On Thu, 12 May 2005, Burke, S (Stephen) wrote:

> Testbed Support for GridPP member institutes
> > [mailto:[log in to unmask]] On Behalf Of Henry Nebrensky said:
> >         - last time I looked, I couldn't find sufficiently idiot-proof
> >           instructions on how to add a specific VO to a site
> >           (not for my level of idiocy!)
>
> It's claimed to be pretty trivial with YAIM.

The "how to include the VO config file" bit was fairly easy when I looked,
- it was the "what to put IN said file" that lost me. Mind you, my brain's
full of R-GMA...

> >         - there isn't a good way to find what possible VOs exist:
>
> See the end for the list of the 43 VOs with replica catalogues visible
> from the RAL BDII. You can also look at the CIC portal to see the ones
> which are known to EGEE:
>
> https://cic.in2p3.fr/index.php?id=vo&js_status=2

I might be using this wrong, but it doesn't really help in finding likely
candidates. For example:

> ldapsearch -x -h lcgbdii02.gridpp.rl.ac.uk -p 2170 -b o=grid | grep
> GlueServiceURI | grep -i replica-cata | cut -d/ -f4 | sort | uniq
...
> baikal
...

If that was using the Grid for numerical reconstruction from digital
holograms as part of a study of aquatic microorganisms in Lake Baikal, I'd
actually be really interested in helping out.

But all the CIC entry gives me is a list of its infrastructure - I can't
even see a VO contact, let alone what the VO does or what sort of
resources they might be interested in - they might only use some sort of
custom hardware accelerator, for example.

...
> >           I'm not even sure offhand what "gridpp VOs" exist apart from
> >           babar.
Er, and mice.


> Probably at least:
...
> herab
> hermes

I've never even heard of these (as VOs)!

...
> cdf
> There should also be dzero, maybe they don't have a catalogue.

Yes, but cdf/dzero are into this SAMgrid thing - i.e. would any "LCG"
based resources I could provide actually be useful(=used by) them?



> >         - Out of possible VOs, I wouldn't pretend to know
> > whether we're
> >           allowed to support ones whose members agreed to
> > conditions of
> >           use other than the LCG ones.
>
> If you check the auth server when you make the grid map file you would
> only get people who have signed the LCG AUP - not that that actually
> means a lot.

True... but I'd really like to know that no-one in a VO can actually use
my resources *before* I go through the rigmarole of installing it!


> > Incidentally, I think it was you that wrote in another post
> > (which I've
> > deleted) something like "It shouldn't be up to Atlas to chase sites to
> > support them, (especially if they don't really need them)",
>
> I think I said the opposite, that the sites shouldn't be chasing VOs to
> install software and run jobs.

Oh. Well, someone said it, although their primary context was the bit I've
parenthesized...

> > In that case, I would say that - in general - it IS up to VOs to go to
> > the sites to ask for resources. We're just fortunate that LCG
> > is acting as a matchmaker for certain VOs and sites.
>
> Well, the VOs go to EGEE as a whole and ask for resources. That
> effectively means the ROC managers. As I understand it the UKI ROC has
> agreed that GridPP will provide a nominal 0.5% allocation to "other" VOs
> approved in EGEE, but I don't think anyone has really either defined
> what that means or done anything to make it happen.

*Some* VOs go to EGEE... I'm sure there are non-LCG, non-EGEE VOs out
there (e.g. in the States) that might end up using UK resources as part of
some collaboration. Of course, the setting up of this would qualify as the
VO going to the site to ask for resources.

But EGEE doesn't seem to be playing the 'matchmaker' role - that would
need exactly things like matching AUPs.


Overall, my point is that sites already publish details about their
capabilities into a single information system, but VOs don't publicise
their needs; so the VOs going to the sites would seem the more efficient
direction, whether the VOs go directly or via the ROCs.


Henry

--
Dr. Henry Nebrensky                     [log in to unmask]
                             http://people.brunel.ac.uk/~eesrjjn
"The opossum is a very sophisticated animal.
 It doesn't even get up until 5 or 6 p.m."