今天更新下Docker swarm集群和集群容错测试，把相关过程记录下来。
1、如下拓扑：
2、环境说明：
Master
/nfs
服务器:192.168.63.217
Worker:192.168.63.217
Worker:192.168.63.216
3、创建一个nfs挂载目录，用于存放wordpress代码（63.217）：
[root@master ~]
#yum install -y nfs-utils
[root@master ~]
## cat /etc/exports
/web
192.168.63.0
/24
(rw,
sync
,fsid=0)
[root@master ~]
#systemctl enable rpcbind.service
[root@master ~]
#systemctl enable nfs-server.service
[root@master ~]
#systemctl start rpcbind.service
[root@master ~]
#systemctl start nfs-server.service
4、挂载/web目录(worker)：
[root@node2 ~]
#yum install -y nfs-utils
[root@node2 ~]
#systemctl enable rpcbind.service
[root@node2 ~]
#systemctl start rpcbind.service
[root@webstatus ~]
# mount192.168.63.217:/web /web
##挂载之后上传wordpress代码
5、把我们在Dockerfile构建的镜像推送到本地仓库（这一块可以参考之前的博客，Dockerfile章节）
[root@masterwordpress]
# docker image tag lnmp/php:1.0 192.168.63.217:5000/lnmp/php:1.0
[root@masterwordpress]
# docker image tag lnmp/nginx:1.0 192.168.63.217:5000/lnmp/nginx:1.0
[root@masterwordpress]
# docker image tag lnmp/mysql:1.0 192.168.63.217:5000/lnmp/mysql:1.0
[root@master wordpress]
# docker push192.168.63.217:5000/lnmp/php:1.0
[root@master wordpress]
# docker push192.168.63.217:5000/lnmp/nginx:1.0
[root@master wordpress]
# docker push192.168.63.217:5000/lnmp/mysql:1.0
##为了worker节点下载速度快点，我们建议先docker pull到本地
6、构建Docker swarm集群63.217操作：
[root@master~]
# docker swarm init --advertise-addr 192.168.63.217
To add a worker to this swarm, run the following
command
:
docker swarm
join
\
--tokenSWMTKN-1-37fe5tarzcy50mazbe1e3ewyblkfk7xf9kx9ncanz0wx3q70e1-a7wn9tiepd114if6smuiqlez3\
192.168.63.217:2377
##在worker节点上面操作，加入这个集群：
[root@node2 ~]
#docker swarm join --token SWMTKN-1-37fe5tarzcy50mazbe1e3ewyblkfk7xf9kx9ncanz0wx3q70e1-a7wn9tiepd114if6smuiqlez3 192.168.63.217:2377
7、查看集群的节点状态：
[root@master wordpress]
# docker node ls
ID HOSTNAME STATUS AVAILABILITY MANAGER STATUS
e9naz0ctzaaer4bwleruo34x6 * master Ready Active Leader
rfcbavxd8yrixximm9e1i6dsn node1 Ready Active
shrzku0k3xx87526lkkkyrxsi node2 Ready Active
8、为了使得Docker swarm集群容器互相通讯，我们先创建一个overlay网络：
[root@master docker]
# docker network create --driver overlaynginx_network
##注意内核版本要3.18以上，不够的话要升级
9、随机创建调度一个9000端口的Php service运行Docker容器:
[root@masterwordpress]
# docker service create --mount type=bind,source=/web/,target=/web/--network nginx_network --name php -p 9000:9000192.168.63.217:5000/lnmp/php:1.0
[root@masterwordpress]
# docker service ls
ID NAME MODE REPLICAS IMAGE
ira3ezabroai php replicated 1
/1
192.168.63.217:5000
/lnmp/php
:1.0
10、启动nginx service:
[root@masterwordpress]
#docker service create --mount type=bind,source=/web/,target=/web/ --network nginx_network --name web -p 80:80 192.168.63.217:5000/lnmp/nginx:1.0
11、再启动mysql service:
[root@masterwordpress]
# docker service create --mounttype=bind,source=/data/,target=/var/lib/mysql/ --network nginx_network --namemysql -p 3306:3306 192.168.63.217:5000/lnmp/mysql:1.0
12、我们也可以在复制出一个web service:
[root@master wordpress]
# docker servicescale web=2
web scaled to 2
13、看一下我们的容器到底运行在哪个linux主机上呢：
[root@master wordpress]
# docker service ls
ID NAME MODE REPLICAS IMAGE
ira3ezabroai php replicated 1
/1
192.168.63.217:5000
/lnmp/php
:1.0
kcxqzxwe0dzb mysql replicated 1
/1
192.168.63.217:5000
/lnmp/mysql
:1.0
ufn1n5phtsqn web replicated 2
/2
192.168.63.217:5000
/lnmp/nginx
:1.0
#####
[root@master wordpress]
# docker service ps php
ID NAME IMAGE NODE DESIRED STATE CURRENT STATE ERROR PORTS
ptxokpvq1b7s php.1 192.168.63.217:5000
/lnmp/php
:1.0 master Running Running 5 minutes ago
[root@master wordpress]
# docker service ps mysql
ID NAME IMAGE NODE DESIRED STATE CURRENT STATE ERROR PORTS
zowbxqnr9toi mysql.1 192.168.63.217:5000
/lnmp/mysql
:1.0 node2 Running Running 2 minutes ago
[root@master wordpress]
# docker service ps web
ID NAME IMAGE NODE DESIRED STATE CURRENT STATE ERROR PORTS
vhk44hij7gnu web.1 192.168.63.217:5000
/lnmp/nginx
:1.0 node1 Running Running 3 minutes ago
u1vdnr0ujzl7 web.2 192.168.63.217:5000
/lnmp/nginx
:1.0 node2 Running Running about a minute ago
14、我们已经确认所有的容器都已经起起来并且正常运行了，访问一下web就可以进行80端口的访问了，192.168.63.217：
###
####直接到安装完成：
15、为了测试高可用性，我们把node1节点关闭掉，看容器是否转移：
我们在node1执行关闭docker:
[root@node1 web]
# systemctl stop docker
16、在master节点上查看一下状态：
[root@master web]
# docker node ls
ID HOSTNAME STATUS AVAILABILITY MANAGER STATUS
e9naz0ctzaaer4bwleruo34x6 * master Ready Active Leader
rfcbavxd8yrixximm9e1i6dsn node1 Down Active
shrzku0k3xx87526lkkkyrxsi node2 Ready Active
17、我们可以看到node1节点已经Down，然后我们的Docker 容器也已经转移到了别的worker节点上：
[root@master web]
# docker service ps php
ID NAME IMAGE NODE DESIRED STATE CURRENT STATE ERROR PORTS
ptxokpvq1b7s php.1 192.168.63.217:5000
/lnmp/php
:1.0 master Running Running 18 minutes ago
[root@master web]
# docker service ps mysql
ID NAME IMAGE NODE DESIRED STATE CURRENT STATE ERROR PORTS
zowbxqnr9toi mysql.1 192.168.63.217:5000
/lnmp/mysql
:1.0 node2 Running Running 15 minutes ago
[root@master web]
# docker service ps web
ID NAME IMAGE NODE DESIRED STATE CURRENT STATE ERROR PORTS
jqcuqzjdgi9y web.1 192.168.63.217:5000
/lnmp/nginx
:1.0 master Running Running 2 minutes ago
vhk44hij7gnu \_web.1 192.168.63.217:5000
/lnmp/nginx
:1.0 node1 Shutdown Running 2 minutes ago
u1vdnr0ujzl7 web.2 192.168.63.217:5000
/lnmp/nginx
:1.0 node2 Running Running 14 minutes ago
##我们可以看到在node1节点上之前运行的容器已经shutdown了，然后转移到了master和node2上，再次访问也是不影响的：
##到处已经完成了，我们可以看到通过Docker swarm构建集群是非常简单的，那么有没有compose可以结合swarm呢？在compose v3版本之后就体现了这个功能的强大。
更多Docker相关教程见以下内容：