docker探索-swarm搭建docker集群(七)

前言

Swarm 在 Docker 1.12 版本之前属于一个独立的项目，在 Docker 1.12 版本发布之后，该项目合并到了 Docker 中，成为 Docker 的一个子命令，docker swarm是创建服务器集群的工具，只需要几条命令就可以创建一个服务器集群。它内置一些服务器集群需要的工具，比如说：服务查找，网络，负载均衡等等

一、环境

centos 7.3

Docker version 1.12.6

ip	角色
192.168.6.130	manager
192.168.6.131	worker
192.168.6.132	worker

二、集群

2.1、在192.168.6.130中，初始化集群

[root@jacky jacky]# docker swarm init --advertise-addr 192.168.6.130:

Swarm initialized: current node (4devzwb6rpimfpteqr17h2jx9) is now a manager.

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

    docker swarm join \

    --token SWMTKN--5r3ujri3th4038jp7q66zrfo56eqo0sabyage8ahc10121evog-ah9bptj9d7txdu6y91w7qxd81 \

    192.168.6.130:

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

[root@jacky jacky]#

说明：--advertise-addr 设置监听Ip和端口号

2.2、初始化完、查看集群上的节点列表

[root@jacky jacky]# docker node ls

ID                           HOSTNAME           STATUS  AVAILABILITY  MANAGER STATUS

4devzwb6rpimfpteqr17h2jx9 *  jacky.localdomain  Ready   Active        Leader

说明：目前只有一个节点，状态为Ready，是个Leader

2.3、192.168.6.131，加入集群

首先，切换到192.168.6.131，然后输入

[root@jacky jacky]#  docker swarm join \

> --token SWMTKN--5r3ujri3th4038jp7q66zrfo56eqo0sabyage8ahc10121evog-ah9bptj9d7txdu6y91w7qxd81 \

> 192.168.6.130:

This node joined a swarm as a worker.

[root@jacky jacky]#

2.4、192.168.6.132，加入集群

首先，切换到192.168.6.132，然后输入

[root@jacky jacky]# docker swarm join \

> --token SWMTKN--5r3ujri3th4038jp7q66zrfo56eqo0sabyage8ahc10121evog-ah9bptj9d7txdu6y91w7qxd81 \

> 192.168.6.130:

This node joined a swarm as a worker.

[root@jacky jacky]#

2.5、接着回到192.168.6.130中，查看集群中节点信息

[root@jacky jacky]# docker node ls

ID                           HOSTNAME           STATUS  AVAILABILITY  MANAGER STATUS

4devzwb6rpimfpteqr17h2jx9 *  jacky.localdomain  Ready   Active        Leader

5mjtda2uzzu43v2xuxdco5ogr    jacky.localdomain  Ready   Active

a8wdtux82dolsbgmv6ff0uu94    jacky.localdomain  Ready   Active

说明：
AVAILABILITY列：
显示调度程序是否可以将任务分配给节点：

Active 意味着调度程序可以将任务分配给节点。
Pause 意味着调度程序不会将新任务分配给节点，但现有任务仍在运行。
Drain 意味着调度程序不会向节点分配新任务。调度程序关闭所有现有任务并在可用节点上调度它们。

MANAGER STATUS列
显示节点是属于manager或者worker

没有值表示不参与群管理的工作节点。
Leader 意味着该节点是使得群的所有群管理和编排决策的主要管理器节点。
Reachable 意味着节点是管理者节点正在参与Raft共识。如果领导节点不可用，则该节点有资格被选为新领导者。
Unavailable 意味着节点是不能与其他管理器通信的管理器。如果管理器节点不可用，您应该将新的管理器节点加入群集，或者将工作器节点升级为管理器。

2.6、创建集群网络overlay

[root@jacky jacky]# docker network create --driver overlay skynet

843z9qb9c6douf7ir7l3iimqn

[root@jacky jacky]# docker network ls

NETWORK ID          NAME                DRIVER              SCOPE

ad5125729239        bridge              bridge              local

5a15f008fb38        host                host                local

6echvokyh2m3        ingress             overlay             swarm

28068704e605        none                null                local

843z9qb9c6do        skynet              overlay             swarm

可以看到新创建的网络名称为skynet，类型为overlay

三、部署测试

3.1、在管理节点中执行

docker service create -p : --name webserver --replicas  httpd

3.2、查看集群中的服务

[root@node1 jacky]# docker service ls

ID            NAME       REPLICAS  IMAGE                 COMMAND

0blhke4vywh8  viz        /       manomarks/visualizer

7batkp4zv9f3  portainer  /       portainer/portainer   -H unix:///var/run/docker.sock

7kw3ovihgqgb  webserver  /       httpd

3.3、查看集群中的webserver服务

[root@node1 jacky]# docker service ps webserver

ID NAME IMAGE NODE DESIRED STATE CURRENT STATE ERROR

e0jqrg479muha7ow8bf34rv39 webserver. httpd node1.jacky.com Running Running hours ago

23n9df2vww079h5rgkxlri4uy \_ webserver. httpd node1.jacky.com Shutdown Complete hours ago

8b8cs13u9yjsoru3ybyzzv9e6 \_ webserver. httpd node2.jacky.com Shutdown Rejected hours ago "No such image: httpd:latest"

8lvx0ynaohlcfyp11jgji4m3q webserver. httpd node1.jacky.com Running Running hours ago

0q8lrzlybo1exl3bngwwfy386 \_ webserver. httpd node1.jacky.com Shutdown Complete hours ago

eoq4a2sqx80a0hly6tizt5ucf \_ webserver. httpd node3.jacky.com Shutdown Shutdown hours ago

10juv2jp1ay9rjbu5hgw6yhs3 \_ webserver. httpd node3.jacky.com Shutdown Failed hours ago "starting container failed: er…"

7xa8uoa8775i5nl0zzi373xbt \_ webserver. httpd node3.jacky.com Shutdown Failed hours ago "starting container failed: er…"

6puw8t22w0exgiwqzt5vi8fc1 webserver. httpd node1.jacky.com Running Running hours ago

74osfdgl5ovp1c3e5s012b3f6 \_ webserver. httpd node1.jacky.com Shutdown Complete hours ago

cwzuewjsolewap28ctvy3jxaa \_ webserver. httpd node3.jacky.com Shutdown Shutdown hours ago

9bb5q38zqk153uqdex1yfcu4e \_ webserver. httpd node3.jacky.com Shutdown Failed hours ago "starting container failed: er…"

1uvoczhsfz5ncp1emdueljwqa \_ webserver. httpd node3.jacky.com Shutdown Failed hours ago "starting container failed: er…"

0j9hsq18v3pzoecmrjg0qtynh webserver. httpd node1.jacky.com Running Running hours ago

dyhyy55xlkm3abw2cqm0k8y6h \_ webserver. httpd node1.jacky.com Shutdown Complete hours ago

8dymtjskyxhw5zj0ombpv0pm1 \_ webserver. httpd node2.jacky.com Shutdown Shutdown hours ago

1b6u7rtknpgmwyfn3j3p94wm6 \_ webserver. httpd node2.jacky.com Shutdown Rejected hours ago "No such image: httpd:latest"

1af72d5vpu1xg3u0qypnvlier \_ webserver. httpd node2.jacky.com Shutdown Rejected hours ago "No such image: httpd:latest"

897au9dm88i94l0scg69slu6z webserver. httpd node1.jacky.com Running Running hours ago

eqt7g4bk6e2kqy6qbnr13gfeh \_ webserver. httpd node1.jacky.com Shutdown Complete hours ago

7vq8u2eycraafwzlmcgg1e80d \_ webserver. httpd node3.jacky.com Shutdown Shutdown hours ago

ehu6f1xun8ha6lw7cyex7jjrw \_ webserver. httpd node2.jacky.com Shutdown Shutdown hours ago

3.4、访问http://192.168.6.130、http://192.168.6.131或http://192.168.6.132

说明：192.168.6.131和192.168.6.132没有部署httpd也能狗访问，swarm搭建docker集群测试成功

四、安装Portainer，图形化管理docker

4.1、关闭selinux

setenforce

4.2、在manager节点中执行，如下：

docker service create \

--name portainer \

--publish : \

--constraint 'node.role == manager' \

--mount type=bind,src=/var/run/docker.sock,dst=/var/run/docker.sock \

portainer/portainer \

-H unix:///var/run/docker.sock

4.3、访问192.168.6.130.9000

五、swarm集群其他操作

5.1、更新节点的可见性状态

[root@node1 jacky]# docker node ls

ID                           HOSTNAME         STATUS  AVAILABILITY  MANAGER STATUS

4devzwb6rpimfpteqr17h2jx9 *  node1.jacky.com  Ready   Active        Leader

5mjtda2uzzu43v2xuxdco5ogr    node3.jacky.com  Ready   Active

a8wdtux82dolsbgmv6ff0uu94    node2.jacky.com  Ready   Active

[root@node1 jacky]# docker node update --availability Drain node2.jacky.com

node2.jacky.com

[root@node1 jacky]# docker node ls

ID                           HOSTNAME         STATUS  AVAILABILITY  MANAGER STATUS

4devzwb6rpimfpteqr17h2jx9 *  node1.jacky.com  Ready   Active        Leader

5mjtda2uzzu43v2xuxdco5ogr    node3.jacky.com  Ready   Active

a8wdtux82dolsbgmv6ff0uu94    node2.jacky.com  Ready   Drain

[root@node1 jacky]# docker node update --availability Active node2.jacky.com

node2.jacky.com

[root@node1 jacky]# docker node ls

ID                           HOSTNAME         STATUS  AVAILABILITY  MANAGER STATUS

4devzwb6rpimfpteqr17h2jx9 *  node1.jacky.com  Ready   Active        Leader

5mjtda2uzzu43v2xuxdco5ogr    node3.jacky.com  Ready   Active

a8wdtux82dolsbgmv6ff0uu94    node2.jacky.com  Ready   Active

[root@node1 jacky]#

5.2、升级或降级节点

您可以将工作程序节点提升为manager角色。这在管理器节点不可用或者您希望使管理器脱机以进行维护时很有用。类似地，您可以将管理器节点降级为worker角色。
无论您升级或降级节点，您应该始终在群中维护奇数个管理器节点。

升级节点

[root@node1 jacky]# docker node promote node3.jacky.com node2.jacky.com

[root@node1 jacky]# docker node ls

ID                           HOSTNAME         STATUS  AVAILABILITY  MANAGER STATUS

4devzwb6rpimfpteqr17h2jx9 *  node1.jacky.com  Ready   Active        Leader

5mjtda2uzzu43v2xuxdco5ogr    node3.jacky.com  Ready   Active        Unreachable

a8wdtux82dolsbgmv6ff0uu94    node2.jacky.com  Ready   Active        Reachable

降级节点

[root@node1 jacky]# docker node demote node3.jacky.com node2.jacky.com

Manager node3.jacky.com demoted in the swarm.

Manager node2.jacky.com demoted in the swarm.

[root@node1 jacky]# docker node ls

ID                           HOSTNAME         STATUS  AVAILABILITY  MANAGER STATUS

4devzwb6rpimfpteqr17h2jx9 *  node1.jacky.com  Ready   Active        Leader

5mjtda2uzzu43v2xuxdco5ogr    node3.jacky.com  Ready   Active

a8wdtux82dolsbgmv6ff0uu94    node2.jacky.com  Ready   Active

5.3、退出docker swarm集群

在对应的worker节点执行如下命令

[root@node1 jacky]# docker swarm leave

如果想把manager节点上已经退出集群的节点信息删除，如下：

[root@node1 jacky]# docker node rm node2.jacky.com