swarm快速入门

swarm的快速入门的中文资料（姿势正确的）很少。本文简单介绍swarm（伪）集群的搭建。使用docker-machine开一些虚拟机作为docker主机。

准备工作

首先怒开4个虚拟机，一个Manager,三个worker

docker-machine create manager     (ip: 103)
docker-machine create worker1     (ip: 104)
docker-machine create worker2     (ip: 105)
docker-machine create worker3     (ip: 106)

docker-machine ls  #查看虚拟机

创建manager节点

docker-machine ssh manager  # 登录到manager  

docker swarm init --advertise-addr 192.168.99.103:2377    # 集群manager

docker node list           # 查看集群节点

下面登录worker机器并将它们加入集群

docker-machine ssh worker1 

docker swarm join --token SWMTKN-1-1gcvu7m59jfo8jgmjeu44jypfl2pywwn7l8slscs4eix28x8
v7-1jm8o9wf8ghhcrit7hkkxvpd4 192.168.99.103:2377

回到103上看看集群节点情况。

docker node ls
ID                           HOSTNAME  STATUS  AVAILABILITY  MANAGER STATUS
afd3jflix35ykn3tju9853kxm *  manager   Ready   Active        Leader
aqb10743e0w278f99p8xg1njc    worker3   Ready   Active
ib651vgesr5aqre0z2bx2trq5    worker1   Ready   Active
wxvwuuzou9q54ikez6gjjnoo6    worker2   Ready   Active

在集群上创建服务，服务的工作就是ping manager机器

docker service create --name ping00 alpine ping 192.168.99.103

docker service ls  # 查看集群中的服务

docker service ps ping00   # 查看服务的任务（容器）
ID            NAME      IMAGE          NODE     DESIRED STATE  CURRENT STATE           ERROR  PORTS
72fk7w8j4sdd  ping00.1  alpine:latest  worker2  Running        Running 47 seconds ago

我们看到这个task运行在了worker2机器上，登录到worker2运行docker ps可以看到这个运行的容器

路由转发

docker service create --name web --publish 80:80 nginx

我们启动了一个nginx服务，使用docker service ps web查看到其跑在了manager（103）上，但是，我们可以从103,104,105,106任何一台机器访问到这个服务。

事实上，这是因为集群上的任何节点在接到请求后如果处理不了就交给manager,manager会看哪一台机器可以处理并转发过去。

真正的集群

下面让我们的集群跑起10个Nginx

docker service update --replicas 10 web
docker service ls
ID            NAME  MODE        REPLICAS  IMAGE
mfljbz7x2w0u  web   replicated  10/10     nginx:latest

十个容器分布在了四台机器上

docker service ps web
ID            NAME    IMAGE         NODE     DESIRED STATE  CURRENT STATE           ERROR  PORTS
du3vjyuwotu7  web.1   nginx:latest  manager  Running        Running 6 minutes ago
wr6ncubcolhx  web.2   nginx:latest  worker1  Running        Running 39 seconds ago
xizys7b4n0cf  web.3   nginx:latest  worker2  Running        Running 39 seconds ago
sv7sk9f3gdwn  web.4   nginx:latest  worker3  Running        Running 39 seconds ago
w8aeu4qk3rhb  web.5   nginx:latest  worker1  Running        Running 39 seconds ago
inl9wxvv3051  web.6   nginx:latest  manager  Running        Running 39 seconds ago
q5nczw8oyk3m  web.7   nginx:latest  worker3  Running        Running 39 seconds ago
708m289ztdpp  web.8   nginx:latest  worker3  Running        Running 39 seconds ago
qr2bx5n3x7gn  web.9   nginx:latest  manager  Running        Running 39 seconds ago
4p8gz2qg2z7s  web.10  nginx:latest  worker2  Running        Running 39 seconds ago

下面我们关掉worker2,我们可以看到，worker2上的容器被跑到了其他节点上。