[ https://issues.apache.org/jira/browse/SPARK-4360?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14313898#comment-14313898 ]
seekerak edited comment on SPARK-4360 at 2/10/15 9:37 AM: ---------------------------------------------------------- i have resolved this issue by configure yarn scheduler like this: <property> <name>yarn.resourcemanager.scheduler.class</name> <value>org.apache.hadoop.yarn.server.resourcemanager.scheduler.fair.FairScheduler</value> </property> or <property> <name>yarn.resourcemanager.scheduler.class</name> <value>org.apache.hadoop.yarn.server.resourcemanager.scheduler.capacity.CapacityScheduler</value> </property> and <property> <name>yarn.scheduler.capacity.resource-calculator</name> <value>org.apache.hadoop.yarn.util.resource.DominantResourceCalculator</value> </property> so the real reason is yarn's resource scheduler, if one node can provide all resource that tasks required, all tasks maybe run on one node only. was (Author: omicronak): i have resolved this issue by configure yarn scheduler like this: <property> <name>yarn.resourcemanager.scheduler.class</name> <value>org.apache.hadoop.yarn.server.resourcemanager.scheduler.fair.FairScheduler</value> </property> or <property> <name>yarn.resourcemanager.scheduler.class</name> <value>org.apache.hadoop.yarn.server.resourcemanager.scheduler.capacity.CapacityScheduler</value> </property> and <property> <name>yarn.scheduler.capacity.resource-calculator</name> <value>org.apache.hadoop.yarn.util.resource.DominantResourceCalculator</value> </property> so the real reason is yarn's resource scheduler, if one node can provide all resource that tasks required, all task maybe run on one node only. > task only execute on one node when spark on yarn > ------------------------------------------------ > > Key: SPARK-4360 > URL: https://issues.apache.org/jira/browse/SPARK-4360 > Project: Spark > Issue Type: Bug > Components: YARN > Affects Versions: 1.0.2 > Reporter: seekerak > > hadoop version: hadoop 2.0.3-alpha > spark version: 1.0.2 > when i run spark jobs on yarn, i found all the task only run on one node, my > cluster has 4 nodes, executors has 3, but only one has task, the others > hasn't, my command like this : > /opt/hadoopcluster/spark-1.0.2-bin-hadoop2/bin/spark-submit --class > org.sr.scala.Spark_LineCount_G0 --executor-memory 2G --num-executors 12 > --master yarn-cluster /home/Spark_G0.jar /data /output/ou_1 > is there any one knows why? -- This message was sent by Atlassian JIRA (v6.3.4#6332) --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org For additional commands, e-mail: issues-h...@spark.apache.org