diff --git a/conf/journal_publisher.yml b/conf/journal_publisher.yml index bff47c2..c64af76 100644 --- a/conf/journal_publisher.yml +++ b/conf/journal_publisher.yml @@ -1,19 +1,22 @@ brokers: - kafka temporary_prefix: swh.tmp.journal.new - -storage: - cls: remote - args: - url: http://swh-storage:5002/ - -max_messages: 1 +final_prefix: swh.journal.objects +consumer_id: swh.journal.publisher +publisher_id: swh.journal.publisher object_types: - content - revision - release - origin - origin_visit - snapshot + +storage: + cls: remote + args: + url: http://swh-storage:5002/ + +max_messages: 1 diff --git a/docker-compose.yml b/docker-compose.yml index 9b8090e..8e89dcd 100644 --- a/docker-compose.yml +++ b/docker-compose.yml @@ -1,356 +1,358 @@ 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 kafka: image: wurstmeister/kafka ports: - 5092:9092 env_file: ./env/kafka.env depends_on: - zookeeper prometheus: image: prom/prometheus 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 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 environment: POSTGRES_PASSWORD: testpassword POSTGRES_DB: swh-scheduler swh-scheduler-api: image: swh/stack build: ./ env_file: ./env/scheduler.env entrypoint: /swh-scheduler-api/entrypoint.sh environment: SWH_CONFIG_FILENAME: /scheduler.yml SWH_SCHEDULER_CONFIG_FILE: /scheduler.yml depends_on: - swh-scheduler-db ports: - 5008:5008 volumes: - "./conf/scheduler.yml:/scheduler.yml:ro" swh-scheduler-listener: image: swh/stack build: ./ entrypoint: /swh-scheduler-worker/entrypoint.sh env_file: ./env/scheduler.env environment: SWH_CONFIG_FILENAME: /scheduler.yml SWH_SCHEDULER_CONFIG_FILE: /scheduler.yml command: listener depends_on: - swh-scheduler-api - amqp volumes: - "./conf/scheduler.yml:/scheduler.yml:ro" swh-scheduler-runner: image: swh/stack build: ./ entrypoint: /swh-scheduler-worker/entrypoint.sh env_file: ./env/scheduler.env environment: SWH_CONFIG_FILENAME: /scheduler.yml SWH_SCHEDULER_CONFIG_FILE: /scheduler.yml command: runner -p 10 depends_on: - swh-scheduler-api - amqp volumes: - "./conf/scheduler.yml:/scheduler.yml:ro" # Graph storage swh-storage-db: image: postgres:11 environment: POSTGRES_PASSWORD: testpassword POSTGRES_DB: swh-storage swh-storage: build: ./ image: swh/stack entrypoint: /swh-storage/entrypoint.sh ports: - 5002:5002 depends_on: - swh-storage-db - swh-objstorage env_file: ./env/storage.env volumes: - "./conf/storage.yml:/storage.yml:ro" # Object storage swh-objstorage: build: ./ image: swh/stack entrypoint: /swh-objstorage/entrypoint.sh ports: - 5003:5003 environment: SWH_CONFIG_FILENAME: /objstorage.yml volumes: - "./conf/objstorage.yml:/objstorage.yml:ro" # Indexer storage swh-idx-storage-db: image: postgres:11 environment: POSTGRES_PASSWORD: testpassword POSTGRES_DB: swh-idx-storage swh-idx-storage: build: ./ image: swh/stack entrypoint: /swh-indexer-storage/entrypoint.sh ports: - 5007:5007 depends_on: - swh-idx-storage-db environment: POSTGRES_PASSWORD: testpassword POSTGRES_DB: swh-idx-storage PGHOST: swh-idx-storage-db PGUSER: postgres SWH_CONFIG_FILENAME: /indexer_storage.yml volumes: - "./conf/indexer_storage.yml:/indexer_storage.yml:ro" # Web interface swh-web: build: ./ image: swh/stack entrypoint: /swh-web/entrypoint.sh 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 PYTHONPATH: /tmp/swh volumes: - "./conf/web.yml:/web.yml:ro" swh-deposit-db: image: postgres:11 env_file: ./env/deposit.env swh-deposit: build: ./ entrypoint: /swh-deposit/entrypoint.sh image: swh/stack ports: - 5006:5006 depends_on: - swh-deposit-db - swh-scheduler-api env_file: ./env/deposit.env environment: PGHOST: swh-deposit-db volumes: - "./conf/deposit.yml:/deposit.yml:ro" swh-vault-db: image: postgres:11 environment: POSTGRES_PASSWORD: testpassword POSTGRES_DB: swh-vault swh-vault-api: build: ./ image: swh/stack entrypoint: /swh-vault/entrypoint.sh env_file: ./env/vault.env command: server ports: - 5005:5005 depends_on: - swh-vault-db - swh-objstorage - swh-storage - swh-scheduler-api volumes: - "./conf/vault-api.yml:/vault-api.yml:ro" swh-vault-worker: build: ./ image: swh/stack entrypoint: /swh-vault/entrypoint.sh command: worker environment: SWH_CONFIG_FILENAME: /cooker.yml depends_on: - swh-vault-api - swh-storage volumes: - "./conf/vault-worker.yml:/cooker.yml:ro" # Lister Celery workers swh-listers-db: image: postgres:11 environment: POSTGRES_PASSWORD: testpassword swh-lister: image: swh/stack build: ./ entrypoint: /swh-listers-worker/entrypoint.sh env_file: ./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 volumes: - "./conf/lister.yml:/lister.yml:ro" # Loader Celery workers swh-loader: image: swh/stack build: ./ entrypoint: /swh-loaders-worker/entrypoint.sh 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 depends_on: - swh-storage - amqp volumes: - "./conf/loader.yml:/loader.yml:ro" # Indexer Celery workers swh-indexer: image: swh/stack build: ./ user: swh entrypoint: /swh-indexer-worker/entrypoint.sh env_file: ./env/indexers.env environment: STATSD_HOST: prometheus-statsd-exporter STATSD_PORT: 9125 depends_on: - swh-scheduler-runner - swh-idx-storage - swh-storage - swh-objstorage - amqp volumes: - "./conf/indexer.yml:/indexer.yml:ro" # Journal related swh-storage-listener: image: swh/stack build: ./ entrypoint: /swh-storage-listener/entrypoint.sh env_file: ./env/storage.env depends_on: - swh-storage-db - kafka volumes: - "./conf/storage_listener.yml:/etc/softwareheritage/storage/listener.yml:ro" swh-journal-publisher: image: swh/stack build: ./ entrypoint: /swh-journal-publisher/entrypoint.sh + environment: + SWH_CONFIG_FILENAME: /journal_publisher.yml depends_on: - kafka - swh-storage-listener volumes: - - "./conf/journal_publisher.yml:/etc/softwareheritage/journal/publisher.yml:ro" + - "./conf/journal_publisher.yml:/journal_publisher.yml:ro" swh-journal-client: image: swh/stack build: ./ entrypoint: /swh-journal-client/entrypoint.sh depends_on: - swh-journal-publisher volumes: - "./conf/journal_client.yml:/etc/softwareheritage/journal/logger.yml:ro" swh-indexer-journal-client: image: swh/stack build: ./ entrypoint: /swh-indexer-journal-client/entrypoint.sh depends_on: - swh-journal-publisher - swh-scheduler-api volumes: - "./conf/journal_client.yml:/etc/softwareheritage/indexer/journal_client.yml:ro" diff --git a/services/swh-journal-publisher/entrypoint.sh b/services/swh-journal-publisher/entrypoint.sh index db1f527..ce19873 100755 --- a/services/swh-journal-publisher/entrypoint.sh +++ b/services/swh-journal-publisher/entrypoint.sh @@ -1,24 +1,24 @@ #!/bin/bash set -e if [[ -d /src ]] ; then for srcrepo in /src/swh-* ; do pushd $srcrepo pip install -e . popd done fi echo Installed Python packages: pip list case "$1" in "shell") exec bash -i ;; *) - echo "Starting swh-journal-publisher" - exec wait-for-it kafka:9092 -s --timeout=0 -- python3 -m swh.journal.publisher + echo "Starting swh-journal-publisher" + exec wait-for-it kafka:9092 -s --timeout=0 -- swh-journal publisher ;; esac