Re: Starcluster SGE usage
I would like to revive this thread as I am in a similar situation as the
original author, I would like to run jobs in a parallel environment where
the slot allocation would not extend beyond a given node.
Here is what John last wrote following a long exchange:
On 10/18/2012 01:55 PM, John St. John wrote:
> Ok just submitted a pull request. I modified the sge template
> "sge_pe_template" so that you can modify the allocation type in a
> new parallel environment. I modified the SGE plugin so that it
> makes by_node as well as ORTE, and by_node uses the $pe_slots
> allocation type. I have tested this out with creating a cluster
> (haven't tried adding/deleting nodes) and it seems to work. The
> changes are pretty minimal to get here so I feel pretty confident
> that I didn't add any new bugs.
>
> Best, John
>
However, I don't see any trace of these modifications in the repos? Was
this abandoned?
Could you suggest a strategy to enable the by_node parallel environment
and have it added to the all.q queue at init time? (I see that I can
initialize it with qconf ap by_node then add it manually with qconf -mq
all.q)
Thanks
-- Marco Blanchette, Ph.D.
Stowers Institute for Medical Research
1000 East 50th Street
Kansas City MO 64110
www.stowers.org
Tel: 816-926-4071
Cell: 816-726-8419
Fax: 816-926-2018
Received on Tue May 28 2013 - 20:12:38 EDT
This archive was generated by
hypermail 2.3.0.