Hi Graeme
And there was me trying to be optimistic! (Is that a chuckle from Dr.
Colling?). Anyway, no parades here so rain away.
You are right that there are other more relevant views within WLCG taken
from the experiment dashboards. I was using the EGEE result. For the
GridPP project map we are taking all the figures into consideration as
you know. Site downtime will feed into all the availability results and
overall improvements here should be noted. Now I know there are some
sites where things have not looked good this last month (or recently),
but I think in all cases people are working hard to improve the
situation.
Now, where was that baton (just in case)?
Cheers,
Jeremy
-----Original Message-----
From: Testbed Support for GridPP member institutes
[mailto:[log in to unmask]] On Behalf Of Graeme Stewart
Sent: 05 November 2008 14:37
To: [log in to unmask]
Subject: Re: October - EGEE site availability & reliability
Jeremy
I hate to rain on your parade, but RAL managed to have at least 3 days
of downtime last month for ATLAS (mainly CASTOR and mainly
unscheduled, e.g., https://goc.gridops.org/downtime/list?id=10455639)
so the figure of 95/99% availability should be taken with a pinch of
salt. 90% maximum and this doesn't count significant periods of
degradation we suffered (plus file loss from CASTOR).
Doing well in these tests is necessary, but far from sufficient.
Tier-2s were very good in October - so well done to all here.
Cheers
Graeme
P.S.
http://dashb-atlas-prodsys-test.cern.ch/dashboard/request.py/summary?sta
rt-date=2008-10-01%2000:00&end-date=2008-10-31%2023:30&cloud=RAL&groupin
g=site&end-date=2008-11-05%2023:59:59&start-date=2008-10-29%2000:00:00
is also of interest.
On Wed, Nov 5, 2008 at 2:59 PM, Coles, J (Jeremy)
<[log in to unmask]> wrote:
> Dear All
>
> It is great to see that UKI as a region leads the ranking for October
in terms of both EGEE site availability and site reliability. The
breakdown is available here:
https://edms.cern.ch/file/963325/1/egee_october2008.pdf. Thank you to
everyone involved for helping us to achieve these results. I am sure
that we can maintain this result and will improve upon them - indeed our
GridPP targets are 95% in both areas.
>
> Some next steps:
>
> 1) In order to understand where we can improve, I will shortly be
circulating a link to new history charts akin to those already used to
gather site comments on availability/reliability issues - for the last
set see
http://www.gridpp.ac.uk/wiki/SAM_availability:_October_2007_-_May_2008.
>
> 2) The autumn 2008 HEPSYSMAN meeting takes place tomorrow in
Manchester: http://hepwww.rl.ac.uk/sysman/nov2008/agenda.html. There
will be plenty of opportunity for discussion. If you are unable to get
to Manchester for this meeting, please consider joining by EVO (if it is
available - hopefully this will be confirmed shortly).
>
> 3) We have an opportunity next week at the WLCG planning workshop to
present UK Tier-2 views on the WLCG project so far and what we would
like to see change over the next 6-12 months. I have asked the Tier-2
coordinators to gather input from their sites for inclusion in the
overview, so please support them and me by coming forward with your
thoughts. If there are things you want addressed then this is a good
time to speak up! The agenda for the meeting is in Indico:
http://indico.cern.ch/conferenceOtherViews.py?view=standard&confId=32660
. Note that an EVO connection will be available if you wish to hear a
specific talk.
>
>
> Thanks again for your continued efforts and support,
> Jeremy
>
>
>
> --
> Scanned by iCritical.
>
--
Dr Graeme Stewart http://www.physics.gla.ac.uk/~graeme/
Department of Physics and Astronomy, University of Glasgow, Scotland
--
Scanned by iCritical.
|