Hi Ewan, I am looking at the long term settings of the FTS channel
shares for ATLAS. If we opened up all Tier2 sites to there full
capacities then we would swamp the services on CASTOR. ( 20 sites each
sending 20 concurrent files at 100MB/s into CASTOR would be ALOT of
resources!!!) Hence trying to establish some metric using
recent-current and nominal usage of a site in this calculation.
Knowing a the fair share and # of CPUs roughly gives the nominal rate
from a CE that needs to uploaded for Tier2 sites. I am making the
assumption that when sites are fully up and running, then the fair
shares will eventually reflect usage of a CE. Obviously the SE
capacity also is needed in consideration for this. Also, since
different VOs have different Computing models, flows of data vary and
so their setup of FTS will also vary. This at the moment is mostly all
in developement ( and may not not even be implemented this way) so if
anyone has any ideas, let me know.
Brian
On 05/11/2007, Ewan MacMahon <[log in to unmask]> wrote:
> > Burke, S (Stephen) wrote:
> > > Testbed Support for GridPP member institutes
> > >> [mailto:[log in to unmask]] On Behalf Of Alessandra
> > Forti said:
> > >> if they are not listed anywhere I don't think you can.
> > >
> > > Can you not do diagnose -f with globus-job-run, or is it privileged?
> > >
>
> > It's privileged.
>
> It looks, however, as though you can do:
> globus-job-run $TARGETCE /bin/cat /var/spool/maui/maui.cfg
> at least on a default yaim generated setup. It obviously won't work
> correctly
> on sites that don't run maui, or run it on a machine other than the CE.
>
> I'm curious to know why you'd want to know though?
>
> Ewan
>
|