[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[cgs-wg] [CGS-WG Bug 14] Heidi Neumann's proposed class/association additions (OpenPBS)
http://bugzilla.gridforum.org/cgs-wg/show_bug.cgi?id=14
flon@isi.edu changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |mihajlov@de.ibm.com
------- Additional Comments From flon@isi.edu 2003-05-08 11:29 -------
Hi,
yesterday Heidi and I had the chance to talk to Waiman Chan (LoadLeveler)
about the proposed extensions to the JSIM model. As a reminder: OpenPBS allows
to specify a group of nodes (execution targets) on which the job can be run
and the proposal was to reflect this appropriately in the model, if other
scheduling systems have this capability too.
We learned that LoadLeveler also has the capability to restrict a job to a
group of nodes. The difference is in the way these execution targets are
specified. For OpenPBS they are statically defined entities, so
called "Resources", that define the nodes belonging to the group. With
LoadLeveler a job can require the presence of certain "Features", where a
feature is an arbitrarily defined property associated to a node. In contrast
to OpenPBS multiple features can be specified for LoadLeveler jobs, thus
allowing a fine grained selection of the execution targets.
Our proposal therefore would be to indeed add the node grouping concept to the
model using a to-be-defined neutral name for the grouping class (i.e., neither
feature nor resource). The only structural change to Heidi's original
proposal would be to change the cardinality of the association between Job and
the "Group of Systems" class from one-to-one to one-to-many.
We're looking forward to your comments and would definitely welcome
suggestions for a nice name for the "Group of Systems" class. I hope that we
can close this item within the next two weeks.
Mit freundlichen Grüßen / Kind Regards
Viktor Mihajlovski
Linux Technology Center
IBM Laboratory Böblingen, Germany
Phone +49-7031-16-2560
E-Mail mihajlov@de.ibm.com
------- You are receiving this mail because: -------
You are the QA contact for the bug, or are watching the QA contact.