[Swift-devel] The effectiveness of new changes in scheduler

[lixi at uchicago.edu]

Hi,

I launched a workflow (including 4001 jobs, the sequential 
execution time should be more than 4001*60s) yesterday. It 
finished successfully, although there is a long waiting time 
(no job submission). The log plot is:
http://www.ci.uchicago.edu/~lixi/Log/report-workflowtest-
20080702-1751-nt031l0f/

I think that, this result could prove the new changes in 
scheduler could prevent certain bad site absorbing all 
retries and improve the success rate of whole workflow 
execution, but potential might increase the workflow 
execution time. Some parameters might be tuned to get the 
tradeoff between success rate and execution time. I hope 
that this feedback could help.

Xi