hadoop - Increasing Java heap size for a class inside a jar -
i need execute java programs contained in jar file error requires increasing java heap size. need increase heap size using command line. tried java -xmx6144m -d64 did not work. on other hand, java -xmx6144m classname requires classname. how can specify class name inside jar? there command increase java heap classes?
here error:
ava.lang.outofmemoryerror: gc overhead limit exceeded @ java.util.regex.pattern.compile(pattern.java:1047) @ java.lang.string.replace(string.java:2180) @ org.apache.hadoop.fs.path.normalizepath(path.java:146) @ org.apache.hadoop.fs.path.initialize(path.java:137) @ org.apache.hadoop.fs.path.<init>(path.java:126) @ org.apache.hadoop.fs.path.makequalified(path.java:296) @ org.apache.hadoop.fs.rawlocalfilesystem$rawlocalfilestatus.<init>(rawlocalfilesystem.java:375) @ org.apache.hadoop.fs.rawlocalfilesystem.getfilestatus(rawlocalfilesystem.java:359) @ org.apache.hadoop.fs.rawlocalfilesystem.liststatus(rawlocalfilesystem.java:290) @ org.apache.hadoop.fs.filesystem.liststatus(filesystem.java:721) @ org.apache.hadoop.fs.filesystem.liststatus(filesystem.java:746) @ org.apache.hadoop.fs.checksumfilesystem.liststatus(checksumfilesystem.java:465) @ org.apache.hadoop.fs.filesystem.liststatus(filesystem.java:721) @ org.apache.hadoop.fs.filesystem.liststatus(filesystem.java:746) @ org.apache.hadoop.mapreduce.lib.input.fileinputformat.liststatus(fileinputformat.java:212) @ org.apache.hadoop.mapreduce.lib.input.fileinputformat.getsplits(fileinputformat.java:241) @ org.apache.hadoop.mapred.localjobrunner$job.run(localjobrunner.java:120)
i believe need increase memory allocation launched map tasks. default 200 mb seems insufficient in case. if driver class subclass of tool, can pass following arguments -dmapred.child.java.opts=-xmx500m . here , bumping memory 200 500.
you can profiling of map reduce job see objects taking memory.
Comments
Post a Comment