[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Condor-users] Requirements not respected

On Thu, May 24, 2007 at 08:13:34AM +0200, Johan Bengtsson wrote:
> Hi,
> A similar discussion was brought up a few months ago. The concluding
> mail from this discussion may be helpful.
> https://lists.cs.wisc.edu/archive/condor-users/2007-January/msg00195.shtml

Indeed it looks very interesting though complicated (and hard to interpret
from a non-admin point of view). 
What I'd like to have is some code containing less "magic numbers": predefined
CPU and memory counts are hard to maintain if you got some inhomogeneous
pool and try to keep everything in one config file (even if there are different
CPU counts, so a redefinition of MEMORY=3*(ORIGINAL_MEMORY) wouldn't help:
with only 2 CPUs there would be 3 VMs, and the factor would be 2, etc.)

Obviously this is some piece of legacy stuff - Condor wasn't developed for
SMP machines in the first place, and it's static mapping of resources to
virtual machines is a serious limitation in certain situations.

I'm afraid this design "feature" isn't easily changed, and the Condor developers 
would have to  modify the current code from the bottom up... but surely it would
be a "nice to have" item for the 7.0 wishlist (now that multi-cores are emerging


Steffen Grunewald * MPI Grav.Phys.(AEI) * Am Mühlenberg 1, D-14476 Potsdam
Cluster Admin * http://pandora.aei.mpg.de/merlin/ * http://www.aei.mpg.de/
* e-mail: steffen.grunewald(*)aei.mpg.de * +49-331-567-{fon:7233,fax:7298}
No Word/PPT mails - http://www.gnu.org/philosophy/no-word-attachments.html