docker-compose.yml

# See usages of this compose file and its commands in the README.md file
volumes:
  local_pg_data:
    driver: local
  local_es_data:
    driver: local

# There are populated by .env when using docker compose
x-application-variables: &application-vars
  # Primary data stores for the API
  DATABASE_URL: ${DATABASE_URL}
  DOWNLOAD_DATABASE_URL: ${DOWNLOAD_DATABASE_URL}
  ES_HOSTNAME: ${ES_HOSTNAME}
  # For connection to Broker
  BROKER_DB: ${BROKER_DB}
  # For local development
  DJANGO_DEBUG: ${DJANGO_DEBUG}
  DJANGO_SETTINGS_MODULE: ${DJANGO_SETTINGS_MODULE}
  IS_LOCAL: ${IS_LOCAL}
  # S3
  RUSTFS_ACCESS_KEY: ${RUSTFS_ACCESS_KEY}
  RUSTFS_SECRET_KEY: ${RUSTFS_SECRET_KEY}
  RUSTFS_HOST: ${RUSTFS_HOST}
  RUSTFS_PORT: ${RUSTFS_PORT}
  # Defined for tracing
  OTEL_EXPORTER_OTLP_TRACES_ENDPOINT: ${OTEL_EXPORTER_OTLP_TRACES_ENDPOINT}
  TOGGLE_OTEL_CONSOLE_LOGGING: ${TOGGLE_OTEL_CONSOLE_LOGGING}

x-spark-submit-variables: &spark-submit-vars
  <<: *application-vars
  AWS_ACCESS_KEY_ID: ${RUSTFS_ACCESS_KEY}
  AWS_ENDPOINT_URL: http://${RUSTFS_HOST}:${RUSTFS_PORT}
  AWS_SECRET_ACCESS_KEY: ${RUSTFS_SECRET_KEY}
  # Needed for the Spark containers to interact
  SPARK_MASTER_HOST: ${SPARK_MASTER_HOST}
  SPARK_MASTER_PORT: ${SPARK_MASTER_PORT}
  SPARK_WORKER_WEBUI_PORT: ${SPARK_WORKER_WEBUI_PORT}
  # This env var needs to be picked up as the config for the spark.sql.warehouse.dir spark conf setting when
  # SparkSessions are created inside a spark-submitted job
  # (i.e., target where host warehouse dir is bound in volume config)
  SPARK_SQL_WAREHOUSE_DIR: ${TARGET_SPARK_SQL_WAREHOUSE_DIR}
  # This env var needs to be picked up as the path part of the config for the
  # spark.hadoop.javax.jdo.option.ConnectionURL spark conf setting when SparkSessions are created inside a
  # spark-submitted job
  # (i.e., a metastore_db sub dir of the target where host warehouse dir is bound in volume config.)
  HIVE_METASTORE_DERBY_DB_DIR: ${TARGET_HIVE_METASTORE_DERBY_DB_DIR}


services:

  usaspending-db:
    command: postgres -c 'max_connections=500'
    profiles:  # must pass --profile usaspending to docker compose for this to come up, or run a service with one of these other profiles
      - ci
      - manage
      - spark
      - test
      - usaspending
    build:
      context: .
      dockerfile: db.Dockerfile
    healthcheck:
      test: [ "CMD-SHELL", "pg_isready -U ${USASPENDING_DB_USER} -d ${USASPENDING_DB_NAME}" ]
      interval: 5s
      timeout: 5s
      retries: 5
    shm_size: 1g
    container_name: usaspending-db
    volumes:
      - type: volume
        source: local_pg_data
        target: /var/lib/postgresql/data
    ports:
      - ${USASPENDING_DB_PORT}:5432
    environment:
      POSTGRES_USER: ${USASPENDING_DB_USER}
      POSTGRES_PASSWORD: ${USASPENDING_DB_PASSWORD}
      POSTGRES_DB: ${USASPENDING_DB_NAME}

  usaspending-manage:
    profiles:
      - manage  # must pass --profile manage to docker compose for this to come up, or use docker compose run
    image: usaspending-development  # when an image by this name is not found in the local repo, and it is forced to build, it will use this as the tag
    build: .
    container_name: usaspending-manage
    volumes:
      - type: bind
        source: .
        target: /usaspending-api
        read_only: false
    depends_on:
      - usaspending-db
    # For an interactive shell, override this command with: docker compose run --rm usaspending-manage python3 -u manage.py shell
    entrypoint: ["uv", "run", "manage.py"]
    environment:
      <<: *application-vars

  usaspending-test:
    profiles:
      - test  # must pass --profile test to docker compose for this to come up, or use docker compose run
    image: usaspending-development  # when an image by this name is not found in the local repo, and it is forced to build, it will use this as the tag
    build:
      context: .
      dockerfile: development.Dockerfile
      args:
        PROJECT_LOG_DIR: ${PROJECT_LOG_DIR}
        SPARK_LOGGING_LEVEL: ${SPARK_LOGGING_LEVEL}
    init: true  # needed for the reaping of processes created for tests
    container_name: usaspending-test
    volumes:
      - type: bind
        source: .
        target: /usaspending-api
        read_only: false
      # Required to interact with host's docker daemon from within this running container,
      # to spin up the data-act-broker-init-test-db container used for broker integration tests (see: conftest.broker_db_setup)
      - /var/run/docker.sock:/var/run/docker.sock
    depends_on:
      - usaspending-s3
      - usaspending-db
      - usaspending-es
    entrypoint: []  # override entrypoint
    environment:
      <<: *spark-submit-vars
      # Location in host machine where broker src code root can be found
      DATA_BROKER_SRC_PATH: "${PWD}/../data-act-broker-backend"
      # Tests will retrieve the root directory from the path and use it to assume the docker network name
      PROJECT_SRC_PATH: "${PWD}"

  usaspending-style-checks:
    profiles:
      - ci  # must pass --profile ci to docker compose for this to come up, or use docker compose run
    image: usaspending-development
    build:
      context: .
      dockerfile: development.Dockerfile
    container_name: usaspending-style-checks
    volumes:
      - type: bind
        source: .
        target: /usaspending-api
        read_only: false
    entrypoint: []  # override entrypoint
    command:
      - sh
      - -c
      - |
        printf -- "==============\nFormatting code:\n"
        git diff --name-only --diff-filter=ACMR | grep '\.py$' | xargs -r uv run ruff format
        printf -- "==============\nChecking code syntax and quality:\n"
        git diff --name-only --diff-filter=ACMR | grep '\.py$' | xargs -r uv run ruff check --fix || exit 1
        printf -- "==============\nChecking API documentation files:\n"
        python3 manage.py check_for_endpoint_documentation
        printf -- "==============\nChecking API contracts:\n"
        dredd > dredd-results.txt && echo '! grep -E "^[warn:|error:]" dredd-results.txt' | bash
    environment:
      <<: *application-vars
      # dredd-hooks (unmaintained since 2018) imports the `imp` module, removed in Python 3.12; patches/imp.py shims it
      PYTHONPATH: patches

  usaspending-ci:
    profiles:
      - ci  # must pass --profile ci to docker compose for this to come up, or use docker compose run
    image: usaspending-development  # when an image by this name is not found in the local repo, and it is forced to build, it will use this as the tag
    build:
      context: .
      dockerfile: development.Dockerfile
    init: true  # needed for the reaping of processes created for tests
    container_name: usaspending-ci
    volumes:
      - type: bind
        source: .
        target: /usaspending-api
        read_only: false
      # Required to interact with host's docker daemon from within this running container,
      # to spin up the data-act-broker-init-test-db container used for broker integration tests (see: conftest.broker_db_setup)
      - /var/run/docker.sock:/var/run/docker.sock
    depends_on:
      usaspending-db:
        condition: service_healthy
      usaspending-es:
        condition: service_healthy
      usaspending-style-checks:
        condition: service_completed_successfully
    entrypoint: []  # override entrypoint
    command:
      - sh
      - -c
      - |
        printf -- "-------\nRunning non-spark integration tests:\n"
        pytest --override-ini=python_files="**/tests/integration/*" -m "(not signal_handling and not spark)" --no-cov --disable-warnings --reuse-db --numprocesses auto
        printf -- "-------\nRunning non-spark integration tests with signal handling:\n"
        pytest --override-ini=python_files="**/tests/integration/*" -m "(signal_handling and not spark)" --no-cov --disable-warnings --reuse-db --numprocesses 0
        printf -- "-------\nRunning spark integration tests -- load to / from delta:\n"
        pytest --override-ini=python_files="test_*.py *_test.py" -m "(spark)" -k "test_load_to_from_delta.py" --no-cov --disable-warnings --reuse-db --numprocesses 2
        printf -- "-------\nRunning spark integration tests -- load transaction fabs / fpds:\n"
        pytest --override-ini=python_files="test_*.py *_test.py" -m "(spark)" -k "test_load_transactions_in_delta_fabs_fpds.py" --no-cov --disable-warnings --reuse-db --numprocesses 0
        printf -- "-------\nRunning spark integration tests -- load transaction lookup:\n"
        pytest --override-ini=python_files="test_*.py *_test.py" -m "(spark)" -k "test_load_transactions_in_delta_lookups.py" --no-cov --disable-warnings --reuse-db --numprocesses 0
        printf -- "-------\nRunning spark integration tests -- other:\n"
        pytest --override-ini=python_files="test_*.py *_test.py" -m "(spark)" -k "(not test_load_to_from_delta.py and not test_load_transactions_in_delta_lookups.py and not test_load_transactions_in_delta_fabs_fpds.py)" --no-cov --disable-warnings --reuse-db --numprocesses auto
        printf -- "-------\nRunning unit tests:\n"
        pytest --override-ini=python_files="test_*.py *_test.py" -m "(not signal_handling)" --ignore-glob="**/tests/integration/*" --no-cov --disable-warnings --reuse-db --numprocesses auto
        printf -- "-------\nRunning unit tests with signal handling:\n"
        pytest --override-ini=python_files="test_*.py *_test.py" -m "(signal_handling)" --ignore-glob="**/tests/integration/*" --no-cov --disable-warnings --reuse-db --numprocesses 0
    environment:
      <<: *spark-submit-vars
      # Location in host machine where broker src code root can be found
      DATA_BROKER_SRC_PATH: "${PWD}/../data-act-broker-backend"
      # Tests will retrieve the root directory from the path and use it to assume the docker network name
      PROJECT_SRC_PATH: "${PWD}"

  usaspending-api:  # builds using the default Dockerfile to mirror deployed environments for local testing
    profiles:
      - usaspending  # must pass --profile usaspending to docker compose for this to come up
    image: usaspending-backend  # when an image by this name is not found in the local repo, and it is forced to build, it will use this as the tag
    build: .
    container_name: usaspending-api
    init: true  # needed for the reaping of processes; mostly related to downloads run in a process
    volumes:
      - type: bind
        source: .
        target: /usaspending-api
        read_only: false
    ports:
      - 8000:8000
    depends_on:
      - usaspending-db
      - usaspending-es
    restart: on-failure:3 # 3 max attempt, and then it will stop restarting
    # Must wait on postgres db to be up (~9s)
    command: /bin/sh -c "sleep 9s; python3 -u manage.py runserver --verbosity 2 0.0.0.0:8000"
    environment:
      <<: *application-vars

  usaspending-bulk-download:
    profiles:
      - usaspending  # must pass --profile usaspending to docker compose for this to come up
    image: usaspending-download  # when an image by this name is not found in the local repo, and it is forced to build, it will use this as the tag
    build:
      context: .
      dockerfile: download.Dockerfile
    init: true  # needed for the reaping of processes
    container_name: usaspending-bulk-download
    restart: on-failure:5 # 5 max attempt, and then it will stop restarting. NOTE: bulk download errors will cause one failure+restart iterations
    volumes:
      - type: bind
        source: .
        target: /usaspending-api
        read_only: false
      # Required to interact with host's docker daemon from within this running container,
      # to submit a spark job via the spark-submit container (see download_sqs_worker._run_spark_download)
      - /var/run/docker.sock:/var/run/docker.sock
    command: python3 manage.py download_sqs_worker
    environment:
      <<: *application-vars

  usaspending-es:
    profiles:
      - usaspending  # must pass --profile usaspending to docker compose for this to come up
      - test
      - ci
    image: opensearchproject/opensearch:2.9.0
    healthcheck:
      test: [ "CMD-SHELL", "curl -f -k http://localhost:9200/_cluster/health || exit 1" ]
      interval: 10s
      timeout: 5s
      retries: 10
      start_period: 60s
    container_name: usaspending-es
    environment:
      - node.name=usaspending-es
      - discovery.seed_hosts=usaspending-es
      - cluster.initial_master_nodes=usaspending-es
      - cluster.name=usaspending
      - network.host=0.0.0.0
      - transport.host=localhost
      - bootstrap.memory_lock=true
      - plugins.security.disabled=true
      - "OPENSEARCH_JAVA_OPTS=-Xms2g -Xmx2g"  # Ensure Docker is allocated plenty of memory, otherwise this will fail
    # Inject plugin install, then resume with orignial entrypoint command
    command: >
      /bin/sh -c "
        if [ ! -d /usr/share/opensearch/plugins/mapper-murmur3 ]; then
          /usr/share/opensearch/bin/opensearch-plugin install mapper-murmur3
        fi
        /usr/share/opensearch/opensearch-docker-entrypoint.sh"
    ulimits:
      memlock:
        soft: -1
        hard: -1
    volumes:
    - type: volume
      source: local_es_data
      target: /usr/share/opensearch/data
    ports:
      - ${ES_PORT}:9200

  usaspending-kibana-es:
    profiles:
      - usaspending  # must pass --profile usaspending to docker compose for this to come up
    image: opensearchproject/opensearch-dashboards:2.9.0
    container_name: usaspending-kibana-es
    depends_on:
      - usaspending-es
    # ELASTICSEARCH_HOSTS should match the port for "usaspending-es"; value will need to be updated if using Windows
    environment:
      - OPENSEARCH_HOSTS="http://usaspending-es:9200"
      - DISABLE_SECURITY_DASHBOARDS_PLUGIN=true
      - DISABLE_SECURITY_PLUGIN=true
    ports:
      - 5601:5601

  usaspending-s3:
    profiles:  # must pass one of these with --profile to docker compose
      - s3
      - spark
      - test
    image: rustfs/rustfs:1.0.0
    container_name: usaspending-s3
    volumes:
      - type: bind
        source: .
        target: /usaspending-api
        read_only: false
      - type: bind
        source: ${S3_DATA_DIR}
        target: /data
    ports:
      - ${RUSTFS_PORT}:10001
      - ${RUSTFS_CONSOLE_PORT}:10002
    environment:
      RUSTFS_ACCESS_KEY: ${RUSTFS_ACCESS_KEY}
      RUSTFS_SECRET_KEY: ${RUSTFS_SECRET_KEY}
    entrypoint: ["rustfs", "server", "--address", ":10001", "--console-address", ":10002", "/data"]
    healthcheck:
      test: ["CMD", "curl", "-f", "http://${RUSTFS_HOST}:${RUSTFS_PORT}/health/ready"]
      interval: 30s
      timeout: 20s
      retries: 3

  # One-shot container that provisions the buckets and seed files RustFS needs for local dev/tests.
  # RustFS (unlike MinIO) does not auto-discover buckets that are created by just placing directories/files directly
  # on its backing filesystem, so this must go through the S3 API.
  usaspending-s3-init:
    profiles:
      - s3
      - spark
      - test
    image: rustfs/rc:v0.1.35
    container_name: usaspending-s3-init
    depends_on:
      usaspending-s3:
        condition: service_healthy
    volumes:
      - type: bind
        source: .
        target: /usaspending-api
        read_only: true
    environment:
      RUSTFS_ACCESS_KEY: ${RUSTFS_ACCESS_KEY}
      RUSTFS_SECRET_KEY: ${RUSTFS_SECRET_KEY}
    entrypoint: ["/bin/sh"]
    command:
      - -c
      - |
        rc alias set local http://usaspending-s3:10001 "$$RUSTFS_ACCESS_KEY" "$$RUSTFS_SECRET_KEY"
        rc mb local/dti-da-public-files-nonprod || true
        rc mb local/bulk-download || true
        rc mb local/monthly-download || true
        rc mb local/data || true
        rc cp /usaspending-api/usaspending_api/data/Data_Dictionary_Crosswalk.xlsx local/dti-da-public-files-nonprod/user_reference_docs/Data_Dictionary_Crosswalk.xlsx
        rc cp /usaspending-api/usaspending_api/data/COVID-19_download_readme.txt local/data/files/COVID-19_download_readme.txt

  spark-master:
    profiles:
      - spark  # must pass --profile spark to docker compose for this to come up
      - test
    image: usaspending-development
    # build context path needs to be relative to project root, from where docker compose will be run
    build:
      context: .
      dockerfile: development.Dockerfile
      args:
        PROJECT_LOG_DIR: ${PROJECT_LOG_DIR}
        SPARK_LOGGING_LEVEL: ${SPARK_LOGGING_LEVEL}
    container_name: spark-master
    environment:
      <<: *spark-submit-vars
    entrypoint: []  # override the entrypoint
    command: >
      /bin/sh -c "
        $${SPARK_HOME}/bin/spark-class org.apache.spark.deploy.master.Master \
        --port $${SPARK_MASTER_PORT} \
        --webui-port $${SPARK_MASTER_WEBUI_PORT}"
    ports:
      - ${SPARK_MASTER_PORT}
      - ${SPARK_MASTER_WEBUI_PORT}
    volumes:
      - type: bind
        source: .
        target: /usaspending-api
        read_only: false
      # Required to interact with host's docker daemon from within this running container and access the Broker DB
      - /var/run/docker.sock:/var/run/docker.sock
  spark-worker:
    profiles:
      - spark  # must pass --profile spark to docker compose for this to come up
      - test
    image: usaspending-development
    # build context path needs to be relative to project root, from where docker compose will be run
    build:
      context: .
      dockerfile: development.Dockerfile
      args:
        PROJECT_LOG_DIR: ${PROJECT_LOG_DIR}
        SPARK_LOGGING_LEVEL: ${SPARK_LOGGING_LEVEL}
    container_name: spark-worker
    depends_on:
      - spark-master
    environment:
      <<: *spark-submit-vars
    entrypoint: []  # override the entrypoint
    command: /bin/sh -c "$${SPARK_HOME}/bin/spark-class org.apache.spark.deploy.worker.Worker --webui-port $${SPARK_WORKER_WEBUI_PORT} spark://$${SPARK_MASTER_HOST}:$${SPARK_MASTER_PORT}"
    ports:
      - ${SPARK_WORKER_WEBUI_PORT}:4041
    volumes:
      - type: bind
        source: .
        target: /usaspending-api
        read_only: false
      # Required to interact with host's docker daemon from within this running container and access the Broker DB
      - /var/run/docker.sock:/var/run/docker.sock
  spark-history-server:
    profiles:
      - spark  # must pass --profile spark to docker compose for this to come up
      - test
    image: usaspending-development
    # build context path needs to be relative to project root, from where docker compose will be run
    build:
      context: .
      dockerfile: development.Dockerfile
      args:
        PROJECT_LOG_DIR: ${PROJECT_LOG_DIR}
        SPARK_LOGGING_LEVEL: ${SPARK_LOGGING_LEVEL}
    container_name: spark-history-server
    environment:
      SPARK_HISTORY_SERVER_PORT: ${SPARK_HISTORY_SERVER_PORT}
    entrypoint: []  # override the entrypoint
    command: /usr/bin/sh -c "$${SPARK_HOME}/bin/spark-class org.apache.spark.deploy.history.HistoryServer"
    ports:
      - ${SPARK_HISTORY_SERVER_PORT}:18080
    volumes:
      - type: bind
        source: .
        target: /usaspending-api
        read_only: false

  # Example of running spark-submit container:
  #  NOTE: double check package dependency versions here with those used in unit tests (conftest_spark.py), as these docs could have gotten stale
  #  (1) Review config values in usaspending_api/config/envs/local.py and override any as needed in a .env file or -e environment variable
  #  (2) Deploy RustFS in docker container (see README.md)
  #  (3) Deploy the postgres DB docker container, if your script connects to a DB.
  #      - If so, also export a JDBC_URL environment variable to pass in to the spark-submit container so it can find its connection
  #  (4) If reading or writing to S3, make sure the bucket given by the value of config setting CONFIG.AWS_S3_BUCKET exists
  #      - e.g. create via UI at http://localhost:10001
  #      - or use RustFS client CLI: rc mb local/data
  #  (4) Run the spark-submit container, citing the dependent packages:
  #    (NOTEs:
  #      - postgresql is needed as a JDBC driver, if connecting to a Postgres DB
  #      - delta-core is needed to read/write in Delta Lake format
  #      - hadoop-aws is needed for the S3AFileSystem.java, used to write data to S3,
  #        - and should use the same hadoop version in your local setup
  #        - NOTE that specifying hadoop-awas should pull in on its own the required version of the aws-java-sdk
  #      - spark-hive is needed to use a hive metastore_db of schemas and tables
  #        - the Docker image at this time only installs spark and hadoop standalone, which does not seem to include all the needed Hive jars
  #
  #    make docker-compose-run profiles="--profile spark" args="--rm -e RUSTFS_HOST=usaspending-s3 -e JDBC_URL -e COMPONENT_NAME='My Spark Prototype Script' spark-submit \
  #      --packages <SEE MAKEFILE> \
  #      /usaspending-api/usaspending_api/etl/tests/path_to_your_spark_prototype_script.py"

  spark-submit:
    profiles:
      - spark  # must pass --profile spark to docker compose for this to come up
      - test
    image: usaspending-development
    # build context path needs to be relative to project root, from where docker compose will be run
    build:
      context: .
      dockerfile: development.Dockerfile
      args:
        PROJECT_LOG_DIR: ${PROJECT_LOG_DIR}
        SPARK_LOGGING_LEVEL: ${SPARK_LOGGING_LEVEL}
    container_name: spark-submit
    depends_on:
      - usaspending-s3
      - spark-history-server
      - spark-master
      - spark-worker
      - usaspending-db
    environment:
      <<: *spark-submit-vars
    # NOTE: entrypoint CANNOT interpolate env vars when processed. They are passed through literally.
    # So in using 1 $ rather than 2 $$, the var is evaluated based on the current SHELL ENV when docker compose is run,
    # and interpolated before accessed as the entrypoint.
    # While this service has values for these interpolated vars in the environment: element, those are not used here,
    # but merely passed into the container. KEEP the two references to these vars and their defaults consistent!
    # To see what it will be, you can run docker compose config (i.e. make docker-compose-config in this project's Makefile)
    entrypoint: /usr/bin/spark-submit --master spark://${SPARK_MASTER_HOST}:${SPARK_MASTER_PORT}
    command: --help
    volumes:
      - type: bind
        source: .
        target: /usaspending-api
        read_only: false
      # NOTE: The hive metastore_db Derby database folder is expected to be configured to show up as a subfolder of the spark-warehouse dir
      - type: bind
        source: ${SOURCE_SPARK_SQL_WAREHOUSE_DIR}
        target: ${TARGET_SPARK_SQL_WAREHOUSE_DIR}
      # Required to interact with host's docker daemon from within this running container and access the Broker DB
      - /var/run/docker.sock:/var/run/docker.sock
  spark-shell:
    profiles:
      - spark  # must pass --profile spark to docker compose for this to come up
      - test
    #image: public.ecr.aws/emr-serverless/spark/emr-7.12.0:latest
    image: usaspending-development
    # build context path needs to be relative to project root, from where docker compose will be run
    build:
      context: .
      dockerfile: development.Dockerfile
      args:
        PROJECT_LOG_DIR: ${PROJECT_LOG_DIR}
        SPARK_LOGGING_LEVEL: ${SPARK_LOGGING_LEVEL}
    container_name: spark-shell
    depends_on:
      - usaspending-s3
    environment:
      <<: *spark-submit-vars
    # NOTE: entrypoint CANNOT interpolate env vars when processed. They are passed through literally.
    # So in using 1 $ rather than 2 $$, the var is evaluated based on the current SHELL ENV when docker compose is run,
    # and interpolated before accessed as the entrypoint.
    # While this service has values for these interpolated vars in the environment: element, those are not used here,
    # but merely passed into the container. KEEP the two references to these vars and their defaults consistent!
    # To see what it will be, you can run docker compose config (i.e. make docker-compose-config in this project's Makefile)
    entrypoint: >
      /usr/bin/pyspark --master local[*]
        --conf spark.sql.extensions=io.delta.sql.DeltaSparkSessionExtension
        --conf spark.sql.catalog.spark_catalog=org.apache.spark.sql.delta.catalog.DeltaCatalog
        --conf spark.hadoop.fs.s3a.endpoint=${RUSTFS_HOST}:${RUSTFS_PORT}
        --conf spark.hadoop.fs.s3a.access.key=${RUSTFS_ACCESS_KEY}
        --conf spark.hadoop.fs.s3a.secret.key=${RUSTFS_SECRET_KEY}
        --conf spark.hadoop.fs.s3a.connection.ssl.enabled=false
        --conf spark.hadoop.fs.s3a.path.style.access=true
        --conf spark.sql.catalogImplementation=hive
        --conf spark.sql.warehouse.dir='${TARGET_SPARK_SQL_WAREHOUSE_DIR}'
        --conf spark.hadoop.javax.jdo.option.ConnectionURL='jdbc:derby:;databaseName=${TARGET_HIVE_METASTORE_DERBY_DB_DIR};create=true'
    volumes:
      - type: bind
        source: .
        target: /usaspending-api
        read_only: false
      # NOTE: The hive metastore_db Derby database folder is expected to be configured to show up as a subfolder of the spark-warehouse dir
      - type: bind
        source: ${SOURCE_SPARK_SQL_WAREHOUSE_DIR}
        target: ${TARGET_SPARK_SQL_WAREHOUSE_DIR}
        read_only: false

  otel-collector:
    profiles:
      - tracing
    image: otel/opentelemetry-collector:0.86.0
    command: [ "--config=/etc/otel-collector.yaml" ]
    volumes:
      - ./otel-collector.yaml:/etc/otel-collector.yaml

  # Tempo runs as user 10001, and docker compose creates the volume as root.
  # As such, we need to chown the volume in order for Tempo to start correctly.
  tempo-init:
    profiles:
      - tracing
    image: &tempoImage grafana/tempo:2.8.2
    user: root
    entrypoint:
      - "chown"
      - "10001:10001"
      - "/var/tempo"
    volumes:
      - ./tempo-data:/var/tempo

  tempo:
    profiles:
      - tracing
    image: *tempoImage
    command: [ "-config.file=/etc/tempo.yaml" ]
    volumes:
      - ./tempo.yaml:/etc/tempo.yaml
      - ./tempo-data:/var/tempo
    ports:
      - "3200" # tempo
      - "4317" # otlp grpc
    depends_on:
      - tempo-init

  grafana:
    profiles:
      - tracing
    image: grafana/grafana:11.0.0
    volumes:
      - ./grafana-datasources.yaml:/etc/grafana/provisioning/datasources/datasources.yaml
    environment:
      - GF_AUTH_ANONYMOUS_ENABLED=true
      - GF_AUTH_ANONYMOUS_ORG_ROLE=Admin
      - GF_AUTH_DISABLE_LOGIN_FORM=true
      - GF_FEATURE_TOGGLES_ENABLE=traceqlEditor
    ports:
      - "3000:3000"