Swarm 集群指定service容器运行节点-实例

Test

---

问题描述

我现在集群里有3太主机分别为163，192，193，现在假设193存在问题service的容器将不在193节点上运行。

[root@sit-cdpapp-163l ~]# docker node ls  
ID                            HOSTNAME            STATUS              AVAILABILITY        MANAGER STATUS      ENGINE VERSION  
ohjdhdzdiebk6ibjoriygfszu     metabase193         Ready               Active              Reachable           19.03.5  
tkoaoslrj2y526qugt3kwzfjo     postgresql192       Ready               Active              Reachable           19.03.5  
t8ixy0wk2mf1owvcsgu6bvt6b *   sit-cdpapp-163l     Ready               Active              Leader              19.03.5

解决方案

将3台node都打上标签来区分--163，192使用标签相同

[root@sit-cdpapp-163l ~]# docker node update --label-add tag=163 sit-cdpapp-163l  
sit-cdpapp-163l  
 

[root@postgresql192 certs.d]# docker node update --label-add tag=163 postgresql192  
postgresql192  
 

[root@metabase193 ~]# docker node update --label-add tag=193 metabase193  
metabase193

• 注意“tag=163”，tag可以为任意字段

[root@sit-cdpapp-163l ~]# docker service create --name worker --hostname worker --constraint 'node.labels.tag == 163' --network sk-net registry.xxxx.com/sk_pbms_worker_dev:alpha3.4.X  
85mgnxhx5m3nwod5any3za1dj  
overall progress: 1 out of 1 tasks   
1/1: running   [==================================================>]   
verify: Service converged 

[root@postgresql192 certs.d]# docker service scale worker=3

[root@sit-cdpapp-163l ~]# docker service ps worker  
ID                  NAME                IMAGE                                              NODE                DESIRED STATE       CURRENT STATE            ERROR                              PORTS  
8xxlea1infyw        worker.1            registry.xxxx.com/sk_pbms_worker_dev:alpha3.4.X   postgresql192       Running             Running 5 minutes ago                                         
92upe13xos13         \_ worker.1        registry.xxxx.com/sk_pbms_worker_dev:alpha3.4.X   postgresql192       Shutdown            Failed 5 minutes ago     "No such container: worker.1.9…"     
ue5vpn47vl9x         \_ worker.1        registry.xxxx.com/sk_pbms_worker_dev:alpha3.4.X   sit-cdpapp-163l     Shutdown            Shutdown 3 minutes ago                                        
nu2w0uzulac3        worker.2            registry.xxxx.com/sk_pbms_worker_dev:alpha3.4.X   postgresql192       Running             Running 4 minutes ago                                         
hx5fjj9nilbb        worker.3            registry.xxxx.com/sk_pbms_worker_dev:alpha3.4.X   sit-cdpapp-163l     Running             Running 21 seconds ago

Test2

---

测试update 替换指定service容器运行节点

[root@sit-cdpapp-163l ~]# docker service update --constraint-add 'node.labels.tag == 193'  worker  
worker  
overall progress: 0 out of 3 tasks   
1/3: no suitable node (scheduling constraints not satisfied on 3 nodes)   
2/3:     
3/3:     
^COperation continuing in background.  
Use `docker service ps worker` to check progress.

• 如果你在创建时候指定过服务约束，现在想换地方，更新的时候 constraint-rm 和 constraint-add 要同时执行，否则就崩了，还好是后台，也有先启再停的策略。

[root@sit-cdpapp-163l ~]# docker service update --constraint-rm 'node.labels.tag == 163'  worker  
worker  
overall progress: 3 out of 3 tasks   
1/3: running   [==================================================>]   
2/3: running   [==================================================>]   
3/3: running   [==================================================>]   
verify: Service converged   
[root@sit-cdpapp-163l ~]# docker service update --constraint-add 'node.labels.tag == 193'  worker  
worker  
overall progress: 3 out of 3 tasks   
1/3: running   [==================================================>]   
2/3: running   [==================================================>]   
3/3: running   [==================================================>]   
verify: Service converged

• 运行结果如下

[root@postgresql192 certs.d]# docker service ps worker  
ID                  NAME                IMAGE                                              NODE                DESIRED STATE       CURRENT STATE             ERROR                              PORTS  
pqe7ffne9sx1        worker.1            registry.xxxx.com/sk_pbms_worker_dev:alpha3.4.X   metabase193         Running             Running 28 minutes ago                                         
j5zoqbqi837s         \_ worker.1        registry.xxxx.com/sk_pbms_worker_dev:alpha3.4.X                       Shutdown            Pending 34 minutes ago    "no suitable node (scheduling …"     
zmleg7w6wiqs         \_ worker.1        registry.xxxx.com/sk_pbms_worker_dev:alpha3.4.X   sit-cdpapp-163l     Shutdown            Shutdown 34 minutes ago                                        
qxga9yumzlte        worker.2            registry.xxxx.com/sk_pbms_worker_dev:alpha3.4.X   metabase193         Running             Running 27 minutes ago                                         
vevxy0pahkpp         \_ worker.2        registry.xxxx.com/sk_pbms_worker_dev:alpha3.4.X   postgresql192       Shutdown            Shutdown 27 minutes ago                                        
nty6x648eixr        worker.3            registry.xxxx.com/sk_pbms_worker_dev:alpha3.4.X   metabase193         Running             Running 27 minutes ago                                         
z18epvm2o2wf         \_ worker.3        registry.xxxx.com/sk_pbms_worker_dev:alpha3.4.X   postgresql192       Shutdown            Shutdown 27 minutes ago

补充注意

• 不排除有人创建时 --constraint 使用了未标记的标签，如果这样的话service会显示running，current state显示pending不分配节点，这时如果任意一个节点标签设置了create时未标记的标签，容器将瞬间部署到其上。