Docker Swarm节点的清空（Drain）和在启用

所有的active节点，都可以接收task。有时候会遇到，一个节点需要维护暂时不能使用，这意味着此节点将不能再接受新的task，且原有的task也要迁走，在其他可用的节点上运行

【注】将节点设置为DRAIN并不会从该节点删除独立容器（docker run/ docker up/ docker Engine API创建的容器）。节点的状态只影响service

1. 查看redis容器分布的节点

[root@node1 ~]# docker service ps redis
ID                  NAME                IMAGE               NODE                DESIRED STATE       CURRENT STATE           ERROR               PORTS
2au9etgoryzy        redis.1             redis:3.0.7         node1               Running             Running 2 hours ago                         
daxio9w01x6f         \_ redis.1         redis:3.0.6         node2               Shutdown            Shutdown 2 hours ago                        
t5na1flecwgb        redis.2             redis:3.0.7         node1               Running             Running 2 hours ago                         
ny2ia69mibek         \_ redis.2         redis:3.0.6         node2               Shutdown            Shutdown 2 hours ago                        
355xlpbugpt2        redis.3             redis:3.0.7         node1               Running             Running 2 hours ago                         
305huk42qp07         \_ redis.3         redis:3.0.6         node1               Shutdown            Shutdown 10 hours ago                       
[root@node1 ~]# 

2. 清空node1节点的任务

[root@node1 ~]# docker node update --availability drain node1 
node1

3. 查看此时任务分配的节点

[root@node1 ~]# docker service ps redis
ID                  NAME                IMAGE               NODE                DESIRED STATE       CURRENT STATE            ERROR               PORTS
3r2nu9exguqf        redis.1             redis:3.0.7         node2               Running             Starting 2 seconds ago                       
2au9etgoryzy         \_ redis.1         redis:3.0.7         node1               Shutdown            Shutdown 2 seconds ago                       
daxio9w01x6f         \_ redis.1         redis:3.0.6         node2               Shutdown            Shutdown 2 hours ago                         
3ie4hycd9m55        redis.2             redis:3.0.7         node2               Running             Starting 1 second ago                        
t5na1flecwgb         \_ redis.2         redis:3.0.7         node1               Shutdown            Shutdown 2 seconds ago                       
ny2ia69mibek         \_ redis.2         redis:3.0.6         node2               Shutdown            Shutdown 2 hours ago                         
ihnnj7k74aud        redis.3             redis:3.0.7         node2               Running             Starting 1 second ago                        
355xlpbugpt2         \_ redis.3         redis:3.0.7         node1               Shutdown            Shutdown 2 seconds ago                       
305huk42qp07         \_ redis.3         redis:3.0.6         node1               Shutdown            Shutdown 10 hours ago                        

可以看到此时任务都分配到了node2节点

4. 检查节点的可用性

[root@node1 ~]# docker node inspect --pretty node1
ID:			28t3r12oxo2k9derz8i7yxw7p
Hostname:              	node1
Joined at:             	2019-06-08 12:20:34.266379324 +0000 utc
Status:
 State:			Ready
 Availability:         	Drain
 Address:		192.168.229.128
...

5. 启用被清空的节点

[root@node1 ~]# docker node update --availability active node1
node1

[root@node1 ~]# docker node inspect --pretty node1
ID:			28t3r12oxo2k9derz8i7yxw7p
Hostname:              	node1
Joined at:             	2019-06-08 12:20:34.266379324 +0000 utc
Status:
 State:			Ready
 Availability:         	Active
 Address:		192.168.229.128

当节点的状态变为Active时，可以接受新的任务

• 在服务更新期间进行扩展
• 在滚动更新期间
• 当您设置另一个节点以耗尽可用性时
• 当另一个活动节点上的任务失败时

参考
https://my.oschina.net/u/915811/blog/1593856
https://docs.docker.com/engine/swarm/swarm-tutorial/drain-node/