FreeBSD/kqueue: DirtyReactorAggregateError

[Tobias Oberstein]

I have been trying to find out more about the last 2 remaining kqueue specific trial errors on the new FreeBSD buildslave (CPy 2.7) - please see here:

​http://buildbot.twistedmatrix.com/builders/freebsd-9.2-amd64-python2.7/builds/13

I can reproduce the errors when running the complete trial, but when I run only the failing test, I *cannot* reproduce. Please see the attached log.

I could need some tips/hints on where to look, what to try or such, since I have no clue whats going on. My suspicion is, that something is left-over from previous tests that make the subsequent fail. Dunno.

[Tobias Oberstein]

Probably related (or not), there is a new fail on all 3 reactors

twisted.conch.test.test_endpoints.NewConnectionHelperTests.test_readExisting

[FAIL]
Traceback (most recent call last):
  File "/home/bbslave_twisted/work/freebsd-9.2-amd64/Twisted/twisted/conch/test/test_endpoints.py", line 1411, in test_readExisting
    self.assertTrue(loaded.hasHostKey("127.0.0.1", key))
  File "/home/bbslave_twisted/work/freebsd-9.2-amd64/Twisted/twisted/trial/_synctest.py", line 308, in assertTrue
    raise self.failureException(msg)
twisted.trial.unittest.FailTest: None

twisted.conch.test.test_endpoints.NewConnectionHelperTests.test_readExisting
===============================================================================

​http://buildbot.twistedmatrix.com/builders/freebsd-9.2-amd64-python2.7/builds/13/steps/select/logs/stdio

which I also cannot reproduce when running trial on that failing test or on complete twisted.conch .. see 2nd attached file.