I ran the pegasus-statistics on my periodograms (very simple jobs using Condor-I/O for staging) of ~1600 clustered jobs (~33k tasks total).

[1] I don't see the number of tasks in the workflow.
[2] When taking a long time, it should have some form of ("I am alive") output, either dots or a rotating bar or something.
[3] The breakdown.txt file created was empty, but the job statistics file had all interesting things.
[4] The computation of the Q length appears to be wrong.
[5] The stage-out jobs don't use kickstart, but the output in jobs claims kickstart time. We think it should appear in Condor time.

I will attach the submit directory as compressed tarball.

As a feature request, it would be useful in my case to have an extra statistics output file that shows hosts and jobs on hosts. In the attached case, the site is always "futuregrid" for compute jobs, but there are actually up to 3 different sites involved. Having a file ordered by the hostaddr from the kickstart record might help me to come up with more statistics, e.g. I can later cluster them by some netmask that I know.