Dear All
It seems that Stuart's mail was distributed wider than intended - i.e.
outside the IC group - but it has usefully highlighted that we could
benefit from feedback. The points raised off list will be considered in
follow up work if not addressed in SSC3. If anyone else wants to make
useful suggestions then please share them and at least let us know the
outcome of any internal discussions.
Anyway, we do not wish to artificially engineer an environment for the
challenge. If most sites support a 48hr then we should have one for the
gridpp VO and use it for the challenge. It sounds to me that the request
should be for a 48hr (rather than 72hr) queue since this matches queues
available for other VOs (thanks for the comment Chris on how things are
affected with faster CPU and Graeme indicating that Glasgow is moving to
48hr to speed up intervention times).
Jeremy
-----Original Message-----
From: Coles, J (Jeremy)
Sent: 20 October 2008 15:04
To: 'Testbed Support for GridPP member institutes'
Subject: RE: Enabling the gridpp VO (and providing a 72hr queue)
Hi Stuart
That's not a helpful thread to start. Please bear in mind when
considering this topic that we are running the challenges to IMPROVE
security for everyone. The challenges themselves have been thought
through but I'm sure afterwards there will be experiences that mean we
can improve them. Starting out with a negative attitude will not improve
the security. If you have (sensible and reasonable) suggestions for
making the sites and grid more secure then please share them but don't
undermine the efforts that are being made.
Thanks,
Jeremy
-----Original Message-----
From: Testbed Support for GridPP member institutes
[mailto:[log in to unmask]] On Behalf Of Stuart Wakefield
Sent: 20 October 2008 14:41
To: [log in to unmask]
Subject: Re: Enabling the gridpp VO (and providing a 72hr queue)
Who else thinks this will be a complete waste of time?
---------- Forwarded message ----------
From: Graeme Stewart <[log in to unmask]>
Date: Mon, Oct 20, 2008 at 2:34 PM
Subject: Re: Enabling the gridpp VO (and providing a 72hr queue)
To: [log in to unmask]
Well, obviously I now have an inkling that I will need to intercept a
running job; but, as I said, Glasgow are going to move to a 48 hour
maximum wallclock for our new kit and we're not going to change this.
Cheers
Graeme
On Mon, Oct 20, 2008 at 3:20 PM, Ma, M (Mingchao)
<[log in to unmask]> wrote:
> Hi Graeme,
>
> I understood what you had said. But I can't tell you why. You will
> understand it once you complete the challenge :-)
>
> Cheers,
>
> Mingchao
>
>
>> -----Original Message-----
>> From: Testbed Support for GridPP member institutes [mailto:TB-
>> [log in to unmask]] On Behalf Of Graeme Stewart
>> Sent: 20 October 2008 14:15
>> To: [log in to unmask]
>> Subject: Re: Enabling the gridpp VO (and providing a 72hr queue)
>>
>> This makes no sense to me. What does having a certain time to respond
>> to the security challenge have to do with the wallclock time on the
>> site's queues?
>>
>> If I can compromise your site it will probably happen <5s after my
job
>> starts to run, even if I run it in a 7 day queue.
>>
>> Graeme
>>
>> On Mon, Oct 20, 2008 at 3:11 PM, Ma, M (Mingchao)
>> <[log in to unmask]> wrote:
>> > Hi All,
>> >
>> > What I can say at this stage is 72 hours mean sites have 72 hours
to
>> > complete the security challenge. And if all sites agree that 48
hours
>> are
>> > enough then I can go with it. But I have to say 24 hours are too
>> short since
>> > sites have only about 8 working hours to response the challenge
>> unless you
>> > want to work around the clock.
>> >
>> > As Jeremy said in his previous email: "But, we want the reference
>> point the
>> > same across sites to allow responsiveness (and actions taken)
across
>> sites
>> > to be measured in a consistent way." Technically I can go with any
>> hours as
>> > long as all sites agree upon it. But I have to say that we are
better
>> to go
>> > with 72, 48 is also ok, but 24 is too short.
>> >
>> > Cheers,
>> >
>> > Mingchao
>> >
>> >> -----Original Message-----
>> >> From: Testbed Support for GridPP member institutes [mailto:TB-
>> >> [log in to unmask]] On Behalf Of Brew, CAJ (Chris)
>> >> Sent: 20 October 2008 12:39
>> >> To: [log in to unmask]
>> >> Subject: Re: Enabling the gridpp VO (and providing a 72hr queue)
>> >>
>> >> And it doesn't have much meaning for sites that scale but CPU
power.
>> >>
>> >> I have 72hr walltime queues but if you end up on a fast CPU that
can
>> >> actually be less than 36hrs real time.
>> >>
>> >> (I know this strikes me as odd as well but it's the way the T1
does
>> it
>> >> and it does sort of make sense.)
>> >>
>> >> Yours,
>> >> Chris.
>> >>
>> >> > -----Original Message-----
>> >> > From: Testbed Support for GridPP member institutes
>> >> > [mailto:[log in to unmask]] On Behalf Of Alessandra Forti
>> >> > Sent: 20 October 2008 12:16
>> >> > To: [log in to unmask]
>> >> > Subject: Re: Enabling the gridpp VO (and providing a 72hr queue)
>> >> >
>> >> > Hi Jeremy,
>> >> >
>> >> > I don't understand why 72hours (apaprt from the fact that it has
>> been
>> >> > the default wall time in yaim for the past 2 years?
>> >> >
>> >> > cheers
>> >> > alessandra
>> >> >
>> >> > Coles, J (Jeremy) wrote:
>> >> > > Dear All
>> >> > >
>> >> > > Thank you to all sites that have now enabled the gridpp VO. A
>> >> number
>> >> > > have still not responded to the request made several times
>> >> > over the last
>> >> > > 4-6 months. If your site does not have the VO enabled
>> >> > please could you
>> >> > > let me know if you (do not) intend to enable it?
>> >> > >
>> >> > > For those sites that have enabled it, please could I ask
>> >> > you to check
>> >> > > the available queues? I have been asked if we can provide a
>> >> > 72hr queue
>> >> > > for use by some jobs submitted under the VO and the current
>> >> > situation
>> >> > > shows that this is only available at half the supporting
sites:
>> >> > >
>> >> > >
>> >> > > GridPP VO 72 hours queue
>> >> > > ScotGrid
>> >> > > UKI-SCOTGRID-DURHAM Yes Yes
>> >> > > UKI-SCOTGRID-ECDF Yes No
>> >> > > UKI-SCOTGRID-GLASGOW Yes Yes
>> >> > >
>> >> > > NorthGrid
>> >> > > UKI-NORTHGRID-LANCS-HEP Yes No
>> >> > > UKI-NORTHGRID-LIV-HEP Yes Yes
>> >> > > UKI-NORTHGRID-MAN-HEP Yes Yes
>> >> > > UKI-NORTHGRID-SHEF-HEP Yes Yes
>> >> > >
>> >> > > SouthGrid
>> >> > > EDFA-JET Yes No
>> >> > > UKI-SOUTHGRID-BHAM-HEP Yes Yes
>> >> > > UKI-SOUTHGRID-BRIS-HEP Yes Yes
>> >> > > UKI-SOUTHGRID-CAM-HEP No No
>> >> > > UKI-SOUTHGRID-OX-HEP Yes Yes
>> >> > > UKI-SOUTHGRID-RALPPD Yes Yes
>> >> > >
>> >> > > LondonGrid
>> >> > > UKI-LT2-Brunel Yes No
>> >> > > UKI-LT2-IC-HEP No No
>> >> > > UKI-LT2-IC-LeSC Yes Yes
>> >> > > UKI-LT2-QMUL No No
>> >> > > UKI-LT2-RHUL Yes No
>> >> > > UKI-LT2-UCL-CENTRAL No No
>> >> > > UKI-LT2-UCL-HEP No No
>> >> > >
>> >> > >
>> >> > > Many thanks for your help,
>> >> > > Jeremy
>> >> > >
>> >> >
>> >> > --
>> >> > Alessandra Forti - NorthGrid Technical Coordinator
>> >> > http://www.hep.manchester.ac.uk/computing/tier2
>> >> >
>> >> > Well you'll still need a tray
>> >> >
>> >> --
>> >> Scanned by iCritical for STFC.
>> >
>>
>>
>>
>> --
>> Dr Graeme Stewart http://www.physics.gla.ac.uk/~graeme/
>> Department of Physics and Astronomy, University of Glasgow, Scotland
>
--
Dr Graeme Stewart http://www.physics.gla.ac.uk/~graeme/
Department of Physics and Astronomy, University of Glasgow, Scotland
--
Scanned by iCritical for STFC.
|