Each worker is allocated *mapred.child.java.opts *memory, which in your case is 4000M. Check if your server doesn't have enough memory for 2 Mappers. Also the out of memory option is available in two forms. 1. Out of core graph 2. Out of core messages.
Currently you are setting only the out of core graph and not the out of core messages. Enable both of them. More information about options can be found here: http://giraph.apache.org/options.html set -D giraph.useOutOfCoreGraph=true -D giraph.useOutOfCoreMessages=true while passing options to GiraphRunner. Thanks, Ameya On Thu, Dec 5, 2013 at 12:39 PM, Sebastian Stipkovic < sebastian.stipko...@gmail.com> wrote: > Hello, > > I had setup giraph 1.1.0 with hadoop-0.20.203.0rc1 on a single > node cluster. It computes a tiny graph successful. But if the > input graph is huge (5 GB), I get an OutOfMemory(Garbage Collector) > exception, although I had turned on the out-of-memory-option. The job with > out-of-memory-option works only well with a tiny graph (0.9 GB). What is > Wrong? Does I have to do furthermore configurations? > > My Configurations are as follows: > > > namevalue *fs.s3n.impl*org.apache.hadoop.fs.s3native.NativeS3FileSystem > *mapred.task.cache.levels*2 *giraph.vertexOutputFormatClass* > org.apache.giraph.examples.MyShortestPaths$MyOutputFormat > *hadoop.tmp.dir*/app/hadoop/tmp *hadoop.native.lib*true *map.sort.class* > org.apache.hadoop.util.QuickSort > *dfs.namenode.decommission.nodes.per.interval*5 > *dfs.https.need.client.auth*false *ipc.client.idlethreshold*4000 > *dfs.datanode.data.dir.perm*755 *mapred.system.dir* > ${hadoop.tmp.dir}/mapred/system > *mapred.job.tracker.persist.jobstatus.hours*0 *dfs.datanode.address* > 0.0.0.0:50010 *dfs.namenode.logging.level*info > *dfs.block.access.token.enable*false *io.skip.checksum.errors*false > *fs.default.name > <http://fs.default.name>*hdfs://localhost:54310 > *mapred.cluster.reduce.memory.mb*-1 *mapred.child.tmp*./tmp > *fs.har.impl.disable.cache*true *dfs.safemode.threshold.pct*0.999f > *mapred.skip.reduce.max.skip.groups*0 *dfs.namenode.handler.count*10 > *dfs.blockreport.initialDelay*0 *mapred.heartbeats.in.second*100 > *mapred.tasktracker.dns.nameserver*default *io.sort.factor*10 > *mapred.task.timeout*600000 *giraph.maxWorkers*1 > *mapred.max.tracker.failures*4 *hadoop.rpc.socket.factory.class.default* > org.apache.hadoop.net.StandardSocketFactory > *mapred.job.tracker.jobhistory.lru.cache.size*5 *fs.hdfs.impl* > org.apache.hadoop.hdfs.DistributedFileSystem > *mapred.queue.default.acl-administer-jobs** > *dfs.block.access.key.update.interval*600 > *mapred.skip.map.auto.incr.proc.count*true > *mapreduce.job.complete.cancel.delegation.tokens*true > *io.mapfile.bloom.size*1048576 *mapreduce.reduce.shuffle.connect.timeout* > 180000 *dfs.safemode.extension*30000 > *mapred.jobtracker.blacklist.fault-timeout-window*180 > *tasktracker.http.threads*40 *mapred.job.shuffle.merge.percent*0.66 > *mapreduce.inputformat.class*org.apache.giraph.bsp.BspInputFormat > *fs.ftp.impl*org.apache.hadoop.fs.ftp.FTPFileSystem *user.name > <http://user.name>*hduser *mapred.output.compress*false > *io.bytes.per.checksum*512 *giraph.isStaticGraph*true > *mapred.healthChecker.script.timeout*600000 > *topology.node.switch.mapping.impl* > org.apache.hadoop.net.ScriptBasedMapping > *dfs.https.server.keystore.resource*ssl-server.xml > *mapred.reduce.slowstart.completed.maps*0.05 *mapred.reduce.max.attempts* > 4 *fs.ramfs.impl*org.apache.hadoop.fs.InMemoryFileSystem > *dfs.block.access.token.lifetime*600 *dfs.name.edits.dir*${dfs.name.dir} > *mapred.skip.map.max.skip.records*0 *mapred.cluster.map.memory.mb*-1 > *hadoop.security.group.mapping* > org.apache.hadoop.security.ShellBasedUnixGroupsMapping > *mapred.job.tracker.persist.jobstatus.dir*/jobtracker/jobsInfo > *mapred.jar* > hdfs://localhost:54310/app/hadoop/tmp/mapred/staging/hduser/.staging/job_201312051827_0001/job.jar > *dfs.block.size*67108864 *fs.s3.buffer.dir*${hadoop.tmp.dir}/s3 > *job.end.retry.attempts*0 *fs.file.impl* > org.apache.hadoop.fs.LocalFileSystem *mapred.local.dir.minspacestart*0 > *mapred.output.compression.type*RECORD *dfs.datanode.ipc.address* > 0.0.0.0:50020 *dfs.permissions*true *topology.script.number.args*100 > *io.mapfile.bloom.error.rate*0.005 *mapred.cluster.max.reduce.memory.mb* > -1 *mapred.max.tracker.blacklists*4 *mapred.task.profile.maps*0-2 > *dfs.datanode.https.address*0.0.0.0:50475 *mapred.userlog.retain.hours*24 > *dfs.secondary.http.address*0.0.0.0:50090 *dfs.replication.max*512 > *mapred.job.tracker.persist.jobstatus.active*false > *hadoop.security.authorization*false *local.cache.size*10737418240 > *dfs.namenode.delegation.token.renew-interval*86400000 > *mapred.min.split.size*0 *mapred.map.tasks*2 *mapred.child.java.opts* > -Xmx4000m *mapreduce.job.counters.limit*120 > *dfs.https.client.keystore.resource*ssl-client.xml *mapred.job.queue.name > <http://mapred.job.queue.name>*default *dfs.https.address*0.0.0.0:50470 > *mapred.job.tracker.retiredjobs.cache.size*1000 > *dfs.balance.bandwidthPerSec*1048576 *ipc.server.listen.queue.size*128 > *mapred.inmem.merge.threshold*1000 *job.end.retry.interval*30000 > *mapred.skip.attempts.to.start.skipping*2 *fs.checkpoint.dir* > ${hadoop.tmp.dir}/dfs/namesecondary *mapred.reduce.tasks*0 > *mapred.merge.recordsBeforeProgress*10000 *mapred.userlog.limit.kb*0 > *mapred.job.reduce.memory.mb*-1 *dfs.max.objects*0 > *webinterface.private.actions*false *io.sort.spill.percent*0.80 > *mapred.job.shuffle.input.buffer.percent*0.70 *mapred.job.name > <http://mapred.job.name>*Giraph: > org.apache.giraph.examples.MyShortestPaths *dfs.datanode.dns.nameserver* > default *mapred.map.tasks.speculative.execution*false > *hadoop.util.hash.type*murmur *dfs.blockreport.intervalMsec*3600000 > *mapred.map.max.attempts*0 *mapreduce.job.acl-view-job* > *dfs.client.block.write.retries*3 *mapred.job.tracker.handler.count*10 > *mapreduce.reduce.shuffle.read.timeout*180000 > *mapred.tasktracker.expiry.interval*600000 *dfs.https.enable*false > *mapred.jobtracker.maxtasks.per.job*-1 > *mapred.jobtracker.job.history.block.size*3145728 > *giraph.useOutOfCoreGiraph*true *keep.failed.task.files*false > *mapreduce.outputformat.class*org.apache.giraph.bsp.BspOutputFormat > *dfs.datanode.failed.volumes.tolerated*0 *ipc.client.tcpnodelay*false > *mapred.task.profile.reduces*0-2 *mapred.output.compression.codec* > org.apache.hadoop.io.compress.DefaultCodec *io.map.index.skip*0 > *mapred.working.dir*hdfs://localhost:54310/user/hduser > *ipc.server.tcpnodelay*false > *mapred.jobtracker.blacklist.fault-bucket-width*15 > *dfs.namenode.delegation.key.update-interval*86400000 > *mapred.used.genericoptionsparser*true *mapred.mapper.new-api*true > *mapred.job.map.memory.mb*-1 *giraph.vertex.input.dir* > hdfs://localhost:54310/user/hduser/output *dfs.default.chunk.view.size* > 32768 *hadoop.logfile.size*10000000 > *mapred.reduce.tasks.speculative.execution*true *mapreduce.job.dir* > hdfs://localhost:54310/app/hadoop/tmp/mapred/staging/hduser/.staging/job_201312051827_0001 > *mapreduce.tasktracker.outofband.heartbeat*false > *mapreduce.reduce.input.limit*-1 *dfs.datanode.du.reserved*0 > *hadoop.security.authentication*simple *fs.checkpoint.period*3600 > *dfs.web.ugi*webuser,webgroup *mapred.job.reuse.jvm.num.tasks*1 > *mapred.jobtracker.completeuserjobs.maximum*100 *dfs.df.interval*60000 > *dfs.data.dir*${hadoop.tmp.dir}/dfs/data > *mapred.task.tracker.task-controller* > org.apache.hadoop.mapred.DefaultTaskController *giraph.minWorkers*1 > *fs.s3.maxRetries*4 *dfs.datanode.dns.interface*default > *mapred.cluster.max.map.memory.mb*-1 *dfs.support.append*false > *mapreduce.job.acl-modify-job* > *dfs.permissions.supergroup*supergroup *mapred.local.dir* > ${hadoop.tmp.dir}/mapred/local *fs.hftp.impl* > org.apache.hadoop.hdfs.HftpFileSystem *fs.trash.interval*0 > *fs.s3.sleepTimeSeconds*10 *dfs.replication.min*1 > *mapred.submit.replication*10 *fs.har.impl* > org.apache.hadoop.fs.HarFileSystem *mapred.map.output.compression.codec* > org.apache.hadoop.io.compress.DefaultCodec > *mapred.tasktracker.dns.interface*default > *dfs.namenode.decommission.interval*30 *dfs.http.address*0.0.0.0:50070 > *dfs.heartbeat.interval*3 *mapred.job.tracker*localhost:54311 > *mapreduce.job.submithost*hduser *io.seqfile.sorter.recordlimit*1000000 > *giraph.vertexInputFormatClass* > org.apache.giraph.examples.MyShortestPaths$MyInputFormat *dfs.name.dir* > ${hadoop.tmp.dir}/dfs/name *mapred.line.input.format.linespermap*1 > *mapred.jobtracker.taskScheduler* > org.apache.hadoop.mapred.JobQueueTaskScheduler > *dfs.datanode.http.address*0.0.0.0:50075 *mapred.local.dir.minspacekill*0 > *dfs.replication.interval*3 *io.sort.record.percent*0.05 *fs.kfs.impl* > org.apache.hadoop.fs.kfs.KosmosFileSystem *mapred.temp.dir* > ${hadoop.tmp.dir}/mapred/temp *mapred.tasktracker.reduce.tasks.maximum*2 > *mapreduce.job.user.classpath.first*true *dfs.replication*1 > *fs.checkpoint.edits.dir*${fs.checkpoint.dir} *giraph.computationClass* > org.apache.giraph.examples.MyShortestPaths > *mapred.tasktracker.tasks.sleeptime-before-sigkill*5000 > *mapred.job.reduce.input.buffer.percent*0.0 > *mapred.tasktracker.indexcache.mb*10 > *mapreduce.job.split.metainfo.maxsize*10000000 *hadoop.logfile.count*10 > *mapred.skip.reduce.auto.incr.proc.count*true > *mapreduce.job.submithostaddress*127.0.1.1 > *io.seqfile.compress.blocksize*1000000 *fs.s3.block.size*67108864 > *mapred.tasktracker.taskmemorymanager.monitoring-interval*5000 > *giraph.minPercentResponded*100.0 *mapred.queue.default.state*RUNNING > *mapred.acls.enabled*false *mapreduce.jobtracker.staging.root.dir* > ${hadoop.tmp.dir}/mapred/staging *mapred.queue.names*default > *dfs.access.time.precision*3600000 *fs.hsftp.impl* > org.apache.hadoop.hdfs.HsftpFileSystem *mapred.task.tracker.http.address* > 0.0.0.0:50060 *mapred.reduce.parallel.copies*5 > *io.seqfile.lazydecompress*true *mapred.output.dir* > /user/hduser/output/shortestpaths *io.sort.mb*100 > *ipc.client.connection.maxidletime*10000 *mapred.compress.map.output* > false *hadoop.security.uid.cache.secs*14400 > *mapred.task.tracker.report.address*127.0.0.1:0 > *mapred.healthChecker.interval*60000 *ipc.client.kill.max*10 > *ipc.client.connect.max.retries*10 *ipc.ping.interval*300000 > *mapreduce.user.classpath.first*true *mapreduce.map.class* > org.apache.giraph.graph.GraphMapper *fs.s3.impl* > org.apache.hadoop.fs.s3.S3FileSystem *mapred.user.jobconf.limit*5242880 > *mapred.job.tracker.http.address*0.0.0.0:50030 *io.file.buffer.size*4096 > *mapred.jobtracker.restart.recover*false *io.serializations* > org.apache.hadoop.io.serializer.WritableSerialization > *dfs.datanode.handler.count*3 *mapred.reduce.copy.backoff*300 > *mapred.task.profile*false *dfs.replication.considerLoad*true > *jobclient.output.filter*FAILED > *dfs.namenode.delegation.token.max-lifetime*604800000 > *mapred.tasktracker.map.tasks.maximum*4 *io.compression.codecs* > org.apache.hadoop.io.compress.DefaultCodec,org.apache.hadoop.io.compress.GzipCodec,org.apache.hadoop.io.compress.BZip2Codec > *fs.checkpoint.size*67108864 > > Additionally, if I have more than one worker I get an Exception, too? Are > my configurations wrong? > > > best regards, > Sebastian >
