On Sat, 17 Jul 2004, Steve Traylen wrote:
> Number 2 here is not strictly true, since the mkgridmap file has the
> sgm VO before the general VO the user is mapped to the sgm account
> overriding membership of other VOs, the first entry is the one that
> counts. Of course it is possible for the LHC VOs software managers
> were removed from the general VO? For babar I know this not to be the
> case and it works.
In LCG-2 there is no "sgm VO" concept: an "sgm" person is just a member
of the normal VO, but happens to be mapped to a static account instead
of a pool account. I suppose we could introduce sgm VOs, but probably
the effort is better spent getting VOMS to work.
> Of course this is all academic and the reality is you must support
> sgm accounts to attract any jobs as previously mentioned.
It depends on whether the job needs the experiment SW pre-installed;
a large variety of jobs may provide themselves everything they need.
|