Re: [Condor-users] hawkeye on dual-processor nodes

[Nick LeRoy <nleroy@xxxxxxxxxxx>]

On Mon November 27 2006 3:40 pm, Junjun Mao wrote:
> Deal all,
Hello

> I installed hawkeye on a Condor test pool. It is intended to monitor
> semaphore arrays and publish the machine specific shared memory state
> to ClassAds. I got the following unexpected outcome as reported by
> condor_status and hawkeye_status. One more machine is added to each
> node, which was already divided into two virtual machines:

In general, you want to have your hawkeye startds reporting to their own 
separate collector.  This prevents the hawkeye startd ad from clobbering the 
condor machine ad.  In your case, because you're running on an MP machine, 
the hawkeye configuration has a default "NUM_CPUS = 1", so that's why you see 
the "node74" and the "vmx@node74" side by side.  On a single CPU machine, 
you'd only see the "node74" ads, but they'd sometimes be from the condor 
startd, and, at other times, from the hawkeye startd -- this is most 
certainly not what you want.

So, bottom line, start a second collector, and point all of the hawkeye 
startds at it.

Hope this helps

-Nick

-- 
           <<< Welcome to the real world. >>>
 /`-_    Nicholas R. LeRoy               The Condor Project
{     }/ http://www.cs.wisc.edu/~nleroy  http://www.cs.wisc.edu/condor
 \    /  nleroy@xxxxxxxxxxx              The University of Wisconsin
 |_*_|   608-265-5761                    Department of Computer Sciences