https://community.cloudera.com/t5/Support-Questions/MapReduce-application-failed-with-OutOfMemoryError/m-p/56031#M36634 <P>Thank You&nbsp;Fawze&nbsp;and sorry for my delayed answer. I was sick.</P><P>Is in both, the map and the reduce phase.</P><P>The MR job takes aprox. 3TB from HBase and adds about 30GB of data per day.</P><P>Cluster architecture:</P><P>Number of NodeManagers = 37</P><P>Resource Managers = 2 with YARN in HA</P><P>5 nodes with 22 cores and 66GB of RAM</P><P>32 nodes with 30 cores and 120GB of RAM</P><P>Total vcores = 1070</P><P>Total memory = 4.07 TB</P><P>&nbsp;</P><P>I realized that decreasing the memory instead of increasing got better results.</P><P>Now with&nbsp;this params worked really well and could get all the cores.</P><P>&nbsp;</P><DIV><DIV><DIV><DIV>"mapreduce.map.memory.mb" 1536<BR />"mapreduce.reduce.memory.mb" 3072</DIV><DIV><DIV><DIV><DIV><DIV><DIV>"mapreduce.map.java.opts" -Xmx1024m<BR />"mapreduce.reduce.java.opts" -Xmx2560m</DIV></DIV></DIV></DIV></DIV></DIV><DIV>Thanks!</DIV><DIV>&nbsp;</DIV><DIV>Guido.</DIV></DIV></DIV></DIV> Fri, 16 Jun 2017 18:29:33 GMT gsalerno 2017-06-16T18:29:33Z