hive-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From <>
Subject Run hive jobs as correct HDFS user instead of as hive user
Date Tue, 19 Jun 2012 15:40:16 GMT

I would like to force all mapreduce jobs run from the Hive shell to run as the hdfs user who
ran them instead of as the "hive" user. For instance, I have HDFS testuser1 logged into the
edge node under their unix user with the same name testuser1. This user begins a hive shell
and kicks off a hive query job. This job is always run as the "hive" user in HDFS and therefore
all temporary files reside in the hive users' HDFS folder. This is a problem.

I'm using the following for temp file location: hive.exec.scratchdir = /user/${}/.hive-temp/
. The intent is to ensure that every user puts their own temp files and intermediate job files
in their own user directory so that we can track user disk usage correctly. Since Hive jobs
are run as the "hive" user, temp files always end up in /user/hive/.hive-temp/.

Is it possible to locate hive temp files in the user who runs the hive job?


Shawn Higgins
Systems Engineer
Thomson Reuters

View raw message