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

Re: [Condor-users] Whole System Scheduling



On Mon, Oct 26, 2009 at 07:25:15PM +0000, Matt Hope wrote:
:would a simple poll testing the cpu load which returns once n consecutive tests under x % total load occur be a sufficient and simple check to be getting on with?
:A limit to the number of iterations would avoid issues with machines which are loaded for some other reason.
:
:If memory usage is the problem checking that instead might be better.

That's crazy enough it just might work, I've been thinking inside the
queue and in some case it would be nice to wait untill specificly for
queued jobs.  My current case likely does want to wait for system load
to drop off even id it is not condor though.

/proc/loadavg here I come

-Jon