> -----Original Message-----
> From: Christopher J. Walker [mailto:[log in to unmask]]
> Sent: 10 January 2014 09:14
>
> On 09/01/14 14:28, Ewan MacMahon wrote:
> >> -----Original Message-----
> >> From: Testbed Support for GridPP member institutes [mailto:TB-
> >> [log in to unmask]] On Behalf Of Jeremy Coles
> >>
> >> The latest WLCG position on this was given by Maarten at the WLCG ops
> >> meeting on Monday
> >>
> (https://twiki.cern.ch/twiki/bin/view/LCG/WLCGDailyMeetingsWeek140106#Mond
> ay):
> >>
> >
> > What's not in Maarten's summary though, is a recommendation of what to
> > actually do about it. My inclination is that everyone should upgrade,
> > and any intermediate services (like WMSes) that are then broken
> > because they've not upgraded, should just be fixed.
>
> I've done some tests on UK WMSs/CEs using dteam based on Maarten's scripts
> at:
>
> https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind1401&L=LCG-
> ROLLOUT&F=&S=&P=7211
>
> Executive summary: at the moment, only Imperial's WMSs are OK.
>
OK, so how does this sound as a plan:
- Get the remaining UK WMSes upgraded,
- Then upgrade all the UK service nodes
At that point the only breakage will be for people who are using
UK resources, via non UK WMSes, and only if those WMSes also haven't
been updated. If I'm following this correctly, there's no downside
to upgrading a WMS since they'll still work with non-upgraded CEs,
the problem is only if you've got upgraded endpoints and out-of-date
intermediate services.
Ewan
|