[
https://issues.apache.org/jira/browse/HADOOP-1750?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12522654
]
Owen O'Malley commented on HADOOP-1750:
---------------------------------------
Sadly, I was wrong about the getInputStream. *sigh* It does always buffer it
and will block the launched process if it fills. Fortunately, only the output
from bash will be written to the stream since the command line redirects all of
the stdout/stderr to files. But it is worth getting the streams for the user if
the process fails. Sorry for the confusion.
I'm not a big fan of Java's process launching API.
> We should log better if something goes wrong with the process fork
> ------------------------------------------------------------------
>
> Key: HADOOP-1750
> URL: https://issues.apache.org/jira/browse/HADOOP-1750
> Project: Hadoop
> Issue Type: Improvement
> Components: mapred
> Reporter: Owen O'Malley
> Assignee: Owen O'Malley
> Fix For: 0.14.1
>
> Attachments: copy-err.patch
>
>
> Currently, if something goes wrong with bash, the user only gets the error
> code out. In the case of error result, we should copy the stdout and stderr
> from the bash process to the user logs.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
