[torquedev] jobid creation

Glen Beane glen.beane at gmail.com
Fri Apr 29 12:07:52 MDT 2011


On Fri, Apr 29, 2011 at 1:58 PM, John Rosenquist <
jrosenquist at adaptivecomputing.com> wrote:

> On 04/29/2011 10:26 AM, John Rosenquist wrote:
> > I've been looking at the job submission code and ran across something I
> find a bit puzzling.
> >
> > When creating a new job id, the number is pulled sequentially from
> > server.sv_qs.sv_jobidnumber
> > This is used in creating the job id. (i.e. system_name.#)
> >
> > Then the system is checked for a duplicate job name.
> > If a duplicate name is found in the old or new jobs, the job is thrown
> out of the system.
> >
> > As the numeric portion of the id is defined by torque, is there a reason
> that we don't simply choose the next sequential number and check if it's
> available? (And so on until we find a valid ID to use)
> >
> To be more specific, I'm not including the condition where the job is from
> another server. This is just referring to jobs initially created in the
> req_quejob function.
>
>
>
if the job is originating on the same server, then the only case there would
be a job ID conflict as far as I know is if the "next_job_number" is set in
qmgr to a value <= a job ID that is currently in use.  In that case I don't
see the harm in just skipping over that number.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.supercluster.org/pipermail/torquedev/attachments/20110429/3ca7ec7e/attachment.html 


More information about the torquedev mailing list