Vinod Kumar Vavilapalli commented on YARN-2934:

This seems like it is same as YARN-2013/YARN-2242.

For the AM, we ended up simply putting a log-message "AM Container for 
\$ApplicationAttemptId exited with exitCode: \$ExitStatus.\nFor more detailed 
output, check application tracking page: \$TrackingUrl, Then, click on links to 
logs of each attempt.\n"

You really don't want to "cat" stderr from containers. Containers may run for a 
very long time, spewing a lot of errors in stderr before finally failing. NM 
unconditionally reading logs in such cases will blow up NM heap. We either do a 
cross-platform way of tailing the last N bytes (not terribly useful if we cut 
lines mid way through) or better simply print a link to take them to the right 
set of logs.

> Improve handling of container's stderr 
> ---------------------------------------
>                 Key: YARN-2934
>                 URL: https://issues.apache.org/jira/browse/YARN-2934
>             Project: Hadoop YARN
>          Issue Type: Improvement
>            Reporter: Gera Shegalov
>            Assignee: Naganarasimha G R
>            Priority: Critical
> Most YARN applications redirect stderr to some file. That's why when 
> container launch fails with {{ExitCodeException}} the message is empty.

This message was sent by Atlassian JIRA

Reply via email to