search

stratosphere / stratosphere

Stratosphere is now Apache Flink.
https://github.com/apache/incubator-flink
Apache License 2.0
197 stars 84 forks source link

Improve JobManager heap space memory configuration for YARN #887

Open rmetzger opened 10 years ago

rmetzger commented 10 years ago

just saw this while testing for the 0.5 release. The JM sometimes fails because I forgot to subtract a few %% from the JM heapspace for extra JVM allocations.

2014-05-29 12:01:53,770 WARN org.apache.hadoop.yarn.server.nodemanager.containermanager.monitor.ContainersMonitorImpl: Process tree for container: container_1401362984347_0002_01_000001 has processes older than 1 iteration running over the configured limit. Limit=1073741824, current usage = 1125031936
2014-05-29 12:01:53,776 WARN org.apache.hadoop.yarn.server.nodemanager.containermanager.monitor.ContainersMonitorImpl: Container [pid=2477,containerID=container_1401362984347_0002_01_000001] is running beyond physical memory limits. Current usage: 1.0 GB of 1 GB physical memory used; 1.7 GB of 5 GB virtual memory used. Killing container.
Dump of the process-tree for container_1401362984347_0002_01_000001 :
    |- PID PPID PGRPID SESSID CMD_NAME USER_MODE_TIME(MILLIS) SYSTEM_TIME(MILLIS) VMEM_USAGE(BYTES) RSSMEM_USAGE(PAGES) FULL_CMD_LINE
    |- 2483 2477 2477 2477 (java) 5021 548 1711837184 274360 /usr/java/latest/bin/java -Xmx1000M -Dlog.file=/mnt/var/log/hadoop/userlogs/application_1401362984347_0002/container_1401362984347_0002_01_000001/jobmanager-log4j.log -Dlog4j.configuration=file:log4j.properties eu.stratosphere.yarn.ApplicationMaster 
    |- 2477 2094 2477 2477 (bash) 0 1 117805056 306 /bin/bash -c /usr/java/latest/bin/java -Xmx1000M  -Dlog.file=/mnt/var/log/hadoop/userlogs/application_1401362984347_0002/container_1401362984347_0002_01_000001/jobmanager-log4j.log -Dlog4j.configuration=file:log4j.properties eu.stratosphere.yarn.ApplicationMaster  1>/mnt/var/log/hadoop/userlogs/application_1401362984347_0002/container_1401362984347_0002_01_000001/jobmanager-stdout.log 2>/mnt/var/log/hadoop/userlogs/application_1401362984347_0002/container_1401362984347_0002_01_000001/jobmanager-stderr.log 

2014-05-29 12:01:53,777 INFO org.apache.hadoop.yarn.server.nodemanager.containermanager.container.Container: Container container_1401362984347_0002_01_000001 transitioned from RUNNING to KILLING