Hi Adrian
At the Toronto Tier-2, I switched from openPBS to Condor to support the
CDF experiment in 2005, but I switched back to Torque/Maui in 2008.
Condor never seemed a good fit for a dedicated cluster where you are not
stealing cycles. There is no native concept of a queue so it was hard to
limit certain users or groups from dominating the cluster when workloads
from the grid were light or vice-versa. Also, the
edg/glite/egee/rgma/whatever support for Condor has always been a little
behind so every new version I had to fix information provisioning by hand.
Condor has a little too much intelligence for a dedicated cluster - it was
hard to predict what it would do sometimes and it had a habit of ignoring
certain schedulers occassionally for no apparent reason that I could ever
ascertain.
I know many of the US sites are running with it, but I have been happy
with the switch back to pbs and we will use Moab/torque on our new 30,000
core Nehalem cluster.
Leslie
On Wed, 13 May 2009, Adrian Sevcenco wrote:
> Hi,
> I really dont want to start a flame .. but i must ask :
> what are advantages and disadvantaged of maui_torque vs condor ?
> At first glance i would think that condor as and newer and monolithic
> approach is more stable and more easy to configure ...
> Is the condor used in gLite environments? What experiences did you had
> with it?
>
> Thank you,
> Adrian
>
--
,-~~-.___. ________________________________________________
/ | ' \ [log in to unmask] Department of Physics
( ) 0 Tel: +1-416-978-2959 University of Toronto
\_/-, ,----' Fax: +1-416-978-8221 60 St. George Street
==== // Toronto, ON M5S 1A7
/ \-'~; /~~~(O) Canada
/ __/~| / | Office: McLennan Physics Lab Room 911
=( _____| (_________| http://home.fnal.gov/~groer
Leslie S. Groer
|