In jobs_horizontal_history, the definition
of remotewallclocktime in Huang et. al. (2007) quotes "cumulative
number of seconds the job has been allocated to a machine".
Is that the number of seconds for the
last run or for all runs e.g. does that number include "badput"
or not ? I ask because I want to calculate goodput vs badput from jobs_horizontal_history
if that's possible in order to come up with a measure of pool efficiency
over a running month, which leads me on to my second question.
Is it possible to calculate goodput
vs badput from the data in quill and if so, what would the formula look
like and which fields would I need to extract from which tables? I
want to use effectively the same calculation as used by "condor_q
-global -goodput" (which reports in real-time) to, in essence, give
me goodput per globaljobid after massaging the data output from a psql
query to our quill database.
Any light you could shed on either problem
would be appreciated.
Best regards
James
# Huang et. al. (2007) An Overview of
Quill: A Passive Operational Data Logging System for Condor
__________________________________________________________________________________
Dr James Osborne, Application Support Engineer, Advanced Research Computing
Division
Cardiff University, Redwood Building,
King Edward VII Avenue, Cardiff, CF10 3NB
Tel: +44(0)2920 874657, Fax: +44(0)2920
870734
__________________________________________________________________________________