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

    rough-minister-58256

    04/06/2016, 12:24 AM
    presumably this is being executed in a multiprocessing.Pool, a child process dies, the pool hangs, the end
  • r

    rough-minister-58256

    04/06/2016, 12:25 AM
    @full-kilobyte-93429 ^
  • r

    rough-minister-58256

    04/06/2016, 12:25 AM
    🚬 🔫
  • f

    full-kilobyte-93429

    04/06/2016, 12:25 AM
    hmm
  • r

    rough-minister-58256

    04/06/2016, 12:26 AM
    whats odd is that it seems to be intermittent, afaict
  • r

    rough-minister-58256

    04/06/2016, 12:26 AM
    I think we can xfail this one test for the moment to unblock travis
  • f

    full-kilobyte-93429

    04/06/2016, 12:27 AM
    +1
  • w

    wide-energy-11069

    04/06/2016, 12:27 AM
    @rough-minister-58256: how do you enable the stacktrace in travis? especially in past/existing builds
  • r

    rough-minister-58256

    04/06/2016, 12:28 AM
    I used the travis admin console to export 
    PANTS_TEST_PYTEST_OPTIONS=“-vvs”
    globally
  • r

    rough-minister-58256

    04/06/2016, 12:28 AM
    and then re-ran the failed jobs until one hung
  • r

    rough-minister-58256

    04/06/2016, 12:28 AM
    travis reapplies the env settings live to each run
  • f

    full-kilobyte-93429

    04/06/2016, 12:28 AM
    wonder if this is to do with the underlying lmdb, anyway, xfail is probably the best we can do, thanks for debugging this!
  • w

    wide-energy-11069

    04/06/2016, 12:28 AM
    nice. thanks!
  • r

    rough-minister-58256

    04/06/2016, 12:29 AM
    that trace is likely from the default python exception handler in the child process which still has a handle to the parent processes stderr
  • r

    rough-minister-58256

    04/06/2016, 12:29 AM
    sgtm, will put out an RB
  • f

    full-kilobyte-93429

    04/06/2016, 12:30 AM
    wait, https://travis-ci.org/peiyuwang/pants/jobs/121026477 also fails, that i started earlier to see if by disabling certain test would make it pass
  • f

    full-kilobyte-93429

    04/06/2016, 12:30 AM
    the one i disabled is 
    test_rerun_with_cache
    , that failed too, meaning we might want to disable more
  • r

    rough-minister-58256

    04/06/2016, 12:31 AM
    ah k
  • r

    rough-minister-58256

    04/06/2016, 12:31 AM
    lets start with that one and see what happens - the verbose flag should be on globally, so any failures form now on out should have more data
  • f

    full-kilobyte-93429

    04/06/2016, 12:32 AM
    i have another https://travis-ci.org/peiyuwang/pants/builds/121027237 that switches to use 
    in_memory=True
    , wonder if that makes a difference
  • r

    rough-minister-58256

    04/06/2016, 12:32 AM
    ah heh, I have one out for that too
  • f

    full-kilobyte-93429

    04/06/2016, 12:32 AM
    cool
  • r

    rough-minister-58256

    04/06/2016, 12:33 AM
    ah tho yours will beat me due to queue
  • f

    full-kilobyte-93429

    04/06/2016, 12:37 AM
    have to run, will be back online probably in 30min, probably either disable one or more tests or if switching to in_memory works
  • r

    rough-minister-58256

    04/06/2016, 12:40 AM
    rgr
  • r

    rough-minister-58256

    04/06/2016, 12:40 AM
  • r

    rough-minister-58256

    04/06/2016, 12:40 AM
    created this to track - and threw a second error in w/ a slightly diff exception
  • f

    full-kilobyte-93429

    04/06/2016, 12:41 AM
    ok, assigned to myself for now
  • b

    bored-art-40741

    04/06/2016, 12:58 AM
    Thanks for running this down, Kris
  • r

    rough-minister-58256

    04/06/2016, 12:58 AM
    np
1...112911301131...2420Latest