dev-tools/cluster/docker-compose.yml

# Licensed to the Apache Software Foundation (ASF) under one or more
# contributor license agreements.  See the NOTICE file distributed with
# this work for additional information regarding copyright ownership.
# The ASF licenses this file to You under the Apache License, Version 2.0
# (the "License"); you may not use this file except in compliance with
# the License.  You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

# A local Storm cluster in cluster (distributed) mode: dev ZooKeeper + Nimbus +
# two Supervisors + UI, plus a Pushgateway + Prometheus + Grafana stack for
# metrics. Built from the locally compiled distribution so it runs your code.
# Prereq: run ./prepare-extlib.sh once. Start with: docker compose up --build

# Build the Storm image once; every storm service reuses it.
# STORM_VERSION must match the project version in the repo root pom.xml. It is
# read from .env (written by build-image.sh from the pom); the default below is
# the fallback when .env is absent.
x-storm: &storm
  image: storm-local:${STORM_VERSION:-3.0.0-SNAPSHOT}
  build:
    context: ../../storm-dist/binary/final-package/target
    dockerfile: ../../../../dev-tools/cluster/Dockerfile
    args:
      STORM_VERSION: "${STORM_VERSION:-3.0.0-SNAPSHOT}"
  volumes:
    - ./storm.yaml:/opt/storm/conf/storm.yaml:ro
    # Sample input read by FileReadSpout. It runs on the workers, so the file
    # must exist on every supervisor host, not just where you submit from.
    - ../../examples/storm-perf/src/main/sampledata/randomwords.txt:/topology/randomwords.txt:ro
  restart: on-failure
  networks:
    - storm

services:
  # Storm's built-in development ZooKeeper (single node, in-memory-ish, under
  # dev.zookeeper.path).
  zookeeper:
    <<: *storm
    hostname: zookeeper
    command: storm dev-zookeeper
    healthcheck:
      test: ["CMD-SHELL", "python3 -c 'import socket; socket.create_connection((\"localhost\", 2181), 2)'"]
      interval: 10s
      timeout: 5s
      retries: 10
      start_period: 20s

  nimbus:
    <<: *storm
    hostname: nimbus
    # Cluster-summary metrics live on the Nimbus leader, so the Prometheus
    # Pushgateway reporter is enabled here (and only here) via -c overrides --
    # keeping storm.yaml clean and avoiding multiple daemons clobbering the same
    # Pushgateway group. The reporter jars come from the extlib-daemon mount.
    command:
      - storm
      - nimbus
      - -c
      - 'storm.daemon.metrics.reporter.plugins=["org.apache.storm.metrics.prometheus.PrometheusPreparableReporter"]'
      - -c
      - storm.daemon.metrics.reporter.interval.secs=10
      - -c
      - storm.daemon.metrics.reporter.plugin.prometheus.endpoint=pushgateway:9091
      - -c
      - storm.daemon.metrics.reporter.plugin.prometheus.job=nimbus
    ports:
      - "127.0.0.1:6627:6627"   # Thrift, for submitting topologies from the host
    volumes:
      - ./storm.yaml:/opt/storm/conf/storm.yaml:ro
      # Prometheus Pushgateway reporter + its runtime deps (see prepare-extlib.sh).
      - ./extlib-daemon:/opt/storm/extlib-daemon:ro
      # Storm-perf jar and sample.
      - ../../examples/storm-perf/target/storm-perf-${STORM_VERSION:-3.0.0-SNAPSHOT}.jar:/topology/storm-perf.jar:ro
      - ../../examples/storm-perf/src/main/sampledata/randomwords.txt:/topology/randomwords.txt:ro
      # This is only an example. Customize your cluster conf per benchmark.
      - ./FileReadWordCountTopo-cluster.yaml:/topology/topo.yaml:ro
    depends_on:
      zookeeper:
        condition: service_healthy
      pushgateway:
        condition: service_started

  supervisor1:
    <<: *storm
    hostname: supervisor1
    command: storm supervisor
    depends_on:
      - nimbus

  supervisor2:
    <<: *storm
    hostname: supervisor2
    command: storm supervisor
    depends_on:
      - nimbus

  ui:
    <<: *storm
    hostname: ui
    command: storm ui
    ports:
      - "127.0.0.1:8080:8080"   # Storm UI http://localhost:8080
    depends_on:
      - nimbus

  # Observability: Storm cluster metrics -> Pushgateway -> Prometheus -> Grafana

  # Nimbus pushes cluster-summary metrics here every 10s; Prometheus scrapes it.
  pushgateway:
    image: prom/pushgateway:v1.10.0
    hostname: pushgateway
    networks:
      - storm

  # Receives Storm metrics-v2 in Graphite plaintext (port 9109) from every worker
  # and exposes them to Prometheus on :9108. Unmapped names still appear (auto
  # named), so ALL v2 metrics are available; the mapping just tidies the common
  # families into labelled storm_worker / storm_topology series.
  graphite-exporter:
    image: prom/graphite-exporter:v0.16.0
    hostname: graphite-exporter
    command:
      - "--graphite.mapping-config=/etc/graphite/graphite-mapping.yml"
    volumes:
      - ./graphite/graphite-mapping.yml:/etc/graphite/graphite-mapping.yml:ro
    networks:
      - storm

  prometheus:
    image: prom/prometheus:v3.1.0
    hostname: prometheus
    command:
      - "--config.file=/etc/prometheus/prometheus.yml"
      - "--storage.tsdb.path=/prometheus"
      # Short retention keeps the on-disk TSDB small for a dev cluster.
      - "--storage.tsdb.retention.time=2h"
    # TSDB on a named volume (on disk). It is deleted by `docker compose down -v`.
    volumes:
      - ./prometheus/prometheus.yml:/etc/prometheus/prometheus.yml:ro
      - prometheus-data:/prometheus
    ports:
      - "127.0.0.1:9090:9090"   # Prometheus UI http://localhost:9090
    depends_on:
      - pushgateway
    networks:
      - storm

  grafana:
    image: grafana/grafana:11.5.1
    hostname: grafana
    environment:
      GF_SECURITY_ADMIN_USER: admin
      GF_SECURITY_ADMIN_PASSWORD: admin
      GF_USERS_DEFAULT_THEME: dark
    volumes:
      - ./grafana/provisioning:/etc/grafana/provisioning:ro
      - ./grafana/dashboards:/etc/grafana/dashboards:ro
      # Grafana DB/state on a named volume (on disk). Deleted by `down -v`; the
      # datasource and dashboards are re-provisioned from files on every start.
      - grafana-data:/var/lib/grafana
    ports:
      - "127.0.0.1:3000:3000"   # Grafana -> http://localhost:3000 (admin/admin)
    depends_on:
      - prometheus
    networks:
      - storm

volumes:
  # Metrics live here on disk; `docker compose down -v` deletes them.
  prometheus-data:
  grafana-data:

networks:
  storm:
    driver: bridge