[GE users] Can we deploy our own cluster scheduling algorithm using sun grid engine ?
rahul8590 at gmail.com
Wed Dec 30 17:29:46 GMT 2009
[ The following text is in the "iso-8859-1" character set. ]
[ Your display is set for the "ISO-8859-10" character set. ]
[ Some special characters may be displayed incorrectly. ]
On Wed, Dec 30, 2009 at 9:54 PM, templedf <dan.templeton at sun.com<mailto:dan.templeton at sun.com>> wrote:
The scheduling algorithm is technically pluggable, but it's a
compile-time operation. The code for the scheduler (the pseudo-daemon
that runs the scheduling algorithm) is in daemons/schedd. The code for
the scheduling algorithm is in libs/sched.
Basically i sense that it vaguely supports it. But can i check the time spent on executing a job in different infrastructure .
By different infrastructure i mean that initially setup cluster with 3 systems and then running the same job with more no of systems , by this i guess i can map the idle number of nodes required in a cluster for executing jobs keeping in mind the utilization of resources and the turn around time for the job.
More information about the gridengine-users