name: messenger-lab5 # Один образ приложения для API, worker, dispatcher, migrate, s3-bootstrap и janitor. Собирает его # только api: параллельная сборка одного тега несколькими сервисами конфликтует. x-app: &app image: mtusi-messenger:lab5 pull_policy: never env_file: - .env environment: &app-network DB_HOST: postgres DB_PORT: "5432" VALKEY_HOST: valkey VALKEY_PORT: "6379" S3_ENDPOINT: http://s3:9000 x-api-resources: &api-resources resources: limits: cpus: "1.0" memory: 512M reservations: cpus: "0.25" memory: 128M x-api: &api <<: *app restart: unless-stopped depends_on: migrate: condition: service_completed_successfully s3-bootstrap: condition: service_completed_successfully postgres: condition: service_healthy valkey: condition: service_started # Только ограниченные ключи своего bucket; административных у API нет secrets: - s3_access_key_id - s3_secret_access_key networks: - edge - backend - storage healthcheck: test: ["CMD", "python", "src/healthcheck.py"] interval: 5s timeout: 3s retries: 2 start_period: 20s start_interval: 1s stop_grace_period: 10s services: nginx: image: nginx:1.28.0-alpine3.21 restart: unless-stopped depends_on: api: condition: service_healthy ports: - "127.0.0.1:${LB_PORT:-8080}:8080" - "127.0.0.1:${LB_EXPIRY_PORT:-8081}:8081" volumes: - ./infra/nginx/nginx.conf:/etc/nginx/nginx.conf:ro - ./infra/nginx/errors.conf:/etc/nginx/errors.conf:ro networks: - edge healthcheck: test: ["CMD", "wget", "-q", "-O", "/dev/null", "http://127.0.0.1:8090/healthz"] interval: 5s timeout: 2s retries: 3 stop_grace_period: 10s deploy: resources: limits: cpus: "0.5" memory: 64M api: <<: *api build: context: . args: APP_REVISION: ${APP_REVISION:-dev} deploy: <<: *api-resources replicas: ${API_REPLICAS:-2} migrate: <<: *app command: ["python", "src/migrate.py"] restart: "no" depends_on: postgres: condition: service_healthy networks: - backend healthcheck: disable: true deploy: resources: limits: cpus: "0.5" memory: 256M postgres: image: postgres:18.2-alpine3.22 restart: unless-stopped environment: POSTGRES_USER: ${DB_USER} POSTGRES_PASSWORD: ${DB_PASS} POSTGRES_DB: ${DB_NAME} volumes: - postgres_data:/var/lib/postgresql ports: - "127.0.0.1:${DB_HOST_PORT:-55433}:5432" healthcheck: test: ["CMD-SHELL", "pg_isready -U \"$$POSTGRES_USER\" -d \"$$POSTGRES_DB\""] interval: 2s timeout: 3s retries: 15 start_period: 10s networks: - backend - db-host stop_grace_period: 30s deploy: resources: limits: cpus: "1.5" memory: 512M valkey: image: valkey/valkey:8.1.4-alpine3.22 restart: unless-stopped environment: VALKEY_PASS: ${VALKEY_PASS} command: - valkey-server - --requirepass - ${VALKEY_PASS} - --save - "" - --appendonly - "no" - --maxmemory - 64mb - --maxmemory-policy - allkeys-lru healthcheck: test: ["CMD-SHELL", "valkey-cli --no-auth-warning -a \"$$VALKEY_PASS\" ping | grep -q PONG"] interval: 2s timeout: 3s retries: 15 networks: - backend deploy: resources: limits: cpus: "0.25" memory: 96M # SeaweedFS всё-в-одном. master, volume, filer и их веб-интерфейсы слушают # только loopback контейнера; в сеть storage открыт лишь S3 API (9000). # Порты наружу не публикуются: bucket доступен только через API. s3: image: chrislusf/seaweedfs:4.48 restart: unless-stopped command: - server - -dir=/data - -ip=127.0.0.1 - -ip.bind=127.0.0.1 - -master.volumeSizeLimitMB=1024 - -volume.max=0 - -s3 - -s3.port=9000 - -s3.ip.bind=0.0.0.0 - -s3.config=/run/secrets/s3_identities - -s3.port.iceberg=0 - -s3.port.lance=0 # Тела загрузок SeaweedFS держит в памяти: без лимита 20 одновременных # PUT по 9 MiB вызывали OOM kill weed (evidence/experiment-uploads.json). # Сверх лимита запросы ждут своей очереди. - -s3.concurrentUploadLimitMB=128 - -volume.concurrentUploadLimitMB=128 environment: # Мягкий предел кучи Go ниже лимита контейнера: GC срабатывает раньше OOM GOMEMLIMIT: 384MiB secrets: - s3_identities volumes: - s3_data:/data networks: - storage healthcheck: test: ["CMD", "wget", "-q", "-O", "/dev/null", "http://127.0.0.1:9000/healthz"] interval: 2s timeout: 3s retries: 30 start_period: 10s stop_grace_period: 30s deploy: resources: limits: cpus: "1.0" memory: 512M # Идемпотентное создание приватного bucket административными ключами s3-bootstrap: <<: *app command: ["python", "src/s3_bootstrap.py"] restart: "no" depends_on: s3: condition: service_healthy secrets: - s3_admin_access_key_id - s3_admin_secret_access_key networks: - storage healthcheck: disable: true deploy: resources: limits: cpus: "0.5" memory: 256M # Сверка S3/БД: незавершённые загрузки старше grace period и сироты janitor: <<: *app command: ["python", "src/janitor.py"] restart: unless-stopped depends_on: migrate: condition: service_completed_successfully s3-bootstrap: condition: service_completed_successfully postgres: condition: service_healthy secrets: - s3_access_key_id - s3_secret_access_key networks: - backend - storage healthcheck: test: ["CMD", "python", "src/healthcheck.py", "janitor"] interval: 30s timeout: 5s retries: 3 start_period: 30s start_interval: 1s stop_grace_period: 10s deploy: resources: limits: cpus: "0.25" memory: 192M broker: image: rabbitmq:4.1.4-alpine hostname: broker restart: unless-stopped environment: RABBITMQ_DEFAULT_USER: ${RABBITMQ_USER:-test} RABBITMQ_DEFAULT_PASS: ${RABBITMQ_PASS:-test} volumes: - rabbitmq_data:/var/lib/rabbitmq networks: - backend healthcheck: test: ["CMD", "rabbitmq-diagnostics", "-q", "ping"] interval: 5s timeout: 5s retries: 20 stop_grace_period: 30s deploy: resources: limits: cpus: "1.0" memory: 512M dispatcher: <<: *app command: ["python", "src/dispatcher.py"] restart: unless-stopped depends_on: migrate: condition: service_completed_successfully broker: condition: service_healthy secrets: - broker_url networks: - backend healthcheck: test: ["CMD", "python", "src/healthcheck.py", "dispatcher"] interval: 5s timeout: 3s retries: 5 start_period: 20s stop_grace_period: 10s deploy: resources: limits: cpus: "0.25" memory: 192M worker: <<: *app command: ["python", "src/worker.py"] restart: unless-stopped depends_on: migrate: condition: service_completed_successfully s3-bootstrap: condition: service_completed_successfully broker: condition: service_healthy secrets: - broker_url - s3_access_key_id - s3_secret_access_key networks: - backend - storage healthcheck: test: ["CMD", "python", "src/healthcheck.py", "worker"] interval: 5s timeout: 3s retries: 5 start_period: 20s stop_grace_period: 10s deploy: resources: limits: cpus: "2.0" memory: 1792M api-short-ttl: <<: *api profiles: - expiry environment: <<: *app-network SESSION_TTL_SECONDS: "2" deploy: <<: *api-resources replicas: 1 networks: edge: {} backend: internal: true # Только localhost-проброс PostgreSQL для локальных клиентов вроде DBeaver db-host: {} # API, janitor и bootstrap ↔ S3; PostgreSQL и Valkey сюда не входят storage: internal: true volumes: postgres_data: s3_data: rabbitmq_data: # Создаются scripts/init-secrets.sh; в git не попадают secrets: broker_url: file: ./secrets/broker_url s3_access_key_id: file: ./secrets/s3_access_key_id s3_secret_access_key: file: ./secrets/s3_secret_access_key s3_admin_access_key_id: file: ./secrets/s3_admin_access_key_id s3_admin_secret_access_key: file: ./secrets/s3_admin_secret_access_key s3_identities: file: ./secrets/s3_identities.json