diff --git a/conf/loader.yml b/conf/loader.yml index 39753ae..285779b 100644 --- a/conf/loader.yml +++ b/conf/loader.yml @@ -1,49 +1,54 @@ storage: cls: filter args: storage: cls: buffer args: storage: cls: remote args: url: http://swh-storage:5002/ min_batch_size: content: 10000 content_bytes: 104857600 directory: 1000 revision: 1000 +scheduler: + cls: remote + args: + url: http://swh-scheduler-api:5008/ + celery: task_broker: amqp://guest:guest@amqp// task_modules: - swh.loader.dir.tasks - swh.loader.git.tasks - swh.loader.mercurial.tasks - swh.loader.npm.tasks - swh.loader.pypi.tasks - swh.loader.svn.tasks - swh.loader.tar.tasks - swh.deposit.loader.tasks - swh.loader.package.tasks task_queues: - swh.loader.package.tasks.LoadDebian - swh.loader.dir.tasks.LoadDirRepository - swh.loader.git.tasks.LoadDiskGitRepository - swh.loader.git.tasks.UncompressAndLoadDiskGitRepository - swh.loader.git.tasks.UpdateGitRepository - swh.loader.mercurial.tasks.LoadArchiveMercurial - swh.loader.mercurial.tasks.LoadMercurial - swh.loader.npm.tasks.LoadNpm - swh.loader.pypi.tasks.LoadPyPI - swh.loader.svn.tasks.DumpMountAndLoadSvnRepository - swh.loader.svn.tasks.LoadSvnRepository - swh.loader.svn.tasks.MountAndLoadSvnRepository - swh.loader.tar.tasks.LoadTarRepository - swh.deposit.loader.tasks.LoadDepositArchiveTsk - swh.deposit.loader.tasks.ChecksDepositTsk lister_db_url: postgresql://postgres@swh-listers-db/swh-listers url: 'http://swh-deposit:5006' diff --git a/docker-compose.yml b/docker-compose.yml index 3fbcf90..44da7d2 100644 --- a/docker-compose.yml +++ b/docker-compose.yml @@ -1,389 +1,390 @@ 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 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: - 5080:5080 # Scheduler swh-scheduler-db: image: postgres:11 env_file: - ./env/scheduler-db.env environment: # unset PGHOST as db service crashes otherwise PGHOST: swh-scheduler-api: image: swh/stack build: ./ env_file: - ./env/scheduler-db.env - ./env/scheduler.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-api/entrypoint.sh:/entrypoint.sh:ro" swh-scheduler-listener: image: swh/stack build: ./ env_file: - ./env/scheduler-db.env - ./env/scheduler.env environment: SWH_CONFIG_FILENAME: /scheduler.yml SWH_SCHEDULER_CONFIG_FILE: /scheduler.yml entrypoint: /entrypoint.sh command: start-listener depends_on: - swh-scheduler-api - 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/scheduler-db.env - ./env/scheduler.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-api - 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/storage-db.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 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/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 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-api environment: VERBOSITY: 3 SWH_CONFIG_FILENAME: /deposit.yml DJANGO_SETTINGS_MODULE: swh.deposit.settings.production env_file: - ./env/deposit-db.env 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-api: image: swh/stack build: ./ env_file: - ./env/vault-db.env environment: SWH_CONFIG_FILENAME: /vault-api.yml command: server ports: - 5005:5005 depends_on: - swh-vault-db - swh-objstorage - swh-storage - swh-scheduler-api entrypoint: /entrypoint.sh volumes: - "./conf/vault-api.yml:/vault-api.yml:ro" - "./services/swh-vault/entrypoint.sh:/entrypoint.sh:ro" swh-vault-worker: image: swh/stack build: ./ command: worker environment: SWH_CONFIG_FILENAME: /cooker.yml depends_on: - swh-vault-api - 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/listers-db.env - ./env/listers.env user: swh environment: STATSD_HOST: prometheus-statsd-exporter STATSD_PORT: 9125 SWH_WORKER_INSTANCE: listers SWH_CONFIG_FILENAME: /lister.yml depends_on: - swh-listers-db - swh-scheduler-api - 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 swh-loader: image: swh/stack build: ./ env_file: - ./env/listers.env user: swh environment: STATSD_HOST: prometheus-statsd-exporter STATSD_PORT: 9125 SWH_WORKER_INSTANCE: loader SWH_CONFIG_FILENAME: /loader.yml entrypoint: /entrypoint.sh depends_on: - swh-storage + - swh-scheduler-api - amqp volumes: - "./conf/loader.yml:/loader.yml:ro" - "./services/swh-loaders-worker/entrypoint.sh:/entrypoint.sh:ro" # Indexer Celery workers swh-indexer: image: swh/stack build: ./ user: swh env_file: - ./env/indexers-db.env - ./env/indexers.env environment: STATSD_HOST: prometheus-statsd-exporter STATSD_PORT: 9125 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 depends_on: - kafka - swh-storage - swh-scheduler-api 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/services/swh-listers-worker/entrypoint.sh b/services/swh-listers-worker/entrypoint.sh index 678f471..3ab19c6 100755 --- a/services/swh-listers-worker/entrypoint.sh +++ b/services/swh-listers-worker/entrypoint.sh @@ -1,46 +1,46 @@ #!/bin/bash set -e source /srv/softwareheritage/utils/pyutils.sh setup_pip source /srv/softwareheritage/utils/pgsql.sh setup_pgsql case "$1" in "shell") exec bash -i ;; *) wait_pgsql echo Setup ${POSTGRES_DB} database for ${SWH_WORKER_INSTANCE} if psql -lqt | cut -d \| -f 1 | grep -qw ${POSTGRES_DB}; then echo Database already exists, nothing to do else echo Creating database createdb ${POSTGRES_DB} fi echo Initialize database swh lister -C ${SWH_CONFIG_FILENAME} db-init - echo Register lister task types in scheduler database - swh lister -C ${SWH_CONFIG_FILENAME} register-task-types + echo Register task types in scheduler database + swh scheduler -C ${SWH_CONFIG_FILENAME} task-type register echo Waiting for RabbitMQ to start wait-for-it amqp:5672 -s --timeout=0 echo Starting the swh-lister 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 diff --git a/services/swh-loaders-worker/entrypoint.sh b/services/swh-loaders-worker/entrypoint.sh index 2a15927..d230f49 100755 --- a/services/swh-loaders-worker/entrypoint.sh +++ b/services/swh-loaders-worker/entrypoint.sh @@ -1,27 +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 ${SWH_WORKER_INSTANCE} exec python -m celery worker \ --app=swh.scheduler.celery_backend.config.app \ --pool=prefork --events \ --concurrency=${CONCURRENCY} \ --maxtasksperchild=${MAX_TASKS_PER_CHILD} \ -Ofair --loglevel=${LOGLEVEL} --without-gossip \ --without-mingle \ --heartbeat-interval 10 \ --hostname "${SWH_WORKER_INSTANCE}@%h" ;; esac