Hi Chris,
Our policy is "PRIORITY", with the current settings being...
NODECFG[DEFAULT] PRIORITYF='SPEED - .0001 * CMEM - 10 * JOBCOUNT'
Cheers,
Rob
> -----Original Message-----
> From: Testbed Support for GridPP member institutes
> [mailto:[log in to unmask]] On Behalf Of Chris Curtis
> Sent: Wednesday, July 29, 2009 12:13 PM
> To: [log in to unmask]
> Subject: Re: UKI HammerCloud Test 2: (Tuesday 28th July):
> This Time It's Personal
>
> Hi -
>
> I increased the number of pilot processes allowed at
> Birmingham, and I think we're starting to see stress effects now.
>
> The first thing we noticed is that the pilot jobs seem to
> cluster on some CPUs, leaving others empty. Our Maui
> NodeAllocationPolicy is "LOAD" at the moment, but maybe this
> isn't the right choice if it takes a while for pilot jobs to
> start the real heavy lifting. What kind of allocation
> policies are other sites using?
>
> Cheers,
>
> Chris
>
> --
> West 326
> Physics and Astronomy
> University of Birmingham
> Edgbaston
> Birmingham
> B15 2TT
>
> (Office) 0121 414 4700
> (Mobile) 0798 666 1959
>
--
Scanned by iCritical.
|