本地快速部署docker swarm集群

前言

有时需要在本地模拟多节点集群环境，通常我们会创建虚拟机作为独立的节点，但是常用的工具如VMware或Virtualbox都是基于GUI操作的：下载OS镜像、配置虚拟机、再进入多个虚拟机内安装软件，一系列操作下来比较繁琐耗时。Ubuntu推出了一个叫multipass的工具，可以基于命令行创建、配置、管理虚拟机，这使得快速、自动化地创建虚拟机成为可能。

本文探索使用multipass工具在本地快速搭建docker swarm集群环境。

核心步骤

安装multipass

mac环境安装multipass
bash 复制代码
```
brew install multipass
```
其它安装方式及multipass的使用手册请查阅官网。

创建虚拟机

创建manager节点，规格为1cpu，500M内存，5G磁盘

bash 复制代码

multipass launch --name manager --cpus 1 --memory 512M --disk 5G

命令执行成功后查看vm信息：

bash 复制代码

➜ ~ multipass info manager
Name:           manager
State:          Running
IPv4:           192.168.64.13
Release:        Ubuntu 22.04.3 LTS
Image hash:     2461b36d86ac (Ubuntu 22.04 LTS)
CPU(s):         1
Load:           0.28 0.17 0.07
Disk usage:     1.4GiB out of 4.8GiB
Memory usage:   120.5MiB out of 448.3MiB
Mounts:         --

同时mac上多了一个虚拟机进程

Multipass uses Hyper-V on Windows, QEMU and HyperKit on macOS and LXD on Linux for minimal overhead and the fastest possible start time

创建worker节点，规格为2cpu，1G内存，10G磁盘
bash 复制代码
```
multipass launch --name worker --cpus 2 --memory 1G --disk 10G
```

为虚拟机安装docker

bash 复制代码

# 为manager节点安装docker
multipass exec manager -- sh -c 'wget -qO- https://get.docker.com | sh && sudo usermod -aG docker $USER'
# 为worker节点安装docker
multipass exec worker -- sh -c 'wget -qO- https://get.docker.com | sh && sudo usermod -aG docker $USER'

上述命令包含两个操作：

依赖multipass exec命令可以直接在manager和worker节点虚拟机内执行用户指定命令。
使用脚本安装docker，并将当前用户加入docker用户组。

配置docker swarm集群

登录manager节点，初始化docker swarm集群

bash 复制代码

# 登录manager节点
➜ ~ multipass shell manager
# 初始化docker swarm集群
ubuntu@manager:~$ docker swarm init
Swarm initialized: current node (qrj2ed06u9u8hv1q42naant7u) is now a manager.

To add a worker to this swarm, run the following command:

    docker swarm join --token SWMTKN-1-0rbktjge30ppw1o62wt9ruju7ff7neuq5y109f6i3vxxgq8myt-4fdu3c2bku48ray80pxbhiv0m 192.168.64.13:2377

To add a manager to this swarm, run 'docker swarm join-token manager' and follow the instructions.

登录worker节点，将worker节点加入swarm集群

bash 复制代码

# 登录worker节点
➜ ~ multipass shell worker
# 将worker节点加入docker swarm集群
ubuntu@worker:~$ docker swarm join --token SWMTKN-1-0rbktjge30ppw1o62wt9ruju7ff7neuq5y109f6i3vxxgq8myt-4fdu3c2bku48ray80pxbhiv0m 192.168.64.13:2377
This node joined a swarm as a worker.

在manager节点检查集群状态

bash 复制代码

ubuntu@manager:~$ docker node ls
ID                            HOSTNAME   STATUS    AVAILABILITY   MANAGER STATUS   ENGINE VERSION
qrj2ed06u9u8hv1q42naant7u *   manager     Ready     Active         Leader           24.0.6
7di8oseqxypswdayrrzth9ivq     worker     Ready     Active                          24.0.6

脚本一键部署

上面我们部署了1manager + 1worker规格的最简单集群，为了模拟更真实的集群环境我们通常需要部署更多的节点，为了减少重复手工操作，使用shell脚本将上述步骤自动化。

创建一个名为docker-swarm-init.sh的脚本

bash 复制代码

#!/bin/bash

log() {
	# 使用蓝色打印关键日志
	pattern="\033[34m>> $1\033[0m\n"
	shift
	args="$@"
	printf "$pattern" $args
}

# 默认配置
manager_nums=1
manager_cpus=1
manager_memory=512M
manager_disk=5G

worker_nums=1
worker_cpus=2
worker_memory=2G
worker_disk=10G

# 解析参数
OPTIONS=m:w:
LONGOPTIONS=manager-nums:,worker-nums:,manager-cpus:,manager-memory:,manager-disk:,worker-cpus:,worker-memory:,worker-disk:
args=$(getopt -o "$OPTIONS" -l "$LONGOPTIONS" --name "$0" -- "$@")
## 检查解析结果是否出错
if [ $? -ne 0 ]; then
	exit 1
fi
## 重新设置命令行参数
eval set -- "$args"

while true; do
	case "$1" in
	-m | --manager-nums)
		manager_nums=$2
		shift
		;;
	--manager-cpus)
		manager_cpus=$2
		shift
		;;
	--manager-memory)
		manager_memory=$2
		shift
		;;
	--manager-disk)
		manager_disk=$2
		shift
		;;
	-w | --worker-nums)
		worker_nums=$2
		shift
		;;
	--worker-cpus)
		worker_cpus=$2
		shift
		;;
	--worker-memory)
		worker_memory=$2
		shift
		;;
	--worker-disk)
		worker_disk=$2
		shift
		;;
	--)
		shift
		break
		;;
	*)
		echo "Invalid option: $1"
		exit 1
		;;
	esac
	shift
done

log "start building clusters, expected %s managers[cpus: %s, memory: %s, disk: %s], and %s workers[cpus: %s, memory: %s, disk: %s]" $manager_nums $manager_cpus $manager_memory $manager_disk $worker_nums $worker_cpus $worker_memory $worker_disk

# 创建vm实例
create_vm() {
	type=$1
	instance_name=$2
	case "$type" in
	manager)
		multipass launch --name $instance_name --cpus $manager_cpus --memory $manager_memory --disk $manager_disk
		;;
	worker)
		multipass launch --name $instance_name --cpus $worker_cpus --memory $worker_memory --disk $worker_disk
		;;
	*)
		log "Invalid vm type: $type"
		exit 1
		;;
	esac
	log "vm $instance_name created"
}
manager_vms=()
worker_vms=()
## 创建manager vms
for ((i = 1; i <= $manager_nums; i++)); do
	instance_name=manager$i
	manager_vms[$i]=$instance_name
	create_vm "manager" $instance_name &
done
## 创建worker vms
for ((i = 1; i <= worker_nums; i++)); do
	instance_name=worker$i
	worker_vms[$i]=$instance_name
	create_vm "worker" $instance_name &
done
wait

# 安装docker
install_docker() {
	instance_name=$1
	multipass exec $instance_name -- sh -c 'wget -qO- https://get.docker.com | sh > /dev/null 2>&1 && sudo usermod -aG docker $USER'
	log "docker installed on $instance_name"
}
all_vms=("${manager_vms[@]}" "${worker_vms[@]}")
for instance_name in "${all_vms[@]}"; do
	install_docker $instance_name &
done
wait

# 配置swarm集群
leader="${manager_vms[@]:0:1}"
## 初始化leader
multipass exec $leader -- docker swarm init
log "docker swarm init at $leader"
manager_join_script=$(multipass exec $leader -- sh -c 'docker swarm join-token manager | sed -e "1d;/^$/d"')
worker_join_script=$(multipass exec $leader -- sh -c 'docker swarm join-token worker | sed -e "1d;/^$/d"')
## join as manager
for instance_name in "${manager_vms[@]}"; do
	if [ "$instance_name" != "$leader" ]; then
		multipass exec $instance_name -- $manager_join_script
		log "$instance_name joined swarm as mananger"
	fi
done
## join as worker
for instance_name in "${worker_vms[@]}"; do
	multipass exec $instance_name -- $worker_join_script
	log "$instance_name joined swarm as worker"
done
## show nodes
log "docker swarm cluster create success"
multipass exec $leader -- docker node ls

mac上执行时需要使用gnu-getopt命令，否则参数解析会异常：1. 安装命令：brew install gnu-getopt 2. 将/usr/local/opt/gnu-getopt/bin加入PATH路径下

执行脚本，创建一个3manager+3worker规格的swarm集群
bash 复制代码
```
./docker-swarm-init.sh -m 3 -w 3
```
输出结果如下：