List the namenode and datanodes of a cluster from any node?

Question

From any node in a Hadoop cluster, what is the command to

• identify the running namenode?
• identify all running datanodes?

I have looked through the commands manual and have not found this.

Alston

List of datanodes is available via namenode's JMX interface -- https://mcmap.net/q/409002/-rest-api-for-getting-dfs-used-for-individual-nodes-in-hadoop/1097572 – Wingspan

Answer 1

Use the dfsadmin command:

bin/hadoop dfsadmin -report

Update (2015):

bin/hdfs dfsadmin -report

Foeman

OK what I don't get is why my namenode only reports itself as a datanode, and none of the datanodes I listed in slaves file show up in dfsadmin -report – Alston

On the namenode url (http://<IP of namenode>:50070/dfshealth.jsp), see if the other nodes are dead / decomissioned. Also, watch the datanode logs to identify the problem – Foeman

This command is now deprecated: DEPRECATED: Use of this script to execute hdfs command is deprecated. – Coppinger

Answer 2

Try this:

hdfs getconf -namenodes
hdfs getconf -secondaryNamenodes

Motionless

Answer 3

Try This. For dfsadmin not every user has permission.

hdfs getconf -confKey fs.defaultFS
hdfs://XYZ

Pneumatophore

Answer 4

2016 update, following Tejas answer:

hadoopuser@gsamaras:~/hadoop/bin$ ./hdfs dfsadmin -report

Ox

Answer 5

hdfs getconf -namenodes
hdfs getconf -secondaryNamenodes

Work only non-HA mode. The simplest way for HA-mode is to use:

hdfs haadmin -getAllServiceState 

haadmin work for HA mode.

Output:

hadoop-adh-1:8020                standby
hadoop-adh-2:8020                active

Enzymology