[jira] [Created] (SPARK-12759) Spark should fail fast if --executor-memory is too small for spark to start

["Imran Rashid (JIRA)" <ji...@apache.org>]

Imran Rashid created SPARK-12759:
------------------------------------

             Summary: Spark should fail fast if --executor-memory is too small for spark to start
                 Key: SPARK-12759
                 URL: https://issues.apache.org/jira/browse/SPARK-12759
             Project: Spark
          Issue Type: Bug
    Affects Versions: 1.6.0
            Reporter: Imran Rashid


With the UnifiedMemoryManager, the minimum memory for executor and driver JVMs was increased to 450MB.  There is code in {{UnifiedMemoryManager}} to provide a helpful warning if less than that much memory is provided.

However if you set {{--executor-memory}} to something less than that, from the driver process you just see executor failures with no warning, since the more meaningful errors are buried in the executor logs.  Eg., on Yarn, you see

{noformat}
16/01/11 13:59:32 WARN cluster.YarnSchedulerBackend$YarnSchedulerEndpoint: Container marked as failed: container_1452548703600_0001_01_000002 on host: imran-adhoc-2.vpc.cloudera.com. Exit status: 1. Diagnostics: Exception from container-launch.
Container id: container_1452548703600_0001_01_000002
Exit code: 1
Stack trace: ExitCodeException exitCode=1: 
	at org.apache.hadoop.util.Shell.runCommand(Shell.java:561)
	at org.apache.hadoop.util.Shell.run(Shell.java:478)
	at org.apache.hadoop.util.Shell$ShellCommandExecutor.execute(Shell.java:738)
	at org.apache.hadoop.yarn.server.nodemanager.DefaultContainerExecutor.launchContainer(DefaultContainerExecutor.java:210)
	at org.apache.hadoop.yarn.server.nodemanager.containermanager.launcher.ContainerLaunch.call(ContainerLaunch.java:302)
	at org.apache.hadoop.yarn.server.nodemanager.containermanager.launcher.ContainerLaunch.call(ContainerLaunch.java:82)
	at java.util.concurrent.FutureTask.run(FutureTask.java:262)
	at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1145)
	at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:615)
	at java.lang.Thread.run(Thread.java:745)


Container exited with a non-zero exit code 1
{noformat}

Though there is already a message from {{UnifiedMemoryManager}} if there isn't enough memory for the driver, as long as this is being changed it would be nice if the message more clearly indicated the {{--driver-memory}} configuration as well.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org