[Beowulf] SGE Queue Priorities
Robert G. Brown
rgb at phy.duke.edu
Mon Dec 5 05:15:58 PST 2005
On Sun, 4 Dec 2005, Mark Hahn wrote:
>> Does anyone have any thoughts?
> egads. when people post this kind of very complicated query about
> configuring SGE/LSF/PBS/etc, I always have to wonder whether they've
> actually asked "do I need a big "full-featured" queue/scheduler?".
> personally, I find that as in many domains, feature-completeness
> is strongly correlated with "so complicated-to-configure that it verges on
> the tail wagging the dog".
Sure Mark, but until you "finish" (polish up) up your personal scheduler
and wrap it up all pretty and well documented in rpm format as an OS
project that others can use, people are stuck either using one of the
ones that exist or writing their own. Not everybody can roll their own.
In fact, damn few people can roll their own, especially if they do a CBA
comparison of RTO and using SGE and suffering through the bloat and aimk
This isn't really critical of you -- I've meant to write a "simple"
cluster scheduler on top of xmlsysd (which can provide all the
information required on a lightweight connection reasonably efficiently)
for some time and I haven't had time either -- too many projects all
uncompleted and pressure on to write grants and do physics and teach and
everything. But at some point SOMEBODY on this list in SOME project --
perhaps warewulf, perhaps wulfware, perhaps hahnware;-) -- needs to
write up a SIMPLE portable queuing system that can be built/installed
via rpm with SIMPLE administrative/policy hooks -- perhaps an xml-based
config file that just defines node "ownership" and priorities.
> Beowulf mailing list, Beowulf at beowulf.org
> To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf
Robert G. Brown http://www.phy.duke.edu/~rgb/
Duke University Dept. of Physics, Box 90305
Durham, N.C. 27708-0305
Phone: 1-919-660-2567 Fax: 919-660-2525 email:rgb at phy.duke.edu
More information about the Beowulf