docker-compose.yml

services:
    db:
        image: postgres:16
        environment:
            POSTGRES_PASSWORD: postgres
            POSTGRES_DB: transcripts
        volumes:
            - dbdata:/var/lib/postgresql/data
            # NOTE: schema.sql is no longer applied automatically on first boot
            # Migrations are now managed by Alembic via the 'migrations' service
            # If you need to bootstrap from schema.sql for testing, apply it manually
        # Expose Postgres to the host for external clients (psql/DBeaver/etc.)
        # If port 5432 is already used on your host, change to something like "5434:5432".
        ports:
            - '${DB_HOST_PORT:-5435}:5432'
        healthcheck:
            test: ['CMD-SHELL', 'pg_isready -U postgres']
            interval: 5s
            timeout: 5s
            retries: 5

    redis:
        image: redis:7-alpine
        ports:
            - '${REDIS_HOST_PORT:-6380}:6379'
        volumes:
            - redis-data:/data
        command: redis-server --save 60 1 --loglevel warning
        healthcheck:
            test: ['CMD', 'redis-cli', 'ping']
            interval: 5s
            timeout: 3s
            retries: 5
        restart: unless-stopped

    migrations:
        image: transcript-create:latest
        env_file: .env
        environment:
            DATABASE_URL: postgresql+psycopg://postgres:postgres@db:5432/transcripts
        depends_on:
            db:
                condition: service_healthy
        command: ['python3', 'scripts/run_migrations.py', 'upgrade']
        restart: 'no'

    api:
        build:
            context: .
            args:
                # Change to rocm6.1 or rocm6.2 if your host ROCm matches those
                ROCM_WHEEL_INDEX: https://download.pytorch.org/whl/rocm6.0
        image: transcript-create:latest
        env_file: .env
        environment:
            # Point app to the compose network Postgres
            DATABASE_URL: postgresql+psycopg://postgres:postgres@db:5432/transcripts
            # Point app to Redis cache inside the compose network
            REDIS_URL: redis://redis:6379/0
            # Point app to OpenSearch service inside the compose network
            OPENSEARCH_URL: http://opensearch:9200
            # Persist model caches to a mounted volume
            HF_HOME: /root/.cache/hf
            HF_HUB_CACHE: /root/.cache/hf/hub
            TRANSFORMERS_CACHE: /root/.cache/hf/transformers
            LOG_LEVEL: DEBUG
            # Disable OpenTelemetry SDK to avoid external telemetry posts
            OTEL_SDK_DISABLED: 'true'
        ports:
            # Map host port 8000 to container port 8000 (matches OAuth callback URLs)
            # To use a different host port, set API_HOST_PORT in .env and update OAuth redirect URIs
            - '${API_HOST_PORT:-8000}:8000'
        depends_on:
            db:
                condition: service_healthy
            redis:
                condition: service_healthy
            migrations:
                condition: service_completed_successfully
        volumes:
            - ./data:/data
            - ./cache:/root/.cache
        devices:
            - '/dev/kfd:/dev/kfd'
            - '/dev/dri:/dev/dri'
        group_add:
            - 'video'

    worker:
        image: transcript-create:latest
        env_file: .env
        environment:
            DATABASE_URL: postgresql+psycopg://postgres:postgres@db:5432/transcripts
            OPENSEARCH_URL: http://opensearch:9200
            # Persist model caches to a mounted volume
            HF_HOME: /root/.cache/hf
            HF_HUB_CACHE: /root/.cache/hf/hub
            TRANSFORMERS_CACHE: /root/.cache/hf/transformers
            LOG_LEVEL: DEBUG
            # Use ROCm PyTorch whisper backend and force GPU
            WHISPER_BACKEND: 'whisper'
            WHISPER_MODEL: 'large-v3'
            FORCE_GPU: 'true'
            # ROCm stability tweaks for some RDNA cards/drivers
            HSA_ENABLE_SDMA: '0'
            # Suppress PyTorch FutureWarnings for trusted models
            PYTHONWARNINGS: 'ignore::FutureWarning'
            # Disable OpenTelemetry SDK to avoid external telemetry posts
            OTEL_SDK_DISABLED: 'true'
            # Optional: yt-dlp tuning
            # Point to a cookies file inside the container to access age-restricted/private videos
            # Example: put cookies at ./cache/yt-dlp/cookies.txt and set YTDLP_COOKIES=/root/.cache/yt-dlp/cookies.txt in .env
            YTDLP_COOKIES: ${YTDLP_COOKIES:-}
            # Extra args passed verbatim to yt-dlp (space-separated)
            YTDLP_EXTRA_ARGS: ${YTDLP_EXTRA_ARGS:-}
        depends_on:
            db:
                condition: service_healthy
            migrations:
                condition: service_completed_successfully
        command: ['python3', '-m', 'worker.loop']
        volumes:
            - ./data:/data
            - ./cache:/root/.cache
        devices:
            - '/dev/kfd:/dev/kfd'
            - '/dev/dri:/dev/dri'
        group_add:
            - 'video'

    opensearch:
        image: opensearchproject/opensearch:2.12.0
        environment:
            - discovery.type=single-node
            - DISABLE_SECURITY_PLUGIN=true
            - DISABLE_INSTALL_DEMO_CONFIG=true
            - bootstrap.memory_lock=true
            - OPENSEARCH_JAVA_OPTS=-Xms512m -Xmx512m
        volumes:
            - ./config/opensearch/analysis:/usr/share/opensearch/config/analysis:ro
        ulimits:
            memlock:
                soft: -1
                hard: -1
        ports:
            - '${OPENSEARCH_HOST_PORT:-9201}:9200'

    dashboards:
        image: opensearchproject/opensearch-dashboards:2.12.0
        environment:
            - OPENSEARCH_HOSTS=["http://opensearch:9200"]
        ports:
            - '${DASHBOARDS_HOST_PORT:-5601}:5601'
        depends_on:
            - opensearch

    prometheus:
        image: prom/prometheus:v2.55.0
        command:
            - '--config.file=/etc/prometheus/prometheus.yml'
            - '--storage.tsdb.path=/prometheus'
            - '--storage.tsdb.retention.time=30d'
            - '--storage.tsdb.retention.size=10GB'
            - '--web.console.libraries=/usr/share/prometheus/console_libraries'
            - '--web.console.templates=/usr/share/prometheus/consoles'
        volumes:
            - ./config/prometheus:/etc/prometheus
            - prometheus-data:/prometheus
        ports:
            - '${PROMETHEUS_HOST_PORT:-9091}:9090'
        depends_on:
            - api
            - worker
        restart: unless-stopped

    grafana:
        image: grafana/grafana:11.4.0
        environment:
            - GF_SECURITY_ADMIN_USER=admin
            - GF_SECURITY_ADMIN_PASSWORD=admin
            - GF_USERS_ALLOW_SIGN_UP=false
            - GF_SERVER_ROOT_URL=http://localhost:3000
            - GF_INSTALL_PLUGINS=
        volumes:
            - grafana-data:/var/lib/grafana
            - ./config/grafana/provisioning/datasources:/etc/grafana/provisioning/datasources:ro
            - ./config/grafana/provisioning/dashboards:/etc/grafana/provisioning/dashboards:ro
            - ./config/grafana/dashboards:/var/lib/grafana/dashboards:ro
        ports:
            - '${GRAFANA_HOST_PORT:-3301}:3000'
        depends_on:
            - prometheus
        restart: unless-stopped

    backup:
        image: postgres:16
        env_file: .env
        environment:
            DATABASE_URL: postgresql+psycopg://postgres:postgres@db:5432/transcripts
            BACKUP_DIR: /backups
            PGHOST: db
            PGPORT: 5432
            PGUSER: postgres
            PGPASSWORD: postgres
            PGDATABASE: transcripts
        depends_on:
            db:
                condition: service_healthy
        volumes:
            - dbdata:/var/lib/postgresql/data:ro
            - ./backups:/backups
            - ./scripts:/scripts:ro
            - ./data:/data:ro
        # Run backup daily at 2 AM UTC
        # Export metrics every 5 minutes for Prometheus
        command: >
            sh -c "
            apt-get update && apt-get install -y cron rsync &&
            mkdir -p /backups/logs &&
            echo '0 2 * * * cd /scripts && ./backup_db.sh >> /backups/logs/cron.log 2>&1' > /etc/cron.d/backup &&
            echo '0 3 * * * cd /scripts && ./backup_media.sh >> /backups/logs/cron.log 2>&1' >> /etc/cron.d/backup &&
            echo '0 4 * * 0 cd /scripts && ./verify_backup.sh >> /backups/logs/cron.log 2>&1' >> /etc/cron.d/backup &&
            echo '*/5 * * * * cd /scripts && ./export_backup_metrics.sh --output-file /backups/metrics.prom 2>&1 | logger -t backup-metrics' >> /etc/cron.d/backup &&
            chmod 0644 /etc/cron.d/backup &&
            crontab /etc/cron.d/backup &&
            echo 'Backup service started. Backups scheduled for 2 AM UTC daily.' &&
            echo 'Metrics export scheduled every 5 minutes.' &&
            cron -f
            "
        restart: unless-stopped

volumes:
    dbdata:
    redis-data:
    prometheus-data:
    grafana-data: