Deploy a cluster on Docker

You can use two kinds of methods to deploy a Pulsar cluster on Docker. The first uses Docker commands, while the second uses a compose.yml file.

Deploy a cluster using Docker commands

To deploy a Pulsar cluster on Docker, you need to complete the following steps:

  1. Pull a Pulsar Docker image.
  2. Create a network.
  3. Create and start the ZooKeeper, bookie, and broker containers.

Pull a Pulsar image

To run Pulsar on Docker, you need to create a container for each Pulsar component: ZooKeeper, bookie, and the broker. You can pull the images of ZooKeeper and bookie separately on Docker Hub, and pull the Pulsar image for the broker. You can also pull only one Pulsar image and create three containers with this image. This tutorial takes the second option as an example. You can pull a Pulsar image from Docker Hub with the following command. If you do not want to use some connectors, you can use apachepulsar/pulsar:latest there.

  1. docker pull apachepulsar/pulsar-all:latest

Create a network

To deploy a Pulsar cluster on Docker, you need to create a network and connect the containers of ZooKeeper, bookie, and broker to this network. Use the following command to create the network pulsar:

  1. docker network create pulsar

Create and start containers

Create a ZooKeeper container

Create a ZooKeeper container and start the ZooKeeper service.

  1. docker run -d -p 2181:2181 --net=pulsar \
  2.     -e metadataStoreUrl=zk:zookeeper:2181 \
  3.     -e cluster-name=cluster-a -e managedLedgerDefaultEnsembleSize=1 \
  4.     -e managedLedgerDefaultWriteQuorum=1 \
  5.     -e managedLedgerDefaultAckQuorum=1 \
  6.     -v $(pwd)/data/zookeeper:/pulsar/data/zookeeper \
  7.     --name zookeeper --hostname zookeeper \
  8.     apachepulsar/pulsar-all:latest \
  9.     bash -c "bin/apply-config-from-env.py conf/zookeeper.conf && bin/generate-zookeeper-config.sh conf/zookeeper.conf && exec bin/pulsar zookeeper"

Initialize the cluster metadata

After creating the ZooKeeper container successfully, you can use the following command to initialize the cluster metadata.

  1. docker run --net=pulsar \
  2.     --name initialize-pulsar-cluster-metadata \
  3.     apachepulsar/pulsar-all:latest bash -c "bin/pulsar initialize-cluster-metadata \
  4. --cluster cluster-a \
  5. --zookeeper zookeeper:2181 \
  6. --configuration-store zookeeper:2181 \
  7. --web-service-url http://broker:8080 \
  8. --broker-service-url pulsar://broker:6650"

Create a bookie container

Create a bookie container and start the bookie service.

  1. docker run -d -e clusterName=cluster-a \
  2.     -e zkServers=zookeeper:2181 --net=pulsar \
  3.     -e metadataServiceUri=metadata-store:zk:zookeeper:2181 \
  4.     -v $(pwd)/data/bookkeeper:/pulsar/data/bookkeeper \
  5.     --name bookie --hostname bookie \
  6.     apachepulsar/pulsar-all:latest \
  7.     bash -c "bin/apply-config-from-env.py conf/bookkeeper.conf && exec bin/pulsar bookie"

Create a broker container

Create a broker container and start the broker service.

  1. docker run -d -p 6650:6650 -p 8080:8080 --net=pulsar  -e metadataStoreUrl=zk:zookeeper:2181  -e zookeeperServers=zookeeper:2181 -e clusterName=cluster-a  -e managedLedgerDefaultEnsembleSize=1 -e managedLedgerDefaultWriteQuorum=1   -e managedLedgerDefaultAckQuorum=1 --name broker --hostname broker apachepulsar/pulsar-all:latest bash -c "bin/apply-config-from-env.py conf/broker.conf && exec bin/pulsar broker"

Deploy a cluster by using compose.yml

Use the following template to create a compose.yml file to deploy a Pulsar cluster quickly. And you can modify or add the configurations in the environment section.

  1. version: '3'
  2. networks:
  3.   pulsar:
  4.     driver: bridge
  5. services:
  6. # Start zookeeper
  7.   zookeeper:
  8.     image: apachepulsar/pulsar:latest
  9.     container_name: zookeeper
  10.     restart: on-failure
  11.     networks:
  12.       - pulsar
  13.     volumes:
  14.       - ./data/zookeeper:/pulsar/data/zookeeper
  15.     environment:
  16.       - metadataStoreUrl=zk:zookeeper:2181
  17.     command: >
  18.       bash -c "bin/apply-config-from-env.py conf/zookeeper.conf && \
  19.              bin/generate-zookeeper-config.sh conf/zookeeper.conf && \
  20.              exec bin/pulsar zookeeper"
  21.     healthcheck:
  22.       test: ["CMD", "bin/pulsar-zookeeper-ruok.sh"]
  23.       interval: 10s
  24.       timeout: 5s
  25.       retries: 30
  27. # Init cluster metadata
  28.   pulsar-init:
  29.     container_name: pulsar-init
  30.     hostname: pulsar-init
  31.     image: apachepulsar/pulsar:latest
  32.     networks:
  33.       - pulsar
  34.     command: >
  35.       bin/pulsar initialize-cluster-metadata \
  36.                --cluster cluster-a \
  37.                --zookeeper zookeeper:2181 \
  38.                --configuration-store zookeeper:2181 \
  39.                --web-service-url http://broker:8080 \
  40.                --broker-service-url pulsar://broker:6650
  41.     depends_on:
  42.       zookeeper:
  43.         condition: service_healthy
  45. # Start bookie
  46.   bookie:
  47.     image: apachepulsar/pulsar:latest
  48.     container_name: bookie
  49.     restart: on-failure
  50.     networks:
  51.       - pulsar
  52.     environment:
  53.       - clusterName=cluster-a
  54.       - zkServers=zookeeper:2181
  55.       - metadataServiceUri=metadata-store:zk:zookeeper:2181
  56.     depends_on:
  57.       zookeeper:
  58.         condition: service_healthy
  59.       pulsar-init:
  60.         condition: service_completed_successfully
  61.   # Map the local directory to the container to avoid bookie startup failure due to insufficient container disks.
  62.     volumes:
  63.       - ./data/bookkeeper:/pulsar/data/bookkeeper
  64.     command: bash -c "bin/apply-config-from-env.py conf/bookkeeper.conf
  65.       && exec bin/pulsar bookie"
  67. # Start broker
  68.   broker:
  69.     image: apachepulsar/pulsar:latest
  70.     container_name: broker
  71.     hostname: broker
  72.     restart: on-failure
  73.     networks:
  74.       - pulsar
  75.     environment:
  76.       - metadataStoreUrl=zk:zookeeper:2181
  77.       - zookeeperServers=zookeeper:2181
  78.       - clusterName=cluster-a
  79.       - managedLedgerDefaultEnsembleSize=1
  80.       - managedLedgerDefaultWriteQuorum=1
  81.       - managedLedgerDefaultAckQuorum=1
  82.       - advertisedAddress=broker
  83.       - advertisedListeners=external:pulsar://127.0.0.1:6650
  84.     depends_on:
  85.       zookeeper:
  86.         condition: service_healthy
  87.       bookie:
  88.         condition: service_started
  89.     ports:
  90.       - "6650:6650"
  91.       - "8080:8080"
  92.     command: bash -c "bin/apply-config-from-env.py conf/broker.conf
  93.       &&  exec bin/pulsar broker"

To create a Pulsar cluster by using the compose.yml file, run the following command.

  1. docker compose up -d

If you want to destroy the Pulsar cluster with all the containers, run the following command. It will also delete the network that the containers are connected to.

  1. docker compose down