F
---------------------------------------------
Francis E. Reyes M.Sc.
215 UCB
University of Colorado at Boulder
gpg --keyserver pgp.mit.edu --recv-keys 67BA8D5D
8AE2 F2F4 90F7 9640 28BC 686F 78FD 6669 67BA 8D5D
What is an art script?
Brian
> --
> You received this message because you are subscribed to the Google Groups
> "PyXG" group.
> To post to this group, send email to py...@googlegroups.com.
> To unsubscribe from this group, send email to
> pyxg+uns...@googlegroups.com.
> For more options, visit this group at
> http://groups.google.com/group/pyxg?hl=en.
>
>
--
Brian E. Granger, Ph.D.
Assistant Professor of Physics
Cal Poly State University, San Luis Obispo
bgra...@calpoly.edu
elli...@gmail.com
It's a preflight script that the controller sends out to agents before
a task is run .. A number of criteria (memory, cpu's,
gpu's,etc,whatever you want) are each assigned a numerical 'score' and
all scores are multiplied together for a given machine. These are sent
back to the controller, which ranks the scores, and assigns tasks
based on the score list.
Presumably this system can rank agents based on whatever criteria you
want.. (i.e. don't run this job on a PPC/Intel or where load/mem/cpu
is <> some value).
I *think* the preflight scripts are assigned just before a task is
assigned allowing the controller to assess cpu load values of each
agent in 'realtime'. (for those agents who accept tasks even when not
idle).
http://www.macresearch.org/xgrid-leopard-scoreboard-rules
F
> Agent reporting tool....
Ahh, OK.
> It's a preflight script that the controller sends out to agents before a
> task is run .. A number of criteria (memory, cpu's, gpu's,etc,whatever you
> want) are each assigned a numerical 'score' and all scores are multiplied
> together for a given machine. These are sent back to the controller, which
> ranks the scores, and assigns tasks based on the score list.
I was not familiar with this, but it should not be too difficult to
add this. I don't have time to work on this myself, but feel free to
have a go yourself. The pyxg code is up on github:
So feel free to fork this and submit a pull request.
> Presumably this system can rank agents based on whatever criteria you want..
> (i.e. don't run this job on a PPC/Intel or where load/mem/cpu is <> some
> value).
>
> I *think* the preflight scripts are assigned just before a task is assigned
> allowing the controller to assess cpu load values of each agent in
> 'realtime'. (for those agents who accept tasks even when not idle).
It would be nice to have this capability for sure. Beat Rupp (the
other pyxg maintainer) may also be interested in this and have more
time to help in the development.
Cheers,
Brian