But RAL also has a variety of other VOs using the T1. It is only the LHC
VOs who have officially asked for SL5 64 bit.
> -----Original Message-----
> From: Testbed Support for GridPP member institutes [mailto:TB-
> [log in to unmask]] On Behalf Of Bly, MJ (Martin)
> Sent: 27 April 2009 17:05
> To: [log in to unmask]
> Subject: Re: Can all sites run SL5 64-bit WNs?
>
> We have lots of old stuff that is 32bit only. We plan at the moment
to
> leave them at SL4/32bit, and retire them in O(12) months. For
> operational convenience, Sl5/32bit would be nice but is not essential.
>
> Martin.
> --
> Martin Bly
> RAL Tier1 Fabric Team
>
> > -----Original Message-----
> > From: Testbed Support for GridPP member institutes
> > [mailto:[log in to unmask]] On Behalf Of Coles, J (Jeremy)
> > Sent: 27 April 2009 15:57
> > To: [log in to unmask]
> > Subject: Can all sites run SL5 64-bit WNs?
> >
> > Dear All
> >
> > It has become apparent that the matter concerning future gLite WN
> > releases was not clear. Specifically, at the moment there is
> > NO plan to
> > provide a gLite 3.2 WN release in a 32-bit version. The gLite 3.2 WN
> > package will ONLY be available on SL5 in a 64-bit version.
> >
> > I am now aware that may cause significant problems for one
> > GridPP site.
> > Are there any other sites running older kit that would not be able
to
> > work with a 64-bit release?
> >
> > Regards,
> > Jeremy
> >
> > P.S At the current time there are no plans to stop supporting
> > gLite 3.1
> > WNs on SL4. Given previous migration periods I believe we can
> > expect the
> > 32-bit SL4 release to be available and used for at least the
> > next year,
> > but much will depend on how quickly the experiments move to SL5 and
> > 64-bit. The 64-bit option has been a major problem for ATLAS due to
> an
> > almost doubling in the memory requirement, but I believe in the last
> > week the underlying cause of this has been understood and this
> barrier
> > will soon be removed. Anyway, the SL5 3.2 release comes with
> > the 32-bit
> > libraries. So perhaps the driving issue is the timescale upon
> > which the
> > experiments will migrate their applications to SL5 only. I am
> > trying to
> > find out.
> >
> >
> >
> > --
> > Scanned by iCritical.
> >
> --
> Scanned by iCritical.
--
Scanned by iCritical.
|