Srikanth Sundarrajan commented on YARN-3735:

When JRE fails with a fatal error during any of the container launchers, an 
error file is created in the working directory and this is being removed by the 
node manager upon container cleanup. It becomes challenging to debug in this 
scenario. It might be useful to collect this and append it to the container 
stderr when such errors are encountered.

> Retain JRE Fatal error logs upon container failure
> --------------------------------------------------
>                 Key: YARN-3735
>                 URL: https://issues.apache.org/jira/browse/YARN-3735
>             Project: Hadoop YARN
>          Issue Type: Improvement
>          Components: nodemanager
>            Reporter: Srikanth Sundarrajan

This message was sent by Atlassian JIRA

Reply via email to