Today's unpleasantry: upgrade tests failed to run. This makes
failure to run to completion on 4 of the last 10 runs. We
should probably look into stabilizing this test environment and/or
the tests if tests are causing the full job to fail as a failed run
is essentially the worst result we can get h
The nose.failure.Failure.runTest was a missing import in a dtest that was
quickly fixed. We'd ideally have a way to stage dtest changes to catch
these sort of things, but there seems to be a reasonable consensus that
this is a lower priority issue than environmental stability and other test
fixes.
