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

Re: [Condor-users] Benchmarking condor

On 6/30/05, Kewley, J (John) <j.kewley@xxxxxxxx> wrote:
> > Take a look at the sum of:
> > condor_status -format "vm%d@" VirtualMachineId -format "%s " Machine
> > -format "%d " KFlops -format "%d\n" Mips
> >
> > Note that this may well report machines which have startd's but do not
> > run jobs, therefore they should be excluded.
> What type of machines have startds but don't execute jobs? Surely that is
> what startds are for?

I have a startd on my machine - not everyone can execute on it - only
myself and one other person.

Therefore blindly taking the sum of the reported values across all
machines would not be valid for my farm.

Using -constraint NODE_TYPE==\"FARM\" as well gives a more reasonable
number since that would include only machines targetted solely for the
farm. that said this segmentation may not be valid for other people...

This is why actually running jobs which will be run on the farm is the
best benchmark (though the most effort) since it reflects the reality
of what machines are available for your job (and indeed what they are
actually doing in the meantime if you are cycle stealing.