Initiating HDFS failover using the Cloudera Manager API

Steps to initiate failover of HDFS using the Cloudera Manager API.

The cluster must have HDFS High Availability enabled.

Determine the ID of the Active and Standby NameNode roles. Run the following command:

curl -X GET "http://myCluster-1:7180/api/v<API VERSION>/clusters/cluster_name/services/HDFS_service_name/roles?view=summary" -H "accept: application/json"

For example:

curl -X GET "http://myCluster-1:7180/api/v43/clusters/Cluster%201/services/HDFS-1/roles?view=summary" -H "accept: application/json"

A data structure is returned that contains the NameNode IDs. Look for output similar to the following:

      "name": "HDFS-1-NAMENODE-b520c25b659296aea5c8249c96a41b79",
      "type": "NAMENODE",
      "serviceRef": {
        "clusterName": "Cluster 1",
        "serviceName": "HDFS-1",
        "serviceDisplayName": "HDFS-1",
        "serviceType": "HDFS"
      },
      "hostRef": {
        "hostId": "dd503ff5-4c27-4261-b0f1-d9273ad6b510",
        "hostname": "nightly7x-cmha-4.nightly7x-cmha.root.hwx.site"
      },
      "roleUrl": "http://nightly7x-cmha-4.nightly7x-cmha.root.hwx.site:7180/cmf/roleRedirect/HDFS-1-NAMENODE-b520c25b659296aea5c8249c96a41b79",
      "roleState": "STARTED",
      "healthSummary": "GOOD",
      "configStalenessStatus": "FRESH",
      "haStatus": "ACTIVE",
      "maintenanceMode": false,
      "commissionState": "COMMISSIONED",
      "roleConfigGroupRef": {
        "roleConfigGroupName": "HDFS-1-NAMENODE-BASE"
      },

In the response above, the line "haStatus": "ACTIVE" indicates that this block defines the Active NameNode. The NameNode ID is the value of the "name" property, in this case the ID is HDFS-1-NAMENODE-b520c25b659296aea5c8249c96a41b79. You will use this value in the next step.

      "name": "HDFS-1-NAMENODE-f9c734734bb24fb4af93576ebeb13ad9",
      "type": "NAMENODE",
      "serviceRef": {
        "clusterName": "Cluster 1",
        "serviceName": "HDFS-1",
        "serviceDisplayName": "HDFS-1",
        "serviceType": "HDFS"
      },
      "hostRef": {
        "hostId": "5dc7ce94-9db0-4a50-bd82-f0bb64da624b",
        "hostname": "nightly7x-cmha-1.nightly7x-cmha.root.hwx.site"
      },
      "roleUrl": "http://nightly7x-cmha-4.nightly7x-cmha.root.hwx.site:7180/cmf/roleRedirect/HDFS-1-NAMENODE-f9c734734bb24fb4af93576ebeb13ad9",
      "roleState": "STARTED",
      "healthSummary": "GOOD",
      "configStalenessStatus": "FRESH",
      "haStatus": "STANDBY",
      "maintenanceMode": false,
      "commissionState": "COMMISSIONED",
      "roleConfigGroupRef": {
        "roleConfigGroupName": "HDFS-1-NAMENODE-BASE"
      },

In the response above, the line "haStatus": "STANDBY" indicates that this block defines the Standby NameNode. The NameNode ID is the value of the "name" property, in this case the ID is HDFS-1-NAMENODE-f9c734734bb24fb4af93576ebeb13ad9. You will use this value in the next step.

Run the following command to initiate the HDFS failover, substituting the NameNode IDs of the Active and Standby NameNodes :

 curl -X POST "Cloudera_Manager_hostname:Cloudera_Manager_port/api/v<API VERSION>/clusters/cluster_name/services/HDFS_service_name/commands/hdfsFailover?force=true|false" -H "accept: application/json" -H "Content-Type: application/json" -d "{ \"items\": [ \"Active_NameNode_ID\", \"Standby_NameNode_ID\" ]}"

For example:

curl -X POST “http://myCluster-1.com:7180/api/v43/clusters/Cluster%201/services/HDFS-1/commands/hdfsFailover?force=true” -H “accept: application/json” -H “Content-Type: application/json” -d “{ \“items\“: [ \“HDFS-1-NAMENODE-b520c25b659296aea5c8249c96a41b79\“, \“HDFS-1-NAMENODE-f9c734734bb24fb4af93576ebeb13ad9\” ]}”