The client machine submitting the job does not need to be running on the same server as the JobTracker (or any other Hadoop/Hive presence). In your Hadoop/Hive configuration files, you just set fs.default.name, mapred.job.tracker, and hive.metastore.uris, and you can connect to MapReduce clusters and Hive metastores anywhere.
- Aaron On Sat, Apr 25, 2009 at 7:04 PM, Raghu R <[email protected]> wrote: > Hi, > Is it possible to have a web interface for hive where jobs could > be automatically submitted in JSP pages, and output retrieved locally, with > the user not even aware the processing is done in mapreduce cluster. > I saw the webUI for Hive, is it possible to control the jobs from a > remote system [ not the master]. > > Thanks > Raghu R >
