[torquedev] Data-Is-String format

Ken Nielson knielson at adaptivecomputing.com
Fri Jun 11 07:17:29 MDT 2010


pbs_sched is not the only scheduler used with TORQUE. In fact it is probably the least used scheduler with TORQUE. (I have no empirical data to confirm that but I would guess Maui and Moab make up the majority of schedulers)

Next the DIS is what is used for server to MOM and MOM to MOM communication. These daemons would have to be rewritten as well to access the database. 

Ken 

----- Original Message -----
From: "ekasit kijsipongse" <ekasit.kijsipongse at nectec.or.th>
To: "Mgr. �imon T๓th" <SimonT at mail.muni.cz>
Cc: "Torque Developers mailing list" <torquedev at supercluster.org>
Sent: Friday, June 11, 2010 12:59:16 AM
Subject: Re: [torquedev] Data-Is-String format

>
> Several questions:
> How do you want to use a database?
> Do you think that a database does not have this overhead?
>
> --

Of course, a database has some overhead.
But, since not all job information is used in scheduling, pbs_sched can
retrieve only the
information it needs such as job id, submission time. So this can
greatly reduce huge amount of data
in communication. What do you think?

Ekasit K.



>>    I'm wondering what is the benefit of using the Data-Is-String
>>    format in the
>> communication between pbs_server and pbs_sched, i.e. for reducing
>> message size
>> while maintaining readability? I've noticed a large overhead from
>> sending and
>> parsing the
>> job information to pbs_sched if the number of jobs > 10,000. Is there
>> any other
>> options such as using database?


_______________________________________________ torquedev mailing list
torquedev at supercluster.org
http://www.supercluster.org/mailman/listinfo/torquedev


More information about the torquedev mailing list