Re: [HTCondor-users] computing element?

[Francesco Prelz <Francesco.Prelz@xxxxxxxxxx>]

On Tue, 22 Oct 2013, Pek Daniel wrote:

> Hi!
>
> Let's suppose there's an existing system with LSF as batch system and
> layers like CREAM as computing element, and ICE for the WMS.
> If I'd like to replace LSF with HTCondor, what can I use instead of CREAM
> in order to stay grid-compliant? (as far as I understood, CREAM doesn't
> support HTCondor as a batch system)

Hi,

the 'BLAH' layer that CREAM uses to ship jobs around *is* able to send 
jobs to Condor, and has been for many years. In fact we do run a 
CREAM-accessible Condor pool at the Milan Tier-2 center. Given the (so 
far) limited interest in this configuration (for a long while only Oxford 
and ourselves were running Condor at a Tier-X center, with RAL joining the 
club about one month ago), there are some manual configuration steps 
involved, depending on how Condor is configured at your place and beyond 
what is captured in existing distributed YAIM scripts.

If it can be of any help, we (the people in the blah@xxxxxxxxxx list) can 
share our current practice, and set of updated YAIM scripts. Just let us 
know.

Just for a quick reference, condor-specific attributes in 
/etc/blah.config for a recent Condor installation from RPM look as 
follows. A Cream 'queue' will then be either a schedd name or a schedd 
name/pool name pair:

supported_lrms=condor
condor_binpath=/usr/bin
condor_nochecksubmission=yes
condor_nologaccess=yes
condor_fallback=no
condor_BLParser=yes

Just let us know exactly what you are trying to accomplish (e.g. whether 
you are installing via YAIM, the EMI version of the installation you'd 
like to move to Condor, the setup you are planning for your Condor pool or 
pools, and the like) and we can probably set you off in a promising 
direction.

Francesco Prelz
INFN Milan