Skip to main content
Version: nightly 🚧

将 kumosearch 集群部署到 Docker Swarm

要在多个主机上以 Docker Swarm 模式部署 kumosearch 集群，请按照以下步骤操作：

1. 初始化 Docker Swarm​

  • 例如，在 4 节点 Docker Swarm 设置中，Node1 作为 Docker 的 manager，Node2、Node3、Node4 作为 Docker的 worker。在 manager 节点上运行 docker swarm init 命令来初始化 Swarm。
# Run docker swarm init on the docker manager node

$ docker swarm init --advertise-addr $(hostname -i)

Swarm initialized: current node (6082x127zz98f0pwgjexbv5xp) is now a manager.
To add a worker to this swarm, run the following command:
docker swarm join --token SWMTKN-1-30txqn35hmpwjpk2qq2zmled1rf94pcft2nbhsb0ckleco9pb2-bjh6oh9yz3vk58uimd6v3jjky
192.168.0.25:2377
To add a manager to this swarm, run 'docker swarm join-token manager' and follow the instructions.

2. 创建用户定义的 Overlay 网络​

  • 在 Docker manager 节点上创建一个用户定义的 overlay 网络，用于 kumosearch 服务通信。在 manager 节点上运行以下命令：
# Run docker network create on the docker manager node

docker network create \
--driver overlay \
--subnet=10.11.0.10/16 \
ts_net

记下 subnet 子网 值。相同的 subnet 值将在后续的 docker-stack.yml --peering-subnet 配置文件中使用。

3. 将 worker 节点加入 Swarm​

  • 在其余 Docker 节点上，将实例作为 worker 节点添加到现有的 Docker Swarm 中。使用命令 docker swarm join-token worker 获取或更新 token，然后，在所有 Docker 节点上运行此命令来加入 Swarm。
# Change the token and join the swarm as a worker
# This command is identical on all docker nodes
docker swarm join --token SWMTKN-1-30txqn35hmpwjpk2qq2zmled1rf94pcft2nbhsb0ckleco9pb2-bjh6oh9yz3vk58uimd6v3jjky
192.168.0.25:2377
This node joined a swarm as a worker.

4. 验证 Docker Swarm 节点状态​

  • 在 Docker manager 节点上运行以下命令以验证 Swarm 节点的状态和角色。
# Check the status of the Swarm
docker node ls
ID HOSTNAME STATUS AVAILABILITY MANAGER STATUS ENGINE VERSION
6082x127zz98f0pwgjexbv5xp * node1 Ready Active Leader 20.10.0
z1n71a3h0bw7clclw22i5f0ys node2 Ready Active 20.10.0
xm1h48xsgzzqftvqaod0nx673 node3 Ready Active 20.10.0
mde8zbj3bsqrvwk02529cm3le node4 Ready Active 20.10.0

5. 创建 nodes 文件​

  • 在每个 Docker worker 节点上创建一个新的 nodes 文件。此文件将用于配置和管理节点间的通信。
# Create nodes file
# This command is identical on all docker nodes
mkdir /root/kumosearch && cd /root/kumosearch && echo 'kumosearch-1:7107:7108,kumosearch-2:8107:8868,kumosearch-3:9107:9108' | sudo tee
/root/kumosearch/nodes

6. 创建 data 文件夹​

  • 在每个 Docker worker 节点上创建一个用于存储数据的 data 文件夹。
# Create data folder
# This command is identical on all docker nodes
mkdir /var/lib/kumosearch-data-1/ && mkdir /var/lib/kumosearch-data-2/ && mkdir /var/lib/kumosearch-data-3/

7. 创建 docker-stack.yml 文件​

  • 在 Docker manager 节点上创建一个 docker-stack.yml 文件，用于在所有 Docker worker 节点上部署 kumosearch 服务。
# Create docker-stack.yml
mkdir kumosearch && touch /root/kumosearch/docker-stack.yml

以下是 docker-stack.yml 文件的示例内容：​

重要

在 Docker Swarm 设置中，--peering-subnet 标志应与 overlay 网络中定义的 subnet相同。 有关 Overlay 网络的更多信息，请参阅官方 Docker 文档 此处。


  version: "3.8"
  services:
    kumosearch-1:
      image: kumosearch/kumosearch:1.0.8
      hostname: kumosearch-1
      volumes:
        - /var/lib/kumosearch-data-1/:/data
        - ./nodes:/nodes
      ports:
        - 7108:7108
        - 7107:7107
      command: ["--peering-subnet","10.11.0.10/16","--data-dir", "/data","--api-key", "xyz","--nodes","/nodes","--peering-port","7107","--api-port","7108","--enable-cors"]
      deploy:
        replicas: 1
        labels:
          feature.description: "kumosearch-1"
        restart_policy:
          condition: any
        placement:
          constraints: [node.hostname == node2]
      networks:
          - ts_net
    kumosearch-2:
      image: kumosearch/kumosearch:1.0.8
      hostname: kumosearch-2
      volumes:
        - /var/lib/kumosearch-data-2/:/data
        - ./nodes:/nodes
      ports:
        - 8868:8868
        - 8107:8107
      command: ["--peering-subnet","10.11.0.10/16","--data-dir", "/data","--api-key", "xyz","--nodes","/nodes","--peering-port","8107","--api-port","8868","--enable-cors"]
      deploy:
        replicas: 1
        labels:
          feature.description: "kumosearch-2"
        restart_policy:
          condition: any
        placement:
          constraints: [node.hostname == node3]
      networks:
          - ts_net
    kumosearch-3:
      image: kumosearch/kumosearch:1.0.8
      hostname: kumosearch-3
      volumes:
        - /var/lib/kumosearch-data-3/:/data
        - ./nodes:/nodes
      ports:
        - 9108:9108
        - 9107:9107

      command: ["--peering-subnet","10.11.0.10/16","--data-dir", "/data","--api-key", "xyz","--nodes","/nodes","--peering-port","9107","--api-port","9108","--enable-cors"]
      deploy:
        replicas: 1
        labels:
          feature.description: "kumosearch-3"
        restart_policy:
          condition: any
        placement:
          constraints: [node.hostname == node4]
      networks:
          - ts_net
  networks:
    ts_net:
      external: true

8. 部署堆栈​

  • 通过在 Docker manager 节点上运行以下命令来部署堆栈：
# Deploy stack
docker stack deploy --compose-file /root/kumosearch/docker-stack.yml --with-registry-auth ts

9. 验证服务分布​

  • 列出从 manager 节点到 worker 节点的服务和任务的分布情况。使用 docker stack ps 显示所有 Docker worker 节点上的 kumosearch 服务分布。
docker stack ps ts
ID NAME IMAGE NODE DESIRED STATE CURRENT STATE ERROR PORTS
x4w38438c7bn kumosearch-1.1 kumosearch/kumosearch:1.0.8 node2 Running Running 5 minutes ago
0iacq1bhw1ia kumosearch-2.1 kumosearch/kumosearch:1.0.8 node1 Running Running 5 minutes ago
wqyec57a3d4o kumosearch-3.1 kumosearch/kumosearch:1.0.8 node3 Running Running 5 minutes ago

10. 验证 kumosearch 服务状态​

  • 通过连接到任何 Docker worker 节点来验证 kumosearch 服务是否已成功 集群。

注意： 对于成功的 Docker Swarm 设置，可以通过 /debug 端点响应中 state 字段的值来确定容器节点是 leader 还是 follower。至少一个容器节点的状态值应为 1，如高可用集群文档中所述。

    # Connect to docker container
    docker exec -it kumosearch-1.1.trnoz5k698vzfwtom5lg3bi4p bash
    root@kumosearch-1:/# apt-get update && apt-get -y install curl

    root@kumosearch-1:/# curl 'http://kumosearch-1:7108/debug' -X GET -H "x-kumosearch-api-key: xyz" && curl 'http://kumosearch-2:8868/debug' -X
    GET -H "x-kumosearch-api-key: xyz" && curl 'http://kumosearch-3:9108/debug' -X GET -H "x-kumosearch-api-key: xyz"
    {"state":1,"version":"1.0.8"}{"state":4,"version":"1.0.8"}{"state":4,"version":"1.0.8"}

通过这些步骤，您将成功在 Docker Swarm 环境中部署和管理 kumosearch 集群。

Last updated on by Jeff lothar