On 30 July 2012 14:10, Stephen Jones <[log in to unmask]> wrote:
> On 07/30/2012 01:53 PM, Sam Skipsey wrote:
>
>> Right, so for VOs using CVMFS, SW_DIR is entirely VO specific, assuming
>> that you mount the CVMFS repositories in the recommended /cvmfs location.
>> For other VOs (in fact, the vast majority of VOs), there's no CVMFS
>> repository, and hence the SW_DIR is up to the site.
>
>
> Hi,
>
> What I'm reading from this is:
>
> a) for VOs using CVMFS, SW_DIR is VO specific.
>
> That invites another question. How does the VO that uses CVMFS convey
> the SW_DIR path? Should it be a tag in the CIC portal? Or some workaround?
> We need to talk about that at some point.
>
This is something which will become an issue soonish (it's not a
problem at the moment purely because the VOs using CVMFS are Big and
LHCish and therefore can afford to tell people to DO THIS via their
reps). I suspect it should be a CIC Portal tag.
> b) For other VOs the SW_DIR is up to the site, but the convention
> (recommendation?) is to use some site-specific mount offset
> (e.g. VO_SW_DIR=/opt/exp_soft_sl5) with a uniq vo specific nickname (e.g.
> for
> vo.sixt.cern.ch it might resolve to SW_DIR=$VO_SW_DIR/sixt).
It's a convention and a very mild recommendation, mostly because it
makes it easier for the site to corral all those software areas
together somewhere sensible (pragmatically, usually on some NFS
server, for non-Lustre/GPFS sites).
Sam
>
>
> Steve
>
>
>
>
> --
> Steve Jones [log in to unmask]
> System Administrator office: 220
> High Energy Physics Division tel (int): 42334
> Oliver Lodge Laboratory tel (ext): +44 (0)151 794 2334
>
> University of Liverpool http://www.liv.ac.uk/physics/hep/
|