[
https://issues.apache.org/jira/browse/HIVE-2422?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13400063#comment-13400063
]
Anurag Tangri commented on HIVE-2422:
-------------------------------------
Looks like scratch did is not being cleaned at lot of locations. Another such
location:
1. ExecDriver.java's execute() function.
Here, if it is created before launching a job and there is error in job launch,
it is not cleaned in exception before returning :
try {
if (ctx == null) {
ctx = new Context(job);
ctxCreated = true;
}
emptyScratchDirStr = ctx.getMRTmpFileURI();
emptyScratchDir = new Path(emptyScratchDirStr);
FileSystem fs = emptyScratchDir.getFileSystem(job);
fs.mkdirs(emptyScratchDir);
} catch (IOException e) {
e.printStackTrace();
console.printError("Error launching map-reduce job", "\n"
+ org.apache.hadoop.util.StringUtils.stringifyException(e));
return 5;
}
Here, ctx.clear() needs to be called in exception.
-Anurag Tangri
> remove the intermediate dir when the hive query finish
> -------------------------------------------------------
>
> Key: HIVE-2422
> URL: https://issues.apache.org/jira/browse/HIVE-2422
> Project: Hive
> Issue Type: Bug
> Reporter: He Yongqiang
> Assignee: He Yongqiang
>
> right now if one hive query got compiled to 2 mr jobs, and the first job's
> output feed the second job. When the query finish, the first job's output
> should be removed.
--
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
