How to configure java memory heap space for hadoop mapreduce? -


i've tried run mapreduce job on 20 gb data, , got error on reduce shuffle phase. says because of memory heap space. then, i've read on many source, have decrease mapreduce.reduce.shuffle.input.buffer.percent property on mapred-site.xml default value 0,7. so, decrease 0,2.

i want ask, property take effect on time performance mapreduce job. so, how can configure make mapreduce job never error?

mapreduce.reduce.shuffle.input.buffer.percent 0.70 percentage of memory allocated maximum heap size storing map outputs during shuffle. looks that if decrease arbitrary value may degrade performance of shuffle phase. there have been reasoning , tests behind default value may check other related properties here http://hadoop.apache.org/docs/current/hadoop-mapreduce-client/hadoop-mapreduce-client-core/mapred-default.xml

what approx data output mappers ,if huge may want increase number of mappers likewise if number of reducers low heap space error happen during reduce phase.

you may want check job counters , increase number of mappers/reducers may try increasing mapper/reducer memory setting properties mapreduce.reduce.memory.mb , mapreduce.map.memory.mb


Comments

Popular posts from this blog

android - MPAndroidChart - How to add Annotations or images to the chart -

javascript - Add class to another page attribute using URL id - Jquery -

firefox - Where is 'webgl.osmesalib' parameter? -