: *common
    image: signoz/signoz-schema-migrator:v0.111.42
    deploy:
      restart_policy:
        condition: on-failure
        delay: 5s
    entrypoint: sh
    command:
      - -c
      - '/signoz-schema-migrator sync --dsn=tcp://clickhouse:9000 --up= && /signoz-schema-migrator async --dsn=tcp://clickhouse:9000 --up='
    depends_on:
      - clickhouse
networks:
  signoz-net:
    name: signoz-net
    attachable: true
  traefik_public:
    external: true
configs:
  otel-collector-config-v4:
    file: /mnt/cephfs/signoz/config/otel-collector-config.v4.yaml
  otel-collector-manager-config-v1:
    file: /mnt/cephfs/signoz/config/signoz/otel-collector-opamp-config.yaml
  clickhouse-config-v1:
    file: /mnt/cephfs/signoz/config/clickhouse/config.v1.xml
  clickhouse-users-v1:
    file: /mnt/cephfs/signoz/config/clickhouse/users.v1.xml
  clickhouse-custom-function-v1:
    file: /mnt/cephfs/signoz/config/clickhouse/custom-function.v1.xml
  clickhouse-cluster-v1:
    file: /mnt/cephfs/signoz/config/clickhouse/cluster.v1.xml
  signoz-prometheus-config-v1:
    file: /mnt/cephfs/signoz/config/signoz/prometheus.v1.yml