Details
-
Bug
-
Status: Open
-
Major
-
Resolution: Unresolved
-
None
-
None
-
None
-
None
Description
I am trying to read 1.1 gb file which has data and i am reading this file and based on the data doing a put to Hbase table , mapper is working fine but reducer is doing till 85% and fails with following error
java.lang.OutOfMemoryError: Java heap space
at java.io.DataOutputStream.<init>(DataOutputStream.java:204)
at org.apache.jute.BinaryOutputArchive.getArchive(BinaryOutputArchive.java:38)
at org.apache.zookeeper.ClientCnxn$Packet.createBB(ClientCnxn.java:282)
at org.apache.zookeeper.ClientCnxnSocketNIO.doIO(ClientCnxnSocketNIO.java:115)
at org.apache.zookeeper.ClientCnxnSocketNIO.doTransport(ClientCnxnSocketNIO.java:355)
at org.apache.zookeeper.ClientCnxn$SendThread.run(ClientCnxn.java:1081)
16/05/31 03:37:50 WARN mapred.LocalJobRunner: job_local1952932043_0001
java.lang.OutOfMemoryError: Java heap space
following are the configuration i have given for the running this mapreduce job
configuration.set("mapreduce.child.java.opts", "-Xmx16g -XX:+UseConcMarkSweepGC -XX:-UseGCOverheadLimit");
configuration.set("mapreduce.reduce.memory.mb", "8192m");
configuration.set("mapred.reduce.child.java.opts", "-Xmx10g");
I try reducing the file size by using some split logic. what i have notice if the file size is 350 Mb Mapreduce work file without issue, but if it is more than 350 Mb we will be getting OOM exception.
can some one please let us know what configuration is am missing or changes is required to run Mapreduce Job