Hunk search problem: Cannot initialize Cluster.

[Ledion_Bitincka]

I've configured Hunk to run searches against my cluster however I keep running into this issue when I try to execute a reporting search (stats, top, chart etc) - streaming searches seem to be working just fine

   INFO  ERP.hadoop -  Cluster - Failed to use org.apache.hadoop.mapred.LocalClientProtocolProvider due to error: Invalid "mapreduce.jobtracker.address" configuration value for LocalJobRunner : "xxxx.splunk.com:8021"
    ERROR ERP.hadoop -  UserGroupInformation - PriviledgedActionException as:lbitincka (auth:SIMPLE) cause:java.io.IOException: Cannot initialize Cluster. Please check your configuration for mapreduce.framework.name and the correspond server addresses.
    ERROR ERP.hadoop -  AsyncMRJob - Cannot initialize Cluster. Please check your configuration for mapreduce.framework.name and the correspond server addresses.
    ERROR ERP.hadoop -  java.io.IOException: Cannot initialize Cluster. Please check your configuration for mapreduce.framework.name and the correspond server addresses.
    ERROR ERP.hadoop -      at org.apache.hadoop.mapreduce.Cluster.initialize(Cluster.java:121)
    ERROR ERP.hadoop -      at org.apache.hadoop.mapreduce.Cluster.<init>(Cluster.java:83)
    ERROR ERP.hadoop -      at org.apache.hadoop.mapreduce.Cluster.<init>(Cluster.java:76)
    ERROR ERP.hadoop -      at org.apache.hadoop.mapreduce.Job$10.run(Job.java:1239)
    ERROR ERP.hadoop -      at org.apache.hadoop.mapreduce.Job$10.run(Job.java:1235)
    ERROR ERP.hadoop -      at java.security.AccessController.doPrivileged(Native Method)
    ERROR ERP.hadoop -      at javax.security.auth.Subject.doAs(Subject.java:415)
    ERROR ERP.hadoop -      at org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1408)
    ERROR ERP.hadoop -      at org.apache.hadoop.mapreduce.Job.connect(Job.java:1234)
    ERROR ERP.hadoop -      at org.apache.hadoop.mapreduce.Job.submit(Job.java:1263)
    ERROR ERP.hadoop -      at com.splunk.mr.AsyncMRJob.run(AsyncMRJob.java:131)
    ERROR ERP.hadoop -      at java.lang.Thread.run(Thread.java:724)

[Ledion_Bitincka]

This issue is caused by the Hadoop client libraries being incompatible with the Hadoop version that the cluster is running. In my case above I was using CDH4.3.1 libraries to talk to CDH4.3.0. The problem went away when using the same client library version as the cluster. To check for compatibility I would recommend running the following commands:

$HADOOP_HOME/bin/hadoop fs -ls hdfs://namenode:ipc_port/

$HADOOP_HOME/bin/hadoop jobs -jt jobtracker:ipc_port -list all