docker-compose.yml

name: spectrarag

services:
  qdrant:
    image: qdrant/qdrant:v1.12.4
    container_name: rag-qdrant
    ports:
      - "6333:6333"
      - "6334:6334"
    volumes:
      - qdrant_storage:/qdrant/storage
    healthcheck:
      test: ["CMD", "bash", "-c", "</dev/tcp/localhost/6333"]
      interval: 5s
      timeout: 3s
      retries: 10

  postgres:
    image: postgres:16
    container_name: rag-postgres
    environment:
      POSTGRES_USER: rag
      POSTGRES_PASSWORD: rag
      POSTGRES_DB: rag
    ports:
      - "5432:5432"
    volumes:
      - postgres_data:/var/lib/postgresql/data
    healthcheck:
      test: ["CMD-SHELL", "pg_isready -U rag -d rag"]
      interval: 5s
      timeout: 3s
      retries: 10

  langfuse-db:
    image: postgres:16
    container_name: rag-langfuse-db
    environment:
      POSTGRES_USER: langfuse
      POSTGRES_PASSWORD: langfuse
      POSTGRES_DB: langfuse
    volumes:
      - langfuse_data:/var/lib/postgresql/data
    healthcheck:
      test: ["CMD-SHELL", "pg_isready -U langfuse -d langfuse"]
      interval: 5s
      timeout: 3s
      retries: 10

  langfuse:
    image: langfuse/langfuse:2
    container_name: rag-langfuse
    depends_on:
      langfuse-db:
        condition: service_healthy
    environment:
      DATABASE_URL: postgresql://langfuse:langfuse@langfuse-db:5432/langfuse
      NEXTAUTH_SECRET: replace-me-with-32-chars-min-aaaaaa
      SALT: replace-me-with-32-chars-min-bbbbbb
      NEXTAUTH_URL: http://localhost:3000
      TELEMETRY_ENABLED: "false"
      LANGFUSE_INIT_ORG_ID: local
      LANGFUSE_INIT_PROJECT_ID: rag
    ports:
      - "3000:3000"

  ollama:
    image: ollama/ollama:latest
    container_name: rag-ollama
    ports:
      - "11434:11434"
    volumes:
      - ollama_models:/root/.ollama
    # No hard GPU reservation, so `docker compose up -d qdrant ollama` works on any box; run Ollama natively for GPU embeddings.
    healthcheck:
      test: ["CMD", "ollama", "list"]
      interval: 10s
      timeout: 5s
      retries: 6

  jaeger:
    image: jaegertracing/all-in-one:1.62.0
    container_name: rag-jaeger
    environment:
      COLLECTOR_OTLP_ENABLED: "true"
    ports:
      - "16686:16686"  # UI
      - "4317:4317"    # OTLP gRPC
      - "4318:4318"    # OTLP HTTP
    healthcheck:
      test: ["CMD-SHELL", "wget -q --spider http://localhost:16686 || exit 1"]
      interval: 10s
      timeout: 5s
      retries: 6

  api:
    build:
      context: .
      dockerfile: Dockerfile
    image: rag-api:dev
    container_name: rag-api
    depends_on:
      qdrant:
        condition: service_healthy
      postgres:
        condition: service_healthy
      jaeger:
        condition: service_healthy
    environment:
      RAG_ENV: dev
      RAG_LOG_LEVEL: INFO
      RAG_QDRANT_URL: http://qdrant:6333
      RAG_OLLAMA_BASE_URL: http://ollama:11434
      RAG_POSTGRES_DSN: postgresql+psycopg://rag:rag@postgres:5432/rag
      RAG_LANGFUSE_HOST: http://langfuse:3000
      OTEL_EXPORTER_OTLP_ENDPOINT: http://jaeger:4317
      OTEL_SERVICE_NAME: rag-api-dev
      # Optional, picked up from the host shell. When set, the server wires the
      # LLMQueryClassifier. Without it, routing falls back to the regex, which
      # misses synonyms like "graph", "image", or "bottom panel of...".
      RAG_OPENROUTER_API_KEY: ${RAG_OPENROUTER_API_KEY:-}
      # SENTRY_DSN intentionally unset; sentry stays in no-op mode locally.
    ports:
      - "8000:8000"

volumes:
  qdrant_storage:
  postgres_data:
  langfuse_data:
  ollama_models: