diff --git a/docker/docker-compose.yml b/docker/docker-compose.yml index 88617de..452dbb6 100644 --- a/docker/docker-compose.yml +++ b/docker/docker-compose.yml @@ -1,422 +1,415 @@ version: '2' services: amqp: image: rabbitmq:3.6-management ports: - 5072:5672 # flower: # image: mher/flower # command: --broker=amqp://guest:guest@amqp:5672// --url_prefix=flower # ports: # - 5055:5555 # depends_on: # - amqp zookeeper: image: wurstmeister/zookeeper restart: always kafka: image: wurstmeister/kafka ports: - "5092:9092" env_file: ./env/kafka.env depends_on: - zookeeper kafka-manager: image: hlebalbau/kafka-manager:stable ports: - "5093:9000" environment: ZK_HOSTS: zookeeper:2181 APPLICATION_SECRET: random-secret command: -Dpidfile.path=/dev/null prometheus: image: prom/prometheus depends_on: - prometheus-statsd-exporter command: # Needed for the reverse-proxy - "--web.external-url=/prometheus" - "--config.file=/etc/prometheus/prometheus.yml" volumes: - "./conf/prometheus.yml:/etc/prometheus/prometheus.yml:ro" restart: unless-stopped prometheus-statsd-exporter: image: prom/statsd-exporter command: - "--statsd.mapping-config=/etc/prometheus/statsd-mapping.yml" volumes: - "./conf/prometheus-statsd-mapping.yml:/etc/prometheus/statsd-mapping.yml:ro" restart: unless-stopped prometheus-rabbitmq-exporter: image: kbudde/rabbitmq-exporter restart: unless-stopped environment: SKIP_QUEUES: "RPC_.*" MAX_QUEUES: 5000 RABBIT_URL: http://amqp:15672 grafana: image: grafana/grafana restart: unless-stopped depends_on: - prometheus environment: GF_SERVER_ROOT_URL: http://localhost:5080/grafana volumes: - "./conf/grafana/provisioning:/etc/grafana/provisioning:ro" - "./conf/grafana/dashboards:/var/lib/grafana/dashboards" nginx: image: nginx volumes: - "./conf/nginx.conf:/etc/nginx/nginx.conf:ro" ports: - 5080:80 # Scheduler swh-scheduler-db: image: postgres:11 env_file: - ./env/common_python.env - ./env/scheduler-db.env environment: # unset PGHOST as db service crashes otherwise PGHOST: swh-scheduler: image: swh/stack build: ./ env_file: - ./env/common_python.env - ./env/scheduler-db.env - ./env/scheduler.env - - ./env/statsd.env environment: SWH_CONFIG_FILENAME: /scheduler.yml SWH_SCHEDULER_CONFIG_FILE: /scheduler.yml entrypoint: /entrypoint.sh depends_on: - swh-scheduler-db ports: - 5008:5008 volumes: - "./conf/scheduler.yml:/scheduler.yml:ro" - "./services/swh-scheduler/entrypoint.sh:/entrypoint.sh:ro" swh-scheduler-listener: image: swh/stack build: ./ env_file: - ./env/common_python.env - ./env/scheduler-db.env - ./env/scheduler.env - - ./env/statsd.env environment: SWH_CONFIG_FILENAME: /scheduler.yml SWH_SCHEDULER_CONFIG_FILE: /scheduler.yml entrypoint: /entrypoint.sh command: start-listener depends_on: - swh-scheduler - amqp volumes: - "./conf/scheduler.yml:/scheduler.yml:ro" - "./services/swh-scheduler-worker/entrypoint.sh:/entrypoint.sh:ro" swh-scheduler-runner: image: swh/stack build: ./ env_file: - ./env/common_python.env - ./env/scheduler-db.env - ./env/scheduler.env - - ./env/statsd.env environment: SWH_CONFIG_FILENAME: /scheduler.yml SWH_SCHEDULER_CONFIG_FILE: /scheduler.yml entrypoint: /entrypoint.sh command: start-runner -p 10 depends_on: - swh-scheduler - amqp volumes: - "./conf/scheduler.yml:/scheduler.yml:ro" - "./services/swh-scheduler-worker/entrypoint.sh:/entrypoint.sh:ro" # Graph storage swh-storage-db: image: postgres:11 env_file: - ./env/storage-db.env environment: # unset PGHOST as db service crashes otherwise PGHOST: swh-storage: image: swh/stack build: ./ ports: - 5002:5002 depends_on: - swh-storage-db - swh-objstorage - kafka env_file: - ./env/common_python.env - ./env/storage-db.env - - ./env/statsd.env environment: SWH_CONFIG_FILENAME: /storage.yml STORAGE_BACKEND: postgresql entrypoint: /entrypoint.sh volumes: - "./conf/storage.yml:/storage.yml:ro" - "./services/swh-storage/entrypoint.sh:/entrypoint.sh:ro" # Object storage swh-objstorage: build: ./ image: swh/stack ports: - 5003:5003 env_file: - ./env/common_python.env - - ./env/statsd.env environment: SWH_CONFIG_FILENAME: /objstorage.yml entrypoint: /entrypoint.sh volumes: - "./conf/objstorage.yml:/objstorage.yml:ro" - "./services/swh-objstorage/entrypoint.sh:/entrypoint.sh:ro" # Indexer storage swh-idx-storage-db: image: postgres:11 env_file: - ./env/indexers-db.env environment: # unset PGHOST as db service crashes otherwise PGHOST: swh-idx-storage: image: swh/stack build: ./ ports: - 5007:5007 depends_on: - swh-idx-storage-db env_file: - ./env/common_python.env - ./env/indexers-db.env environment: SWH_CONFIG_FILENAME: /indexer_storage.yml entrypoint: /entrypoint.sh volumes: - "./conf/indexer_storage.yml:/indexer_storage.yml:ro" - "./services/swh-indexer-storage/entrypoint.sh:/entrypoint.sh:ro" # Web interface swh-web: build: ./ image: swh/stack ports: - 5004:5004 depends_on: - swh-objstorage - swh-storage - swh-idx-storage env_file: - ./env/common_python.env - - ./env/statsd.env environment: VERBOSITY: 3 DJANGO_SETTINGS_MODULE: swh.web.settings.development SWH_CONFIG_FILENAME: /web.yml entrypoint: /entrypoint.sh volumes: - "./conf/web.yml:/web.yml:ro" - "./services/swh-web/entrypoint.sh:/entrypoint.sh:ro" swh-deposit-db: image: postgres:11 env_file: - ./env/deposit-db.env environment: # unset PGHOST as db service crashes otherwise PGHOST: swh-deposit: image: swh/stack build: ./ ports: - 5006:5006 depends_on: - swh-deposit-db - swh-scheduler env_file: - ./env/common_python.env - - ./env/statsd.env - ./env/deposit-db.env environment: VERBOSITY: 3 SWH_CONFIG_FILENAME: /deposit.yml DJANGO_SETTINGS_MODULE: swh.deposit.settings.production entrypoint: /entrypoint.sh volumes: - "./conf/deposit.yml:/deposit.yml:ro" - "./services/swh-deposit/entrypoint.sh:/entrypoint.sh:ro" swh-vault-db: image: postgres:11 env_file: - ./env/vault-db.env environment: # unset PGHOST as db service crashes otherwise PGHOST: swh-vault: image: swh/stack build: ./ env_file: - ./env/common_python.env - ./env/vault-db.env - - ./env/statsd.env environment: SWH_CONFIG_FILENAME: /vault.yml command: server ports: - 5005:5005 depends_on: - swh-vault-db - swh-objstorage - swh-storage - swh-scheduler entrypoint: /entrypoint.sh volumes: - "./conf/vault.yml:/vault.yml:ro" - "./services/swh-vault/entrypoint.sh:/entrypoint.sh:ro" swh-vault-worker: image: swh/stack build: ./ command: worker env_file: - ./env/common_python.env - - ./env/statsd.env environment: SWH_CONFIG_FILENAME: /cooker.yml depends_on: - swh-vault - swh-storage entrypoint: /entrypoint.sh volumes: - "./conf/vault-worker.yml:/cooker.yml:ro" - "./services/swh-vault/entrypoint.sh:/entrypoint.sh:ro" # Lister Celery workers swh-listers-db: image: postgres:11 env_file: - ./env/listers-db.env environment: # unset PGHOST as db service crashes otherwise PGHOST: swh-lister: image: swh/stack build: ./ env_file: - ./env/common_python.env - ./env/listers-db.env - - ./env/listers.env - - ./env/statsd.env + - ./env/workers.env user: swh environment: SWH_WORKER_INSTANCE: listers SWH_CONFIG_FILENAME: /lister.yml depends_on: - swh-listers-db - swh-scheduler - swh-scheduler-runner - swh-storage - amqp entrypoint: /entrypoint.sh volumes: - "./conf/lister.yml:/lister.yml:ro" - "./services/swh-listers-worker/entrypoint.sh:/entrypoint.sh:ro" -# Loader Celery workers +# Loader + deposit checker Celery workers swh-loader: image: swh/stack build: ./ env_file: - ./env/common_python.env - - ./env/listers.env - - ./env/statsd.env + - ./env/workers.env user: swh environment: SWH_WORKER_INSTANCE: loader SWH_CONFIG_FILENAME: /loader.yml entrypoint: /entrypoint.sh depends_on: - swh-storage - swh-scheduler + - swh-deposit - amqp volumes: - "./conf/loader.yml:/loader.yml:ro" - - "./services/swh-loaders-worker/entrypoint.sh:/entrypoint.sh:ro" + - "./services/swh-worker/entrypoint.sh:/entrypoint.sh:ro" # Indexer Celery workers swh-indexer: image: swh/stack build: ./ user: swh env_file: - ./env/common_python.env - ./env/indexers-db.env - - ./env/indexers.env - - ./env/statsd.env + - ./env/workers.env + environment: + SWH_WORKER_INSTANCE: indexer + SWH_CONFIG_FILENAME: /indexer.yml + CONCURRENCY: 4 entrypoint: /entrypoint.sh depends_on: - swh-scheduler-runner - swh-idx-storage - swh-storage - swh-objstorage - amqp volumes: - "./conf/indexer.yml:/indexer.yml:ro" - "./services/swh-indexer-worker/entrypoint.sh:/entrypoint.sh:ro" # Journal related swh-indexer-journal-client: image: swh/stack build: ./ entrypoint: /entrypoint.sh env_file: - ./env/common_python.env depends_on: - kafka - swh-storage - swh-scheduler volumes: - "./conf/indexer_journal_client.yml:/etc/softwareheritage/indexer/journal_client.yml:ro" - "./services/swh-indexer-journal-client/entrypoint.sh:/entrypoint.sh:ro" diff --git a/docker/env/common_python.env b/docker/env/common_python.env index 2e2011e..b425d4a 100644 --- a/docker/env/common_python.env +++ b/docker/env/common_python.env @@ -1,5 +1,10 @@ # Forces Python's stdout to be shown in docker logs before they exit: PYTHONUNBUFFERED=1 +# Send statsd probes to prometheus-statsd-exporter so they are visible in +# http://localhost:5080/grafana +STATSD_HOST=prometheus-statsd-exporter +STATSD_PORT=9125 + # Uncomment and set this to a Sentry DSN to report errors to Sentry: # SWH_SENTRY_DSN= diff --git a/docker/env/indexers.env b/docker/env/indexers.env deleted file mode 100644 index 5db2f4b..0000000 --- a/docker/env/indexers.env +++ /dev/null @@ -1,5 +0,0 @@ -CONCURRENCY=4 -MAX_TASKS_PER_CHILD=10 -LOGLEVEL=DEBUG -SWH_WORKER_INSTANCE=indexer -SWH_CONFIG_FILENAME=/indexer.yml diff --git a/docker/env/statsd.env b/docker/env/statsd.env deleted file mode 100644 index bbb9a22..0000000 --- a/docker/env/statsd.env +++ /dev/null @@ -1,2 +0,0 @@ -STATSD_HOST=prometheus-statsd-exporter -STATSD_PORT=9125 diff --git a/docker/env/listers.env b/docker/env/workers.env similarity index 100% rename from docker/env/listers.env rename to docker/env/workers.env diff --git a/docker/services/swh-loaders-worker/entrypoint.sh b/docker/services/swh-worker/entrypoint.sh similarity index 91% rename from docker/services/swh-loaders-worker/entrypoint.sh rename to docker/services/swh-worker/entrypoint.sh index 47f4aa3..8be9d0c 100755 --- a/docker/services/swh-loaders-worker/entrypoint.sh +++ b/docker/services/swh-worker/entrypoint.sh @@ -1,30 +1,30 @@ #!/bin/bash set -e source /srv/softwareheritage/utils/pyutils.sh setup_pip case "$1" in "shell") exec bash -i ;; *) echo Waiting for RabbitMQ to start wait-for-it amqp:5672 -s --timeout=0 echo Register task types in scheduler database swh scheduler -C ${SWH_CONFIG_FILENAME} task-type register - echo Starting the swh-loader Celery worker for ${SWH_WORKER_INSTANCE} + echo Starting the swh Celery worker for ${SWH_WORKER_INSTANCE} exec python -m celery worker \ --app=swh.scheduler.celery_backend.config.app \ --pool=prefork --events \ --concurrency=${CONCURRENCY} \ --maxtasksperchild=${MAX_TASKS_PER_CHILD} \ -Ofair --loglevel=${LOGLEVEL} --without-gossip \ --without-mingle \ --heartbeat-interval 10 \ --hostname "${SWH_WORKER_INSTANCE}@%h" ;; esac