At the risk of speaking for John, I believe he's referring to the
description that Graeme gave in the last dteam meeting (this Tuesday).
The minutes summarise the comments as:
"
Tomorrow UK hammercloud test – more extreme than normal though. Load
sites with as much analysis tests as possible. Start at 10am Wednesday
22 April. Initially run through WMS, but towards end of week
hammertest through Panda. Panda test does not need Role=Pilot to be
set up. Pilot jobs will have Peter's DN.
User analysis jobs not as efficient as production jobs. Is Maui fair
share based on Wall time or cputime?
For STEP want to load sites for 2 weeks.
"
He also started an email thread in this very mailing list, entitled
"ATLAS STEP09 Site Request(s) + Hammercloud Tomorrow", which includes
some discussion, including an explanation of why the initial date for
the HammerCloud (the 22nd) didn't happen (bugs in the submission
code).
That information is actually about as much as I know about why the
HammerCloud broke (and I don't know why the new attempt broke, which
it seems to have, based on the monitoring page).
(Google doesn't seem to spider the mail archive for TB-SUPPORT, which
makes searching it less useful.)
BTW, the HammerCloud tests started near the end of 2008 - hence the
vague aura of panic in some circles about how much load User Analysis
puts on storage over LAN.
Sam
2009/4/24 Daniela Bauer <[log in to unmask]>:
> Where was that sent to ? When I search my inbox for hammer all I get is
>
> "We are also trying to ramp-up the analysis tests so, to this end, we
> plan to run a more intense version of hammer cloud tomorrow in the UK.
> This will send up to 1200 jobs per site, which should keep the site
> busy for most of the day running many analysis jobs."
>
> What I was looking for was something like the email from Sam that tells
> somebody who is new (or maybe not even a HEP person), what this is all about
> - documentation and all that.
>
> From my understanding these tests have been running for a while (as in
> months if not years), so there is no reason they can't have an introduction
> somewhere on a webpage.
>
> Daniela
>
>
> 2009/4/24 Gordon, JC (John) <[log in to unmask]>
>>
>> The original mail from Graeme announcing this round was quite detailed.
>> John
>>
>> On 24 Apr 2009, at 15:50, "Daniela Bauer"
>> <[log in to unmask]> wrote:
>>
>> Sigh, that last message should not have gone to the mailing list, but the
>> point I was trying to make was:
>>
>> I have never seen so far any (written) documentation what a HammerCloud
>> test is (i.e we are submitting MC production/analysis/whatever jobs in a
>> mixture we think is sensible ?)/what's it trying to achieve etc etc. Somehow
>> it's always assumed that this information reaches the lower ranks somehow,
>> but it just doesn't.
>>
>> I need some sugar.
>>
>> Daniela
>>
>> 2009/4/24 Daniela Bauer <[log in to unmask]>
>>>
>>> Sometimes I think Scots don't understand what life on the grid without a
>>> direct channel to Graeme looks like....
>>>
>>> 2009/4/24 Douglas McNab <[log in to unmask]>
>>>>
>>>> Hi,
>>>>
>>>> The Ganga Robot Page is here: http://gangarobot.cern.ch/st/
>>>> There are a few pages where you can see submit logs etc etc.
>>>>
>>>> Cheers,
>>>>
>>>> Dug
>>>>
>>>> 2009/4/24 Daniela Bauer <[log in to unmask]>
>>>>>
>>>>> Is there somewhere where we can see what these tests are actually
>>>>> meant to be doing ? Googling it didn't come up with much.
>>>>>
>>>>> Daniela
>>>>>
>>>>>
>>>>> 2009/4/24 Sam Skipsey <[log in to unmask]>:
>>>>> > Are you sure this isn't just the Extreme HammerCloud stressing your
>>>>> > SE? I notice that it (the HammerCloud) seems to be in a bit of an odd
>>>>> > state now, but it definitely submitted lots of jobs to everyone - we
>>>>> > got the normal high load at Glasgow when it passed through.
>>>>> >
>>>>> > Sam
>>>>> >
>>>>> > 2009/4/24 Christopher J.Walker <[log in to unmask]>:
>>>>> >> We have a high load (225) on our SE machine, and there are a large
>>>>> >> number of
>>>>> >> rfiod in the "D" state. We were also failing SAM tests earlier
>>>>> >> today -
>>>>> >> possibly because of this.
>>>>> >>
>>>>> >> Is this just that our SE is underpowered, or is it indicative of
>>>>> >> something
>>>>> >> having got stuck - would restarting dpm help for example?
>>>>> >>
>>>>> >> Chris
>>>>> >>
>>>>> >
>>>>>
>>>>>
>>>>>
>>>>> --
>>>>> -----------------------------------------------------------
>>>>> HEP Group
>>>>> Physics Dep
>>>>> Imperial College
>>>>> Tel: +44-(0)20-75947810
>>>>> http://www.hep.ph.ic.ac.uk/~dbauer/
>>>>
>>>>
>>>>
>>>> --
>>>> ScotGrid, Room 481, Kelvin Building, University of Glasgow
>>>> tel: +44(0)141 330 6439
>>>
>>>
>>>
>>> --
>>> -----------------------------------------------------------
>>> HEP Group
>>> Physics Dep
>>> Imperial College
>>> Tel: +44-(0)20-75947810
>>> http://www.hep.ph.ic.ac.uk/~dbauer/
>>>
>>
>>
>>
>> --
>> -----------------------------------------------------------
>> HEP Group
>> Physics Dep
>> Imperial College
>> Tel: +44-(0)20-75947810
>> http://www.hep.ph.ic.ac.uk/~dbauer/
>>
>>
>> ________________________________
>> Scanned by iCritical.
>
>
>
> --
> -----------------------------------------------------------
> HEP Group
> Physics Dep
> Imperial College
> Tel: +44-(0)20-75947810
> http://www.hep.ph.ic.ac.uk/~dbauer/
>
>
|