 |
 |
Yo, These things are very useful. In particular they show how difficult it is to compare numbers coming from one group of people. Just compare the information today for ATLAS and LHCb ... The total success rate of ATLAS was 63% and 76% of the jobs that succeeded, succeeded only after resubmission. In the same day, LHCb reported 91% success, and none of the successful jobs required resubmission. Closer inspection shows that the data for ATLAS and LHCb jobs going thru the brokers gdrb01 and gdrb02 at CERN are essentially identical; the bad ATLAS results are due to heavy use of gdrb09 and gdrb10, which both had extremely poor reliability. J "statistics indeed" T Dr D J Colling wrote: > Sear All, > > > In order to display jobs moving around the Grid we have needed to gather > information about the times that each jobs spends in different states and > at different sites. We decided to harvest a subset of this data and > analyse it in an attempt to understand usage patterns and possible > inefficiencies/problems at each site. We have started to produce daily > summaries of these statistics which we hope are useful. These summaries > will be archived. > > > In the future we plan to make the data that we harvest available to > others who may wish to perform analyses (either as ASCII text or root > trees). If you would like the data now then please contact us. > > > You can find these reports at: > http://gridportal.hep.ph.ic.ac.uk/rtm/reports.html > > They are also linked from our main monitoring page: > http://gridportal.hep.ph.ic.ac.uk/rtm/ > > All feedback would be welcome. > > All the best, > david