On 30/07/2012 13:48, Rob Fay wrote:
> On 30/07/2012 11:49, Alessandra Forti wrote:
>> SW_DIR, DEFAULT_SE, STORAGE_DIR are site specific and therefore
>> should not be
>> decided by the VO.
>>
> Is that entirely true for SW_DIR? The VO does have at least some input
> into it, specifically whether they're using cvmfs or not. And am I
> right in thinking they specify more than that as well? E.g. ATLAS's
> cvmfs SW_DIR path is set to '/cvmfs/atlas.cern.ch/repo/sw' whereas
> LHCb's is set to '/cvmfs/lhcb.cern.ch' - that's not site specific is it?
>
there are sites that cannot use cvmfs yet not even for the LHC VOs
because they don't have enough space on the WN. Also it is true that
atlas must use the same path everywhere but lhcb doesn't so for them you
could use a different mount point as IC-HEP did at the beginning. And
smaller VOs aren't using CVMFS, so even if you can argue that CVMFS has
predetermined paths for the LHC VOs due to CVMFS and for simplicity that
is not true for anybody else.
--
Facts aren't facts if they come from the wrong people. (Paul Krugman)
|