You are viewing a plain text version of this content. The canonical link for it is here.
Posted to mapreduce-dev@hadoop.apache.org by "Adam Kramer (Created) (JIRA)" <ji...@apache.org> on 2011/10/22 03:00:38 UTC
[jira] [Created] (MAPREDUCE-3244) Make nodes fail/succeed based on
return codes (regardless of pending data)
Make nodes fail/succeed based on return codes (regardless of pending data)
--------------------------------------------------------------------------
Key: MAPREDUCE-3244
URL: https://issues.apache.org/jira/browse/MAPREDUCE-3244
Project: Hadoop Map/Reduce
Issue Type: Improvement
Reporter: Adam Kramer
Priority: Minor
If I were to run a MR job with a script like 'head -n10' for some reason, each node would execute head -n10 (good), and output 10 rows (good) and then the 'head' program would return success and quit (good), and all the nodes would fail (bad) because there's a ton of pending data (irrelevant).
For purposes of streaming data through programs, Hadoop should trust the program's return code.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira