Hi Alessandro
>> 2./ What's the correct value for GlueSubClusterLogicalCPUs?
>>
>> I've set it up to 904 (the total number of cores) but since we have
>> sufficient memory per core, we allow some overbooking, and I wonder
>> if I should publish the total number of slots (allowed jobs) instead...
> if each of your WN has 8 core, 226*8=1808 is the right number; but, if
> I've well understood your case, if on some WN the overbooking makes to
> see to the O.S. more "computing units", you should take them into
> account in the "LogicalCPU number"
Just to clarify, in my case, each WN has 8 cores distributed in 2
quadcore CPUS. So, the total number of Logical CPUS should be 226 (2
quadcore CPUS per WN) * 4 (4 cores per quadcore CPU) = 904. On top of
that, I should apply the overbooking factor.
>>
>> 3./ I've set up "GlueHostProcessorOtherDescription:
>> Cores=8,Benchmark=54.90-HEP-SPEC06". I assume that the benchmark
>> value should be published by server (and not by core), and that the
>> number of cores is the total number of cores per server, and not per
>> CPU as stated in the docs (which in my case would be 4 since each
>> host has 2 quadcore cpus). Is my assumption right?
> actually you should publish "Cores=4,Benchmark=54.90-HEP-SPEC06",
> because 4 is the number of cores per cpu on the typical WN
After Gonzalo comment, I realized that the Benchmark value should be per
core.
Thanks in Advance
Cheers
Goncalo
|