docker-compose.yml

networks:
  rca-net:
    driver: bridge

volumes:
  tempo-data:
  prometheus-data:
  grafana-data:
  ollama-data:

services:

  tempo:
    image: grafana/tempo:2.6.1
    container_name: tempo
    command: ["-config.file=/etc/tempo.yml"]
    volumes:
      - ./infra/tempo.yml:/etc/tempo.yml
      - tempo-data:/var/tempo
    ports:
      - "3200:3200"
      - "4317:4317"
    networks:
      - rca-net
    healthcheck:
      test: ["CMD", "wget", "--spider", "-q", "http://localhost:3200/ready"]
      interval: 5s
      timeout: 5s
      retries: 10

  prometheus:
    image: prom/prometheus:v2.54.1
    container_name: prometheus
    command:
      - --config.file=/etc/prometheus/prometheus.yml
      - --storage.tsdb.path=/prometheus
      - --web.enable-remote-write-receiver
    volumes:
      - ./infra/prometheus.yml:/etc/prometheus/prometheus.yml
      - prometheus-data:/prometheus
    ports:
      - "9090:9090"
    networks:
      - rca-net
    healthcheck:
      test: ["CMD", "wget", "--spider", "-q", "http://localhost:9090/-/ready"]
      interval: 5s
      timeout: 5s
      retries: 10

  grafana:
    image: grafana/grafana:11.3.0
    container_name: grafana
    environment:
      - GF_SECURITY_ADMIN_USER=admin
      - GF_SECURITY_ADMIN_PASSWORD=admin
      - GF_AUTH_ANONYMOUS_ENABLED=true
    volumes:
      - grafana-data:/var/lib/grafana
      - ./infra/grafana/provisioning:/etc/grafana/provisioning
    ports:
      - "3000:3000"
    depends_on:
      tempo:
        condition: service_healthy
      prometheus:
        condition: service_healthy
    networks:
      - rca-net

  otel-collector:
    image: otel/opentelemetry-collector-contrib:0.111.0
    container_name: otel-collector
    command: ["--config=/etc/otel-collector-config.yml"]
    volumes:
      - ./infra/otel-collector-config.yml:/etc/otel-collector-config.yml
    ports:
      - "4318:4318"
      - "13133:13133"
    depends_on:
      tempo:
        condition: service_healthy
      prometheus:
        condition: service_healthy
    networks:
      - rca-net

  ollama:
    image: ollama/ollama:latest
    container_name: ollama
    ports:
      - "11434:11434"
    volumes:
      - ollama-data:/root/.ollama
    networks:
      - rca-net

  kafka:
    image: confluentinc/cp-kafka:7.7.0
    container_name: kafka
    ports:
      - "9092:9092"
    environment:
      KAFKA_NODE_ID: 1
      KAFKA_PROCESS_ROLES: broker,controller
      KAFKA_LISTENERS: PLAINTEXT://0.0.0.0:9092,CONTROLLER://0.0.0.0:9093
      KAFKA_ADVERTISED_LISTENERS: PLAINTEXT://kafka:9092
      KAFKA_CONTROLLER_QUORUM_VOTERS: 1@kafka:9093
      KAFKA_CONTROLLER_LISTENER_NAMES: CONTROLLER
      KAFKA_LISTENER_SECURITY_PROTOCOL_MAP: PLAINTEXT:PLAINTEXT,CONTROLLER:PLAINTEXT
      KAFKA_OFFSETS_TOPIC_REPLICATION_FACTOR: 1
      KAFKA_AUTO_CREATE_TOPICS_ENABLE: "true"
      CLUSTER_ID: "MkU3OEVBNTcwNTJENDM2Qg"
    networks:
      - rca-net
    healthcheck:
      test: ["CMD", "kafka-broker-api-versions", "--bootstrap-server", "localhost:9092"]
      interval: 10s
      timeout: 10s
      retries: 10

  order-service:
    build:
      context: .
      dockerfile: Dockerfile
      args:
        SERVICE: order-service
    container_name: order-service
    ports:
      - "8081:8081"
    environment:
      - PORT=8081
      - PAYMENT_SERVICE_URL=http://payment-service:8082
      - INVENTORY_SERVICE_URL=http://inventory-service:8083
      - KAFKA_BOOTSTRAP_SERVERS=kafka:9092
      - JAVA_TOOL_OPTIONS=-javaagent:/otel-agent/opentelemetry-javaagent.jar
      - OTEL_SERVICE_NAME=order-service
      - OTEL_EXPORTER_OTLP_ENDPOINT=http://otel-collector:4318
      - OTEL_EXPORTER_OTLP_PROTOCOL=http/protobuf
      - OTEL_METRICS_EXPORTER=otlp
      - OTEL_LOGS_EXPORTER=otlp
    volumes:
      - ./infra/otel-agent:/otel-agent:ro
    depends_on:
      otel-collector:
        condition: service_started
      kafka:
        condition: service_healthy
    networks:
      - rca-net

  payment-service:
    build:
      context: .
      dockerfile: Dockerfile
      args:
        SERVICE: payment-service
    container_name: payment-service
    ports:
      - "8082:8082"
    environment:
      - PORT=8082
      - INVENTORY_SERVICE_URL=http://inventory-service:8083
      - JAVA_TOOL_OPTIONS=-javaagent:/otel-agent/opentelemetry-javaagent.jar
      - OTEL_SERVICE_NAME=payment-service
      - OTEL_EXPORTER_OTLP_ENDPOINT=http://otel-collector:4318
      - OTEL_EXPORTER_OTLP_PROTOCOL=http/protobuf
      - OTEL_METRICS_EXPORTER=otlp
      - OTEL_LOGS_EXPORTER=otlp
    volumes:
      - ./infra/otel-agent:/otel-agent:ro
    depends_on:
      otel-collector:
        condition: service_started
      inventory-service:
        condition: service_started
    networks:
      - rca-net

  inventory-service:
    build:
      context: .
      dockerfile: Dockerfile
      args:
        SERVICE: inventory-service
    container_name: inventory-service
    ports:
      - "8083:8083"
    environment:
      - PORT=8083
      - JAVA_TOOL_OPTIONS=-javaagent:/otel-agent/opentelemetry-javaagent.jar
      - OTEL_SERVICE_NAME=inventory-service
      - OTEL_EXPORTER_OTLP_ENDPOINT=http://otel-collector:4318
      - OTEL_EXPORTER_OTLP_PROTOCOL=http/protobuf
      - OTEL_METRICS_EXPORTER=otlp
      - OTEL_LOGS_EXPORTER=otlp
    volumes:
      - ./infra/otel-agent:/otel-agent:ro
    depends_on:
      otel-collector:
        condition: service_started
    networks:
      - rca-net

  notification-service:
    build:
      context: .
      dockerfile: Dockerfile
      args:
        SERVICE: notification-service
    container_name: notification-service
    ports:
      - "8084:8084"
    environment:
      - PORT=8084
      - KAFKA_BOOTSTRAP_SERVERS=kafka:9092
      - JAVA_TOOL_OPTIONS=-javaagent:/otel-agent/opentelemetry-javaagent.jar
      - OTEL_SERVICE_NAME=notification-service
      - OTEL_EXPORTER_OTLP_ENDPOINT=http://otel-collector:4318
      - OTEL_EXPORTER_OTLP_PROTOCOL=http/protobuf
      - OTEL_METRICS_EXPORTER=otlp
      - OTEL_LOGS_EXPORTER=otlp
    volumes:
      - ./infra/otel-agent:/otel-agent:ro
    depends_on:
      otel-collector:
        condition: service_started
      kafka:
        condition: service_healthy
    networks:
      - rca-net

  rca-agent:
    build:
      context: .
      dockerfile: Dockerfile.agent
    container_name: rca-agent
    ports:
      - "${RCA_AGENT_PORT:-8080}:8080"
    environment:
      - TEMPO_URL=http://tempo:3200
      - PROMETHEUS_URL=http://prometheus:9090
      - OLLAMA_URL=http://ollama:11434
      - LLM_MODE=${LLM_MODE:-ollama}
      - LLM_LOCAL_MODEL=${LLM_LOCAL_MODEL:-llama3.2}
      - LLM_STANDARD_MODEL=${LLM_STANDARD_MODEL:-gemini-3.1-flash-lite}
      - GEMINI_API_KEY=${GEMINI_API_KEY:-}
      - RCA_CONFIDENCE_THRESHOLD=${RCA_CONFIDENCE_THRESHOLD:-0.75}
    depends_on:
      tempo:
        condition: service_healthy
      prometheus:
        condition: service_healthy
      otel-collector:
        condition: service_started
    networks:
      - rca-net
    healthcheck:
      test: ["CMD", "wget", "--spider", "-q", "http://127.0.0.1:8080/actuator/health"]
      interval: 10s
      timeout: 5s
      retries: 10