Comment 3 for bug 899972

I'm waiting for qastaging to update before I can actually tag this bug, but I've QAed it on dogfood and everything seems fine.

cron.germinate runtime before (r14515):

  real 7m39.023s
  user 2m29.057s
  sys 0m40.159s

cron.germinate runtime after (r14517):

  real 3m13.435s
  user 1m24.801s
  sys 0m9.617s

(A good amount of the remaining time is maintenance-check.py, which I have plans to rewrite in the future but which doesn't need to be part of this bug.)

I had some initial problems because dogfood's database has an ambiguity about what the Ubuntu development series might be: all of rusty, precise, and oneiric are in status FROZEN. This caused generate-extra-overrides.py to get confused as apparently its find_operable_series produces different results from 'lp-query-distro.py development'. I changed the status of rusty and precise to FUTURE temporarily in order to do this test, and then everything went through cleanly. This will not be a problem on production because there will only ever be one Ubuntu series there in either DEVELOPMENT or FROZEN status. I've filed bug 904538 for this, though, as we should clean it up at some point.

I compared the old and new ubuntu-misc and ubuntu-germinate trees with these commands:

  diff -u <(<ubuntu-misc.r14515/more-extra.override.oneiric.main sed 's/Task */Task /' | sort) <(<ubuntu-misc/more-extra.override.oneiric.main sort)
  # ... no output
  diff -bru ubuntu-germinate.r14515 ubuntu-germinate
  # ... considerable output, but it's all either in _maintenance-check.*.stderr (don't care) or expected changes to the germinate.output log file due to the new and more efficient ordering of operations

Thus, all is well. qa-ok.