Cloudbreak: How to enable High Availability for NameNode in Blueprint

[#1]

I want to enable NameNode high availability in my cluster but can't seem to find the right way to enable this from the blueprint I'm supplying to Cloudbreak.

[#2 · Accepted solution]

Hi,

Here's and example blueprint, below. In the cluster creation wizard you must disable blueprint validation (it's under the advanced options)(don't worry about the localhost references in the configuration, it will be resolved by Ambari)

{
  "configurations": [
    {
      "hdfs-site": {
        "properties": {
          "dfs.namenode.http-address": "localhost:50070",
          "dfs.namenode.https-address": "localhost:50470",
          "dfs.namenode.rpc-address": "localhost:8020",
          "dfs.namenode.secondary.http-address": "localhost:50090",
          "dfs.nameservices": "mycluster",
          "dfs.internal.nameservices": "mycluster",
          "dfs.ha.namenodes.mycluster": "nn1,nn2",
          "dfs.client.failover.proxy.provider.mycluster": "org.apache.hadoop.hdfs.server.namenode.ha.ConfiguredFailoverProxyProvider",
          "dfs.ha.fencing.methods": "shell(/bin/true)",
          "dfs.ha.automatic-failover.enabled": "true",
          "dfs.namenode.shared.edits.dir": "qjournal://localhost:8485/mycluster",
          "dfs.namenode.https-address.mycluster.nn1": "%HOSTGROUP::master1%:50470",
          "dfs.namenode.http-address.mycluster.nn1": "%HOSTGROUP::master1%:50070",
          "dfs.namenode.rpc-address.mycluster.nn1": "%HOSTGROUP::master1%:8020",
          "dfs.namenode.https-address.mycluster.nn2": "%HOSTGROUP::master2%:50470",
          "dfs.namenode.http-address.mycluster.nn2": "%HOSTGROUP::master2%:50070",
          "dfs.namenode.rpc-address.mycluster.nn2": "%HOSTGROUP::master2%:8020"
        }
      }
    },
    {
      "core-site": {
        "properties": {
          "fs.defaultFS": "hdfs://mycluster",
          "ha.zookeeper.quorum": "localhost:2181"
        }
      }
    }
  ],
  "host_groups": [
    {
      "name": "master1",
      "configurations": [],
      "components": [
        {
          "name": "ZOOKEEPER_SERVER"
        },
        {
          "name": "ZOOKEEPER_CLIENT"
        },
        {
          "name": "NAMENODE"
        },
        {
          "name": "DATANODE"
        },
        {
          "name": "JOURNALNODE"
        },
        {
          "name": "ZKFC"
        },
        {
          "name": "HDFS_CLIENT"
        },
        {
          "name": "YARN_CLIENT"
        },
        {
          "name": "MAPREDUCE2_CLIENT"
        }
      ],
      "cardinality": "1"
    },
    {
      "name": "master2",
      "configurations": [],
      "components": [
        {
          "name": "ZOOKEEPER_SERVER"
        },
        {
          "name": "ZOOKEEPER_CLIENT"
        },
        {
          "name": "NAMENODE"
        },
        {
          "name": "DATANODE"
        },
        {
          "name": "ZKFC"
        },
        {
          "name": "JOURNALNODE"
        },
        {
          "name": "RESOURCEMANAGER"
        },
        {
          "name": "HISTORYSERVER"
        },
        {
          "name": "APP_TIMELINE_SERVER"
        },
        {
          "name": "HDFS_CLIENT"
        },
        {
          "name": "YARN_CLIENT"
        }
      ],
      "cardinality": "1"
    },
    {
      "name": "master3",
      "configurations": [],
      "components": [
        {
          "name": "ZOOKEEPER_SERVER"
        },
        {
          "name": "ZOOKEEPER_CLIENT"
        },
        {
          "name": "DATANODE"
        },
        {
          "name": "JOURNALNODE"
        },
        {
          "name": "HDFS_CLIENT"
        },
        {
          "name": "NODEMANAGER"
        },
        {
          "name": "YARN_CLIENT"
        }
      ],
      "cardinality": "1"
    },
    {
      "name": "worker",
      "configurations": [],
      "components": [
        {
          "name": "DATANODE"
        },
        {
          "name": "HDFS_CLIENT"
        },
        {
          "name": "NODEMANAGER"
        },
        {
          "name": "YARN_CLIENT"
        },
        {
          "name": "MAPREDUCE2_CLIENT"
        }
      ],
      "cardinality": "1"
    }
  ],
  "Blueprints": {
    "blueprint_name": "nn-ha",
    "stack_name": "HDP",
    "stack_version": "2.5"
  } 

}

[#3]

Thanks for the fast answer. If I used cloudbreakshell, is there a way to tell it to ignore the validation?

[#4]

In shell it is disabled by default: https://github.com/hortonworks/cloudbreak/blob/master/shell/src/main/java/com/sequenceiq/cloudbreak/...