Re: [lead-all] WxChallenge Testing

NOTE: The leadusers mailing list is no longer active. The list archives are made available for historical reasons.

Hi Dennis,

Actually the success rate I mentioned is only from March 2nd and we need all the testers in high gear to bring it down to 95% (I wish not too lower than this).

Hi Testers,

I just want to emphasize again that the percentage numbers I am throwing around are from rehearsals and to encourage you guys but you have to help get the actual numbers. Please follow Tom's documentation to do a end to end testing. Also the 42 hour 5Km WRF output is around 5 Giga bytes and after you see workflow complete in Xbaya or finished status in experiment builder you will have to wait a while before your data appears in your workspace as the storage system has to move the file from compute clusters scratch location to a permanent storage location. But if you don't see the files appearing in your workspace within an hour (or at most 2 hours) then report it.

And please don't launch workflows just for the sake of launching, they are running on teragrid resources stopping other science simulations and eating away LEAD allocated usage (or indirect money). So if you run a workflow, please verify the the forecast and report the success/failure and other feed back using feed back form on portal or to Rich, Everett & Tom in however way you are asked to do so.

Thanks for all your time and help with testing,
Suresh

Dennis Gannon wrote:
LEAD - Master Mailing List

hi suresh,
i am in my hotel room in redmond and things are indeed looking
better.  you and the rest of the extended gang have done a great job!
(but i would not say we are at 100% yet.... unless i measure
from march 1.  ... if i measure from feb 24, i have done 16 runs
and 4 failed and one just vanished.  but so far with the 2.2 version
it is all working.  and do we have big red in the picture yet?
(also, i am talking with pcs about how we can use their machine.
we will need to run in the special protected shell and they want
us to us 550 cores! we can work on that after we get the ibm folks
to finish helping us with bigred.

in general 100% is not achievable.  but if we can get 95% over
a long period to complete in reasonable times, i will be very happy.
thanks again!
dennis

On Sun, 4 Mar 2007, Suresh Marru wrote:

LEAD - Master Mailing List

Lead Testers,

I am happy to say that after the portal system is back from maintenance
on Friday all the infrastructure and WRF looks very stable and 100% of
the workflows ended successfully. Also since the new wrf 2.2 workflows
are working good and since they need more testing, I have removed the
old wrf 2.1 workflows and other development ones from sample workflow
repository. So test more of the weather challenge scenario (5km regional
configuration) with both NAM and ADAS workflows which will run latest
ARPS and WRF.

Thanks to NCSA & IU resource providers, starting tomorrow we do have
reservation of compute resources for 5 hours each day until end of
wxchallenge from Noon CST - ie from 18Z to 23Z. So try to launch your
workflows before 18Z to finish them on time. After you see the WRF
component turn green and start to see *Active* messages in final
notifications in xbaya, the estimated time of completion for a 36 hour
5km forecast is 3 hours 20 mins.

Note: We are still working on making the nested configurations and post
processing services available and also looking at refining WRF timings.

LEAD Met. Group Testers,

Please look at your workspace for input files and wrf output in IDV and
verify if the forecasts and configurations are valid.

HU & MU students,

Prof. Gannon from IU has been the leading the race of frequent testing
and finding problems immediatly after they occur (50% of the workflows
are run by him). I urge all of you to challenge him and do more testing
and  report problems. (Sorry for more emails Tom). The winner gets a
free bonus additional hour of forecast. So when you run a 36 hour
forecast and if you see 37 hour results then you win, shoot for it guys
:-) .

Thanks,
Suresh


==============================================================================
To unsubscribe leadusers, visit:
http://www.unidata.ucar.edu/mailing-list-delete-form.html
==============================================================================


  • 2007 messages navigation, sorted by:
    1. Thread
    2. Subject
    3. Author
    4. Date
    5. ↑ Table Of Contents
  • Search the leadusers archives: