[ 
https://issues.apache.org/jira/browse/YARN-7381?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Eric Yang resolved YARN-7381.
-----------------------------
    Resolution: Fixed

When NM_LOG_CONTAINER_DEBUG_INFO is enabled, and there is a problem to execute 
container script.  A ExitCodeException would be thrown to notify the execution 
failure.  By default, I think this is correct to make sure failure are notified 
to caller.  The past behavior swallows the exception, which is not exactly 
correct.  Most people don't use DefaultContainerExecutor, and container 
launcher code is usually successful.  This is the reason that swallowed 
exception was not noticeable.

TestContainerLaunch unit test is not quite accurate because the script 
instructed to execute "hello" script, which does not exist.  Hence, the 
throwing of exception is the proper behavior.  I am inclined to close this 
issue as fixed.  When ExitCodeException triggers more exception else where, it 
will help developer to look at the root causes that trigger launcher failures 
more closely.

> Enable the configuration: yarn.nodemanager.log-container-debug-info.enabled 
> by default in yarn-default.xml
> ----------------------------------------------------------------------------------------------------------
>
>                 Key: YARN-7381
>                 URL: https://issues.apache.org/jira/browse/YARN-7381
>             Project: Hadoop YARN
>          Issue Type: Bug
>    Affects Versions: 2.9.0, 3.0.0, 3.1.0
>            Reporter: Xuan Gong
>            Assignee: Xuan Gong
>            Priority: Critical
>             Fix For: 3.0.0
>
>         Attachments: 
> TEST-org.apache.hadoop.yarn.server.nodemanager.containermanager.launcher.TestContainerLaunch.xml,
>  YARN-7381.1.patch, 
> org.apache.hadoop.yarn.server.nodemanager.containermanager.launcher.TestContainerLaunch-output.txt,
>  
> org.apache.hadoop.yarn.server.nodemanager.containermanager.launcher.TestContainerLaunch.txt
>
>
> Enable the configuration "yarn.nodemanager.log-container-debug-info.enabled", 
> so we can aggregate launch_container.sh and directory.info



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

---------------------------------------------------------------------
To unsubscribe, e-mail: yarn-dev-unsubscr...@hadoop.apache.org
For additional commands, e-mail: yarn-dev-h...@hadoop.apache.org

Reply via email to