Hi Francisco, By any chance, is your aggregation runner getting messages from all vertices, with each message being the vector from each vertex? That is 35000*65537*4(assuming integer) = roughly 9GB. Apologies if you are not, because assuming you have 15 tasks for the purpose, the memory usage for vertex is coming to be around 600 MB. I am looking into limiting the memory used during Vertex#compute. Apologies for the delay.
On Wed, Jan 30, 2013 at 2:01 AM, Edward J. Yoon <[email protected]>wrote: > Have you tried to increase the number of tasks? > > On Sat, Jan 26, 2013 at 2:34 AM, Francisco Sanches > <[email protected]> wrote: > > Hi > > > > With graphs with up to 15,000 vertices can run. I believe it is a > > quick look at memory saw this being used enough memory swap. > > > > 2013/1/24 Edward J. Yoon <[email protected]>: > >> Have you checked your algorithm with small input? I'm not sure that is > >> a memory issue. > >> > >> On Thu, Jan 24, 2013 at 7:53 PM, Francisco Sanches > >> <[email protected]> wrote: > >>> Hi thanks for the reply, > >>> > >>> My input is a graph of 35 000 vertices, but my application uses a > >>> vector of 65537 to store the distance. Each node has a vector with the > >>> number of positions (All-pair). > >>> > >>> 2013/1/24 Edward J. Yoon <[email protected]>: > >>>> Hi, > >>>> > >>>> How large your input data is? > >>>> > >>>> On Tue, Jan 22, 2013 at 11:21 PM, Francisco Sanches > >>>> <[email protected]> wrote: > >>>>> Hi, > >>>>> > >>>>> When I try to run a larger graph in the application that I developed > >>>>> here first get this message: > >>>>> > >>>>> hduser@node28:/usr/local/hama-trunk$ time bin/hama jar > aplic19jan.jar > >>>>> 65537 /hdfs/disco/dados/internet_a /hdfs/disco/outinternet_a 42 30 > >>>>> 13/01/22 10:59:14 INFO bsp.FileInputFormat: Total input paths to > process : 1 > >>>>> 13/01/22 10:59:14 INFO bsp.FileInputFormat: Total input paths to > process : 1 > >>>>> 13/01/22 10:59:17 INFO bsp.BSPJobClient: Running job: > job_201301221058_0002 > >>>>> 13/01/22 10:59:20 INFO bsp.BSPJobClient: Current supersteps number: 0 > >>>>> 13/01/22 10:59:32 INFO bsp.BSPJobClient: Current supersteps number: 1 > >>>>> 13/01/22 10:59:32 INFO bsp.BSPJobClient: The total number of > supersteps: 1 > >>>>> 13/01/22 10:59:32 INFO bsp.BSPJobClient: Counters: 6 > >>>>> 13/01/22 10:59:32 INFO bsp.BSPJobClient: > >>>>> org.apache.hama.bsp.JobInProgress$JobCounter > >>>>> 13/01/22 10:59:32 INFO bsp.BSPJobClient: SUPERSTEPS=1 > >>>>> 13/01/22 10:59:32 INFO bsp.BSPJobClient: LAUNCHED_TASKS=1 > >>>>> 13/01/22 10:59:32 INFO bsp.BSPJobClient: > >>>>> org.apache.hama.bsp.BSPPeerImpl$PeerCounter > >>>>> 13/01/22 10:59:32 INFO bsp.BSPJobClient: SUPERSTEP_SUM=1 > >>>>> 13/01/22 10:59:32 INFO bsp.BSPJobClient: TIME_IN_SYNC_MS=70 > >>>>> 13/01/22 10:59:32 INFO bsp.BSPJobClient: IO_BYTES_READ=3467587 > >>>>> 13/01/22 10:59:32 INFO bsp.BSPJobClient: TASK_INPUT_RECORDS=42089 > >>>>> 13/01/22 10:59:32 INFO bsp.FileInputFormat: Total input paths to > process : 42 > >>>>> 13/01/22 10:59:35 INFO bsp.BSPJobClient: Running job: > job_201301221058_0001 > >>>>> 13/01/22 10:59:38 INFO bsp.BSPJobClient: Current supersteps number: 0 > >>>>> 13/01/22 11:00:38 INFO bsp.BSPJobClient: Current supersteps number: 1 > >>>>> attempt_201301221058_0001_000001_0: SLF4J: Class path contains > >>>>> multiple SLF4J bindings. > >>>>> attempt_201301221058_0001_000001_0: SLF4J: Found binding in > >>>>> > [jar:file:/usr/local/hama-trunk/lib/slf4j-log4j12-1.5.8.jar!/org/slf4j/impl/StaticLoggerBinder.class] > >>>>> attempt_201301221058_0001_000001_0: SLF4J: Found binding in > >>>>> > [file:/tmp/hama-hduser/bsp/local/groomServer/attempt_201301221058_0001_000001_0/work/org/slf4j/impl/StaticLoggerBinder.class] > >>>>> attempt_201301221058_0001_000001_0: SLF4J: See > >>>>> http://www.slf4j.org/codes.html#multiple_bindings for an > explanation. > >>>>> attempt_201301221058_0001_000001_0: 13/01/22 11:00:14 INFO > >>>>> sync.ZKSyncClient: Initializing ZK Sync Client > >>>>> attempt_201301221058_0001_000001_0: 13/01/22 11:00:14 INFO > >>>>> sync.ZooKeeperSyncClientImpl: Start connecting to Zookeeper! At > >>>>> node30/10.16.2.30:61003 > >>>>> attempt_201301221058_0001_000001_0: 13/01/22 11:00:16 INFO > ipc.Server: > >>>>> Starting SocketReader > >>>>> attempt_201301221058_0001_000001_0: 13/01/22 11:00:16 INFO > ipc.Server: > >>>>> IPC Server Responder: starting > >>>>> attempt_201301221058_0001_000001_0: 13/01/22 11:00:16 INFO > ipc.Server: > >>>>> IPC Server listener on 61003: starting > >>>>> attempt_201301221058_0001_000001_0: 13/01/22 11:00:16 INFO > >>>>> message.HadoopMessageManagerImpl: BSPPeer address:node30 port:61003 > >>>>> attempt_201301221058_0001_000001_0: 13/01/22 11:00:16 INFO > ipc.Server: > >>>>> IPC Server handler 0 on 61003: starting > >>>>> attempt_201301221058_0001_000001_0: 13/01/22 11:00:34 INFO > >>>>> graph.GraphJobRunner: Loading finished at 1 steps. > >>>>> 13/01/22 11:04:46 INFO bsp.BSPJobClient: Job failed. > >>>>> > >>>>> real 5m34.523s > >>>>> user 0m4.832s > >>>>> sys 0m0.960s > >>>>> > >>>>> Analyzing logs the one that fails has this log: > >>>>> > >>>>> /************************************************************ > >>>>> STARTUP_MSG: Starting GroomServer > >>>>> STARTUP_MSG: host = node30/10.16.2.30 > >>>>> STARTUP_MSG: args = [node30] > >>>>> STARTUP_MSG: version = 1.0.0 > >>>>> STARTUP_MSG: build = > >>>>> https://svn.apache.org/repos/asf/hadoop/common/branches/branch-1.0-r > >>>>> 1214675; compiled by 'hortonfo' on Fri Dec 16 20:01:27 UTC 2011 > >>>>> ************************************************************/ > >>>>> 2013-01-22 10:58:51,118 INFO org.apache.hama.bsp.GroomServer: groom > start > >>>>> 2013-01-22 10:58:51,420 INFO org.apache.hama.bsp.GroomServer: > >>>>> /tmp/hama-hduser/bsp/local > >>>>> 2013-01-22 10:58:55,417 INFO org.apache.hadoop.ipc.Server: Starting > SocketReader > >>>>> 2013-01-22 10:58:55,452 INFO org.apache.hadoop.ipc.Server: IPC Server > >>>>> listener on 50000: starting > >>>>> 2013-01-22 10:58:55,452 INFO org.apache.hadoop.ipc.Server: IPC Server > >>>>> Responder: starting > >>>>> 2013-01-22 10:58:55,455 INFO org.apache.hama.bsp.GroomServer: Worker > >>>>> rpc server --> node30:50000 > >>>>> 2013-01-22 10:58:55,455 INFO org.apache.hadoop.ipc.Server: IPC Server > >>>>> handler 0 on 50000: starting > >>>>> 2013-01-22 10:58:55,560 INFO org.mortbay.log: Logging to > >>>>> org.slf4j.impl.Log4jLoggerAdapter(org.mortbay.log) via > >>>>> org.mortbay.log.Slf4jLog > >>>>> 2013-01-22 10:58:55,645 INFO org.apache.hama.bsp.GroomServer: > starting > >>>>> webserver: node30 > >>>>> 2013-01-22 10:58:55,645 INFO org.apache.hama.http.HttpServer: Port > >>>>> returned by webServer.getConnectors()[0].getLocalPort() before open() > >>>>> is -1. Opening the listener on 40015 > >>>>> 2013-01-22 10:58:55,646 INFO org.apache.hama.http.HttpServer: > >>>>> listener.getLocalPort() returned 40015 > >>>>> webServer.getConnectors()[0].getLocalPort() returned 40015 > >>>>> 2013-01-22 10:58:55,646 INFO org.apache.hama.http.HttpServer: Jetty > >>>>> bound to port 40015 > >>>>> 2013-01-22 10:58:55,646 INFO org.mortbay.log: jetty-6.1.14 > >>>>> 2013-01-22 10:58:56,252 INFO org.mortbay.log: Started > >>>>> SelectChannelConnector@node30:40015 > >>>>> 2013-01-22 10:59:01,574 INFO org.apache.hadoop.ipc.Server: Starting > SocketReader > >>>>> 2013-01-22 10:59:01,576 INFO org.apache.hadoop.ipc.Server: IPC Server > >>>>> Responder: starting > >>>>> 2013-01-22 10:59:01,576 INFO org.apache.hadoop.ipc.Server: IPC Server > >>>>> listener on 59102: starting > >>>>> 2013-01-22 10:59:01,576 INFO org.apache.hadoop.ipc.Server: IPC Server > >>>>> handler 0 on 59102: starting > >>>>> 2013-01-22 10:59:01,580 INFO org.apache.hadoop.ipc.Server: IPC Server > >>>>> handler 1 on 59102: starting > >>>>> 2013-01-22 10:59:01,577 INFO org.apache.hadoop.ipc.Server: IPC Server > >>>>> handler 2 on 59102: starting > >>>>> 2013-01-22 10:59:01,584 INFO org.apache.hadoop.ipc.Server: IPC Server > >>>>> handler 4 on 59102: starting > >>>>> 2013-01-22 10:59:01,583 INFO org.apache.hadoop.ipc.Server: IPC Server > >>>>> handler 3 on 59102: starting > >>>>> 2013-01-22 10:59:01,584 INFO org.apache.hadoop.ipc.Server: IPC Server > >>>>> handler 5 on 59102: starting > >>>>> 2013-01-22 10:59:01,584 INFO org.apache.hadoop.ipc.Server: IPC Server > >>>>> handler 6 on 59102: starting > >>>>> 2013-01-22 10:59:01,585 INFO org.apache.hadoop.ipc.Server: IPC Server > >>>>> handler 7 on 59102: starting > >>>>> 2013-01-22 10:59:01,585 INFO org.apache.hadoop.ipc.Server: IPC Server > >>>>> handler 8 on 59102: starting > >>>>> 2013-01-22 10:59:01,585 INFO org.apache.hadoop.ipc.Server: IPC Server > >>>>> handler 9 on 59102: starting > >>>>> 2013-01-22 10:59:06,899 INFO org.apache.hama.bsp.GroomServer: > >>>>> GroomServer up at: 127.0.0.1/127.0.0.1:59102 > >>>>> 2013-01-22 10:59:06,900 INFO org.apache.hama.bsp.GroomServer: > Starting > >>>>> groom: node30:50000 > >>>>> 2013-01-22 10:59:18,083 INFO org.apache.hama.bsp.GroomServer: Launch > 1 tasks. > >>>>> 2013-01-22 10:59:18,283 WARN org.apache.hadoop.util.NativeCodeLoader: > >>>>> Unable to load native-hadoop library for your platform... using > >>>>> builtin-java classes where applicable > >>>>> 2013-01-22 10:59:23,057 INFO org.apache.hama.bsp.GroomServer: Task > >>>>> 'attempt_201301221058_0002_000000_0' has started. > >>>>> 2013-01-22 10:59:31,403 INFO org.apache.hama.bsp.GroomServer: Task > >>>>> attempt_201301221058_0002_000000_0 is done. > >>>>> 2013-01-22 10:59:35,757 INFO org.apache.hama.bsp.GroomServer: Launch > 3 tasks. > >>>>> 2013-01-22 10:59:56,486 INFO org.apache.hama.bsp.GroomServer: Task > >>>>> 'attempt_201301221058_0001_000000_0' has started. > >>>>> 2013-01-22 10:59:56,487 INFO org.apache.hama.bsp.GroomServer: Launch > 3 tasks. > >>>>> 2013-01-22 11:00:06,973 INFO org.apache.hama.bsp.GroomServer: Task > >>>>> 'attempt_201301221058_0001_000002_0' has started. > >>>>> 2013-01-22 11:00:06,974 INFO org.apache.hama.bsp.GroomServer: Launch > 3 tasks. > >>>>> 2013-01-22 11:00:12,927 INFO org.apache.hama.bsp.GroomServer: Task > >>>>> 'attempt_201301221058_0001_000001_0' has started. > >>>>> 2013-01-22 11:03:50,330 ERROR org.apache.hama.bsp.TaskRunner: Error > >>>>> when executing BSPPeer process. > >>>>> java.io.IOException: BSP task process exit with nonzero status of > 137. > >>>>> command = [/usr/lib/jvm/jdk1.7.0_11/jre/bin/java, -Xmx6144m, > >>>>> -classpath, > /usr/local/hama-trunk/bin/../conf:/usr/lib/jvm/java-7-oracle//lib/tools.jar:/usr/local/hama-trunk/bin/../core/target/classes:/usr/local/hama-trunk/bin/../hama-**.jar:/usr/local/hama-trunk/bin/../lib/ant-1.7.1.jar:/usr/local/hama-trunk/bin/../lib/ant-launcher-1.7.1.jar:/usr/local/hama-trunk/bin/../lib/avro-1.6.0.jar:/usr/local/hama-trunk/bin/../lib/avro-ipc-1.6.0.jar:/usr/local/hama-trunk/bin/../lib/commons-cli-1.2.jar:/usr/local/hama-trunk/bin/../lib/commons-configuration-1.7.jar:/usr/local/hama-trunk/bin/../lib/commons-httpclient-3.0.1.jar:/usr/local/hama-trunk/bin/../lib/commons-lang-2.6.jar:/usr/local/hama-trunk/bin/../lib/commons-logging-1.1.1.jar:/usr/local/hama-trunk/bin/../lib/commons-math3-3.0.jar:/usr/local/hama-trunk/bin/../lib/guava-10.0.1.jar:/usr/local/hama-trunk/bin/../lib/hadoop-core-1.0.0.jar:/usr/local/hama-trunk/bin/../lib/hadoop-test-1.0.0.jar:/usr/local/hama-trunk/bin/../lib/jackson-core-asl-1.9.2.jar:/usr/local/hama-trunk/bin/../lib/jackson-mapper-asl-1.9.2.jar:/usr/local/hama-trunk/bin/../lib/jetty-6.1.14.jar:/usr/local/hama-trunk/bin/../lib/jetty-annotations-6.1.14.jar:/usr/local/hama-trunk/bin/../lib/jetty-util-6.1.14.jar:/usr/local/hama-trunk/bin/../lib/jsp-2.1-6.1.14.jar:/usr/local/hama-trunk/bin/../lib/jsp-api-2.1-6.1.14.jar:/usr/local/hama-trunk/bin/../lib/junit-4.8.1.jar:/usr/local/hama-trunk/bin/../lib/log4j-1.2.16.jar:/usr/local/hama-trunk/bin/../lib/netty-3.2.6.Final.jar:/usr/local/hama-trunk/bin/../lib/servlet-api-6.0.32.jar:/usr/local/hama-trunk/bin/../lib/slf4j-api-1.5.8.jar:/usr/local/hama-trunk/bin/../lib/slf4j-log4j12-1.5.8.jar:/usr/local/hama-trunk/bin/../lib/snappy-java-1.0.4.1.jar:/usr/local/hama-trunk/bin/../lib/zookeeper-3.4.5.jar::/tmp/hama-hduser/bsp/local/groomServer/attempt_201301221058_0001_000001_0/work/classes:/tmp/hama-hduser/bsp/local/groomServer/attempt_201301221058_0001_000001_0/work, > >>>>> org.apache.hama.bsp.GroomServer$BSPPeerChild, 127.0.0.1, 59102, > >>>>> attempt_201301221058_0001_000001_0, node30, -1, RUNNING] > >>>>> at > org.apache.hama.bsp.TaskRunner$BspChildRunner.call(TaskRunner.java:126) > >>>>> at > java.util.concurrent.FutureTask$Sync.innerRun(FutureTask.java:334) > >>>>> at java.util.concurrent.FutureTask.run(FutureTask.java:166) > >>>>> at > java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.access$201(ScheduledThreadPoolExecutor.java:178) > >>>>> at > java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.run(ScheduledThreadPoolExecutor.java:292) > >>>>> at > java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1110) > >>>>> at > java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:603) > >>>>> at java.lang.Thread.run(Thread.java:722) > >>>>> 2013-01-22 11:04:39,789 INFO org.apache.hama.bsp.GroomServer: adding > >>>>> purge task: attempt_201301221058_0001_000001_0 > >>>>> 2013-01-22 11:04:39,884 INFO org.apache.hama.bsp.GroomServer: About > to > >>>>> purge task: attempt_201301221058_0001_000001_0 > >>>>> 2013-01-22 11:04:41,170 INFO org.apache.hama.bsp.GroomServer: Kill 1 > tasks. > >>>>> 2013-01-22 11:04:41,219 INFO org.apache.hama.bsp.GroomServer: Kill 1 > tasks. > >>>>> 2013-01-22 11:04:41,589 INFO org.apache.hama.bsp.GroomServer: Kill 1 > tasks. > >>>>> > >>>>> > >>>>> What can it be? Some configuration? To not get the message of Java > >>>>> Heap Space configured: > >>>>> > >>>>> <name> bsp.child.java.opts </ name> > >>>>> <value>-Xmx6144m </ value> > >>>>> > >>>>> I have only 4 GB of memory in each Groomserver, can it be that? I am > >>>>> using the trunk version of Hama. > >>>>> > >>>>> -- > >>>>> Francisco Sanches > >>>> > >>>> > >>>> > >>>> -- > >>>> Best Regards, Edward J. Yoon > >>>> @eddieyoon > >>> > >>> > >>> > >>> -- > >>> Francisco Sanches > >> > >> > >> > >> -- > >> Best Regards, Edward J. Yoon > >> @eddieyoon > > > > > > > > -- > > Francisco Sanches > > > > -- > Best Regards, Edward J. Yoon > @eddieyoon >
