diff --git a/docker/conf/indexer.yml b/docker/conf/indexer.yml index 8197a5b..8b18005 100644 --- a/docker/conf/indexer.yml +++ b/docker/conf/indexer.yml @@ -1,33 +1,40 @@ storage: cls: remote url: http://swh-storage:5002/ objstorage: cls: remote url: http://swh-objstorage:5003/ indexer_storage: cls: remote url: http://swh-idx-storage:5007/ scheduler: cls: remote url: http://swh-scheduler:5008/ celery: broker_heartbeat: null task_broker: amqp://guest:guest@amqp// task_modules: - swh.indexer.tasks task_queues: - swh.indexer.tasks.ContentFossologyLicense - swh.indexer.tasks.ContentLanguage - swh.indexer.tasks.ContentMimetype - swh.indexer.tasks.ContentRangeFossologyLicense - swh.indexer.tasks.ContentRangeMimetype - swh.indexer.tasks.Ctags - swh.indexer.tasks.OriginHead - swh.indexer.tasks.OriginMetadata - swh.indexer.tasks.RecomputeChecksums - swh.indexer.tasks.RevisionMetadata +journal: + brokers: + - kafka + group_id: swh.indexer.journal_client + prefix: swh.journal.objects +max_messages: 50 + tools: name: swh-metadata-detector version: 0.0.2 configuration: {} diff --git a/docker/docker-compose.yml b/docker/docker-compose.yml index ecf9cc7..4d5dd55 100644 --- a/docker/docker-compose.yml +++ b/docker/docker-compose.yml @@ -1,559 +1,568 @@ version: "2.1" services: amqp: image: rabbitmq:3.6-management ports: - 5072:5672 zookeeper: image: wurstmeister/zookeeper restart: always ports: - "5081:2181" environment: ZOO_LOG4J_PROP: WARN,CONSOLE kafka: image: wurstmeister/kafka ports: - "5092:5092" env_file: ./env/kafka.env environment: KAFKA_BROKER_ID: 1 KAFKA_CREATE_TOPICS: swh.journal.objects.content:16:1, swh.journal.objects.directory:16:1, swh.journal.objects.origin:16:1, swh.journal.objects.origin_visit:16:1, swh.journal.objects.origin_visit_status:16:1, swh.journal.objects.revision:16:1, swh.journal.objects.release:16:1, swh.journal.objects.skipped_content:16:1, swh.journal.objects.snapshot:16:1, swh.journal.indexed.origin_intrinsic_metadata:16:1 KAFKA_ADVERTISED_LISTENERS: PLAINTEXT://kafka:9092,PLAINTEXT_HOST://127.0.0.1:5092 KAFKA_LISTENERS: PLAINTEXT://:9092,PLAINTEXT_HOST://:5092 KAFKA_LISTENER_SECURITY_PROTOCOL_MAP: PLAINTEXT:PLAINTEXT,PLAINTEXT_HOST:PLAINTEXT KAFKA_INTER_BROKER_LISTENER_NAME: PLAINTEXT depends_on: - zookeeper healthcheck: test: "[ `JMX_PORT= kafka-topics.sh --list --zookeeper zookeeper:2181 | wc -l` -ge 6 ]" interval: 10s timeout: 5s retries: 10 cmak: # Note: CMAK does not work out of the box, you need to run this first: # $ docker-compose exec zookeeper ./bin/zkCli.sh # create /kafka-manager/mutex "" # create /kafka-manager/mutex/locks "" # create /kafka-manager/mutex/leases "" # See: https://github.com/yahoo/CMAK/issues/731#issuecomment-643880544 image: hlebalbau/kafka-manager:stable environment: ZK_HOSTS: "zookeeper:2181" 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 LOG_LEVEL: warning 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" mailhog: image: mailhog/mailhog ports: - "1025:1025" - "8025:8025" memcache: image: memcached restart: unless-stopped nginx: image: nginx volumes: - "./conf/nginx.conf:/etc/nginx/nginx.conf:ro" ports: - 5080:5080 # Scheduler swh-scheduler-db: image: postgres:12 env_file: - ./env/common_python.env - ./env/scheduler-db.env swh-scheduler: image: swh/stack build: ./ env_file: - ./env/common_python.env - ./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/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 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 environment: SWH_CONFIG_FILENAME: /scheduler.yml SWH_SCHEDULER_CONFIG_FILE: /scheduler.yml entrypoint: /entrypoint.sh command: start-runner --period 10 depends_on: - swh-scheduler - amqp volumes: - "./conf/scheduler.yml:/scheduler.yml:ro" - "./services/swh-scheduler-worker/entrypoint.sh:/entrypoint.sh:ro" swh-scheduler-runner-priority: image: swh/stack build: ./ env_file: - ./env/common_python.env - ./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 --period 10 --with-priority depends_on: - swh-scheduler - amqp volumes: - "./conf/scheduler.yml:/scheduler.yml:ro" - "./services/swh-scheduler-worker/entrypoint.sh:/entrypoint.sh:ro" swh-scheduler-schedule-recurrent: image: swh/stack build: ./ env_file: - ./env/common_python.env - ./env/scheduler-db.env - ./env/scheduler.env environment: SWH_CONFIG_FILENAME: /scheduler.yml SWH_SCHEDULER_CONFIG_FILE: /scheduler.yml LOGLEVEL: INFO entrypoint: /entrypoint.sh command: schedule-recurrent 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:12 env_file: - ./env/storage-db.env 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.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 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:12 env_file: - ./env/indexers-db.env 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 - ./env/indexers.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-db: image: postgres:12 env_file: - ./env/01-web-db.env swh-web: build: ./ image: swh/stack ports: - 3000:3000 - 5004:5004 depends_on: - swh-idx-storage - swh-scheduler - swh-storage - swh-web-db - memcache env_file: - ./env/common_python.env - ./env/01-web-db.env - ./env/02-web-db.env environment: VERBOSITY: 3 DJANGO_SETTINGS_MODULE: swh.web.settings.production SWH_CONFIG_FILENAME: /web.yml entrypoint: /entrypoint.sh volumes: - "./conf/web.yml:/web.yml:ro" - "./services/swh-web/entrypoint.sh:/entrypoint.sh:ro" swh-web-cron: build: ./ image: swh/stack depends_on: - swh-web env_file: - ./env/common_python.env - ./env/01-web-db.env - ./env/02-web-db.env environment: VERBOSITY: 2 DJANGO_SETTINGS_MODULE: swh.web.settings.production SWH_CONFIG_FILENAME: /web.yml entrypoint: /entrypoint.sh command: cron volumes: - "./conf/web.yml:/web.yml:ro" - "./services/swh-web/entrypoint.sh:/entrypoint.sh:ro" swh-deposit-db: image: postgres:12 env_file: - ./env/deposit-db.env swh-deposit: image: swh/stack build: ./ ports: - 5006:5006 depends_on: - swh-deposit-db - swh-scheduler env_file: - ./env/common_python.env - ./env/deposit-db.env - ./env/deposit.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:12 env_file: - ./env/vault-db.env swh-vault: image: swh/stack build: ./ env_file: - ./env/common_python.env - ./env/vault-db.env - ./env/vault.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/workers.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-lister: image: swh/stack build: ./ env_file: - ./env/common_python.env - ./env/listers.env - ./env/workers.env user: swh environment: SWH_WORKER_INSTANCE: listers SWH_CONFIG_FILENAME: /lister.yml depends_on: - swh-scheduler - swh-scheduler-runner - amqp - swh-lister-maven-nginx entrypoint: /entrypoint.sh volumes: - "./conf/lister.yml:/lister.yml:ro" - "./services/swh-listers-worker/entrypoint.sh:/entrypoint.sh:ro" swh-lister-maven-nginx: # Http server to host the maven extracted index for the maven lister image: nginx volumes: - "./conf/maven_index/:/usr/share/nginx/html:ro" ports: - 8880:80 # Loader + deposit checker Celery workers swh-loader: image: swh/stack build: ./ env_file: - ./env/common_python.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 - amqp volumes: - "./conf/loader.yml:/loader.yml:ro" - "./services/swh-worker/entrypoint.sh:/entrypoint.sh:ro" swh-loader-deposit: image: swh/stack build: ./ env_file: - ./env/common_python.env - ./env/workers.env user: swh environment: SWH_WORKER_INSTANCE: loader-deposit SWH_CONFIG_FILENAME: /loader-deposit.yml entrypoint: /entrypoint.sh depends_on: - swh-storage - swh-scheduler - swh-deposit - amqp volumes: - "./conf/loader-deposit.yml:/loader-deposit.yml:ro" - "./services/swh-worker/entrypoint.sh:/entrypoint.sh:ro" swh-loader-opam: image: swh/stack build: ./ env_file: - ./env/common_python.env - ./env/workers.env user: swh environment: SWH_WORKER_INSTANCE: loader-opam SWH_CONFIG_FILENAME: /loader-opam.yml entrypoint: /entrypoint.sh depends_on: - swh-storage - swh-scheduler - swh-deposit - amqp volumes: - "./conf/loader-opam.yml:/loader-opam.yml:ro" - "./services/swh-worker/entrypoint.sh:/entrypoint.sh:ro" - # Indexer Celery workers + # Indexer workers - swh-indexer: + swh-indexer-worker-celery: image: swh/stack build: ./ user: swh env_file: - ./env/common_python.env - ./env/indexers-db.env - ./env/indexers.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" + - "./services/swh-indexer-worker-celery/entrypoint.sh:/entrypoint.sh:ro" - # Journal related - - swh-indexer-journal-client: + swh-indexer-worker-journal: image: swh/stack build: ./ - entrypoint: /entrypoint.sh + user: swh env_file: - ./env/common_python.env + - ./env/indexers-db.env + - ./env/indexers.env + - ./env/workers.env + environment: + SWH_WORKER_INSTANCE: indexer + SWH_CONFIG_FILENAME: /indexer.yml + CONCURRENCY: 4 + entrypoint: /entrypoint.sh depends_on: kafka: condition: service_healthy swh-storage: condition: service_started - swh-scheduler: + swh-idx-storage: + condition: service_started + swh-objstorage: condition: service_started - volumes: - - "./conf/indexer_journal_client.yml:/etc/softwareheritage/indexer/journal_client.yml:ro" - - "./services/swh-indexer-journal-client/entrypoint.sh:/entrypoint.sh:ro" + - "./conf/indexer.yml:/indexer.yml:ro" + - "./services/swh-indexer-worker-journal/entrypoint.sh:/entrypoint.sh:ro" + + # Journal related swh-scheduler-journal-client: image: swh/stack build: ./ entrypoint: /entrypoint.sh env_file: - ./env/common_python.env depends_on: kafka: condition: service_healthy swh-scheduler: condition: service_started volumes: - "./conf/scheduler_journal_client.yml:/etc/softwareheritage/scheduler/journal_client.yml:ro" - "./services/swh-scheduler-journal-client/entrypoint.sh:/entrypoint.sh:ro" diff --git a/docker/services/swh-indexer-worker/entrypoint.sh b/docker/services/swh-indexer-worker-celery/entrypoint.sh similarity index 92% rename from docker/services/swh-indexer-worker/entrypoint.sh rename to docker/services/swh-indexer-worker-celery/entrypoint.sh index d285fbb..9d39862 100755 --- a/docker/services/swh-indexer-worker/entrypoint.sh +++ b/docker/services/swh-indexer-worker-celery/entrypoint.sh @@ -1,31 +1,31 @@ #!/bin/bash set -e source /srv/softwareheritage/utils/pyutils.sh setup_pip source /srv/softwareheritage/utils/pgsql.sh setup_pgsql case "$1" in "shell") exec bash -i ;; *) echo Waiting for RabbitMQ to start wait-for-it amqp:5672 -s --timeout=0 wait_pgsql - echo Starting swh-indexer worker + echo Starting swh-indexer Celery-based worker exec python -m celery \ --app=swh.scheduler.celery_backend.config.app \ worker \ --pool=prefork --events \ --concurrency=${CONCURRENCY} \ --max-tasks-per-child=${MAX_TASKS_PER_CHILD} \ -Ofair --loglevel=${LOGLEVEL} \ --hostname "${SWH_WORKER_INSTANCE}@%h" ;; esac diff --git a/docker/services/swh-indexer-worker-journal/entrypoint.sh b/docker/services/swh-indexer-worker-journal/entrypoint.sh new file mode 100755 index 0000000..382bb63 --- /dev/null +++ b/docker/services/swh-indexer-worker-journal/entrypoint.sh @@ -0,0 +1,24 @@ +#!/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 + ;; + *) + echo Waiting for Kafka to start + wait-for-it kafka:9092 -s --timeout=0 + + wait_pgsql + + echo Starting swh-indexer journal-based worker + swh --log-level ${LOGLEVEL} indexer --config-file /indexer.yml journal-client '*' + ;; +esac