Re: [Condor-users] breadth-first depth-first order control in dagman

[Stuart Anderson <anderson@xxxxxxxxxxxxxxxx>]

Kent,
	Strangely enough, I just had this discussion with one of my users
yesterday and was going to ask for the same DAGman enhancement on Monday.
In our case the motivation is to get an option for "depth first" so that
users developing new algorithms/pipelines can get to the end game sooner
rather than later to decide if they need to abort the whole dag and refine
their pipeline.

Thanks.

On Fri, Nov 17, 2006 at 09:47:12AM -0600, R. Kent Wenger wrote:
> On Fri, 17 Nov 2006, [ISO-8859-1] Horvátth Szabolcs wrote:
> 
> > Is it possible to control whether new DAG jobs are submitted in
> > breadth-first or depth-first order?
> > I limit the job submission rate using the maxidle switch and while it
> > works nicely in
> > limiting the number of idle jobs in the queue, it has the unlikely
> > effect of always submitting
> > the parent tasks first and only start submitting the child jobs after
> > all "first level" parents are submitted.
> > With bigger DAG structures this can delay the completion of the first
> > "branch" a lot.
> 
> I just looked at the code, and the way things work is that nodes are
> submitted in the order in which they become ready.  So any node at a
> higher level in the DAG will be submitted before a node at a lower level.
> 
> I think it would be fairly easy to add a "depth first vs. breadth first"
> configuration macro.  If anyone else out there would also like this
> feature, speak up -- the more people who want it, the sooner it's likely
> to happen...
> 
> Kent Wenger
> Condor Team
> 

> _______________________________________________
> Condor-users mailing list
> To unsubscribe, send a message to condor-users-request@xxxxxxxxxxx with a
> subject: Unsubscribe
> You can also unsubscribe by visiting
> https://lists.cs.wisc.edu/mailman/listinfo/condor-users
> 
> The archives can be found at either
> https://lists.cs.wisc.edu/archive/condor-users/
> http://www.opencondor.org/spaces/viewmailarchive.action?key=CONDOR

-- 
Stuart Anderson  anderson@xxxxxxxxxxxxxxxx
http://www.ligo.caltech.edu/~anderson