https://pantsbuild.org/ logo
Powered by
#general
  • b

    bored-art-40741

    04/05/2016, 11:29 PM
    Or locally
  • b

    bored-art-40741

    04/05/2016, 11:30 PM
    But it's very persistent on CI
  • f

    full-kilobyte-93429

    04/05/2016, 11:30 PM
    travis?
  • r

    rough-minister-58256

    04/05/2016, 11:31 PM
  • r

    rough-minister-58256

    04/05/2016, 11:31 PM
    last 4 master commits error’d on a test_engine.py hang
  • f

    full-kilobyte-93429

    04/05/2016, 11:31 PM
    yea, just saw
  • b

    bored-art-40741

    04/05/2016, 11:31 PM
    Both Travis and Jenkins
  • r

    rough-minister-58256

    04/05/2016, 11:32 PM
    it was super occasional - then all of the sudden seems to have gotten worse with no real explanation
  • r

    rough-minister-58256

    04/05/2016, 11:33 PM
    @bored-art-40741: could we do a 
    export PANTS_TEST_PYTEST_OPTIONS="-vvs”
    pre-jenkins repro?
  • b

    bored-art-40741

    04/05/2016, 11:33 PM
    Sure
  • r

    rough-minister-58256

    04/05/2016, 11:33 PM
    that’d probably give us more useful info than an 
    -ldebug
  • b

    bored-art-40741

    04/05/2016, 11:33 PM
    Yeah and it doesn't fail the tests, too
  • b

    bored-art-40741

    04/05/2016, 11:34 PM
  • b

    bored-art-40741

    04/05/2016, 11:37 PM
    It got past it :|
  • b

    bored-art-40741

    04/05/2016, 11:38 PM
    I'll rerun
  • r

    rough-minister-58256

    04/05/2016, 11:38 PM
    doh
  • b

    bored-art-40741

    04/05/2016, 11:39 PM
    Same thing happened with ldebug. Feels like the output might shake loose the deadlock somehow
  • r

    rough-minister-58256

    04/05/2016, 11:40 PM
    for this one - as far as the test is concerned its still outputting to a similar fd tho, no? the options only change what pytest reveals to us
  • b

    bored-art-40741

    04/05/2016, 11:41 PM
    No clue
  • b

    bored-art-40741

    04/05/2016, 11:41 PM
    One would think?
  • b

    bored-art-40741

    04/05/2016, 11:41 PM
    hm, here's a question
  • r

    rough-minister-58256

    04/05/2016, 11:41 PM
    and -ldebug should only affect the log level
  • b

    bored-art-40741

    04/05/2016, 11:41 PM
    Are these locks fork-safe?
  • b

    bored-art-40741

    04/05/2016, 11:41 PM
    I'm guessing no
  • b

    bored-art-40741

    04/05/2016, 11:41 PM
    The PID based one almost certainly isn't
  • b

    bored-art-40741

    04/05/2016, 11:42 PM
    The file lock one is trickier since you're potentially also inheriting the fd
  • r

    rough-minister-58256

    04/05/2016, 11:42 PM
    im not sure there are any locks involved in the exp engine case
  • b

    bored-art-40741

    04/05/2016, 11:43 PM
    Also
  • b

    bored-art-40741

    04/05/2016, 11:43 PM
  • b

    bored-art-40741

    04/05/2016, 11:44 PM
    It looks like at the end of that pytest is ... running its own tests?
1...112711281129...2420Latest