hadoop-mapreduce-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Bertrand Dechoux <decho...@gmail.com>
Subject Re: Configuration of hadoop.job.history.user.location on a user basis
Date Fri, 14 Sep 2012 05:38:54 GMT
Hi Hemanth,

I thought of that but I didn't test it because I need the same for the job
id. Of course, I could generate my own id but I don't like that solution. I
will look at hadoop.tmp.dir to see if it can solve my issue differently.



On Fri, Sep 14, 2012 at 5:54 AM, Hemanth Yamijala <yhemanth@thoughtworks.com
> wrote:

> Hi Bretrand,
> Could you try using user.name as a property in the value of the config
> item ? Something like /user/${user.name}/history ?
> Thanks
> hemanth
> On Thu, Sep 13, 2012 at 5:46 PM, Bertrand Dechoux <dechouxb@gmail.com>wrote:
>> Hi,
>> I would like to be able to fully isolate the quota of each user.
>> That is, I would like to put everything related to a user into her/his
>> home directory /user/<user>.
>> Among other points of configuration, I would like to be able to configure
>> hadoop.job.history.user.location.
>> But I can't see how it could be done.
>> How can I have a different path for each user?
>> How do I make sure that each job will still have its own directory? (Even
>> when no output dir is defined for the job.)
>> Other question, slightly related. If I deactivate those logs using the
>> "none" configuration, it should only impact the copy of the user, should it
>> not?
>> The user should still be able to browse the logs using the version
>> provided through the jobtracker web interface, should it not?
>> Regards
>> Bertrand

Bertrand Dechoux

View raw message