通过RESTAPI实现对cdh集群某个进程进行单独的启停

通过cm api 单独启动和关闭某台主机的datanode进程。

测试环境

• cdh版本 Cloudera Express 5.11.0
• 需求：停止hadoop03主机上面的datanode进程

具体命令

1. 先获取集群各个主机所对应的hostId
   这个可以从数据库里面获取

SELECT name,HOST_IDENTIFIER FROM `scm`.`HOSTS`;

可以看到hadoop03上面对于的hostId为e2c11728-faf8-45ea-abe5-35932fbc2749。

2. 通过hostId 去获取datanode的roleName

curl -u admin:admin http://cm_ip:7180/api/v16/hosts/e2c11728-faf8-45ea-abe5-35932fbc2749

得到的结果

➜  ~ curl -u admin:admin http://cm_ip:7180/api/v16/hosts/e2c11728-faf8-45ea-abe5-35932fbc2749
{
  "hostId" : "e2c11728-faf8-45ea-abe5-35932fbc2749",
  "ipAddress" : "cm_ip",
  "hostname" : "hadoop03",
  "rackId" : "/OK3",
  "lastHeartbeat" : "2019-08-08T07:57:32.893Z",
  "healthSummary" : "GOOD",
  "healthChecks" : [ {
    "name" : "HOST_AGENT_LOG_DIRECTORY_FREE_SPACE",
    "summary" : "GOOD",
    "suppressed" : false
  }, , {
    "name" : "HOST_SCM_HEALTH",
    "summary" : "GOOD",
    "suppressed" : false
  } ],
  "roleRefs" : [ {
    "clusterName" : "cluster",
    "serviceName" : "hive",
    "roleName" : "hive-HIVEMETASTORE-f2e757abb270ac60de67ad4a42599e4b"
  }, {
    "clusterName" : "cluster",
    "serviceName" : "hdfs",
    "roleName" : "hdfs-NAMENODE-f2e757abb270ac60de67ad4a42599e4b"
  }....
 ],
  "maintenanceMode" : false,
  "maintenanceOwners" : [ ],
  "commissionState" : "COMMISSIONED",
  "numCores" : 24,
  "numPhysicalCores" : 12,
  "totalPhysMemBytes" : 135301316608,
  "entityStatus" : "GOOD_HEALTH",
  "clusterRef" : {
    "clusterName" : "cluster"
  }
}%

然后找到datanode的roleName为：
hdfs-DATANODE-f2e757abb270ac60de67ad4a42599e4b
3. 停止该进程

curl -u admin:admin -X POST -H "Content-Type: application/json" -d '{"items":["hdfs-DATANODE-f2e757abb270ac60de67ad4a42599e4b"]}' "http://cm_ip:7180/api/v16/clusters/cluster/services/hdfs/roleCommands/stop"

登陆cm页面发现该节点上datanode已经停止。