1.17 appears buggy

[jreback]

we have been using pytest-xdist quite sucessfully for a while on Travis CI (other CI's). Appears the 1.17 release broke things. Not sure entirely sure why.

here is a snipped (the bottom linked will take you there). I was also able to repro this (locally on macosx).

The command "ci/script_single.sh" exited with 0.
41.85s$ ci/script_multi.sh
[script multi]
PYTHONHASHSEED=4076114954
pytest -n 2 -r xX -m not single --junitxml=/tmp/multiple.xml --skip-slow --skip-network pandas
============================= test session starts ==============================
platform linux -- Python 3.6.1, pytest-3.1.1, py-1.4.33, pluggy-0.4.0
rootdir: /home/travis/build/pandas-dev/pandas, inifile: setup.cfg
plugins: xdist-1.17.0
gw0 [11171] / gw1 [11171]
scheduling tests via LoadScheduling
................................................................................................................................................................................................................................................................................sss...........sINTERNALERROR> Traceback (most recent call last):
INTERNALERROR>   File "/home/travis/miniconda3/envs/pandas/lib/python3.6/site-packages/_pytest/main.py", line 105, in wrap_session
INTERNALERROR>     session.exitstatus = doit(config, session) or 0
INTERNALERROR>   File "/home/travis/miniconda3/envs/pandas/lib/python3.6/site-packages/_pytest/main.py", line 141, in _main
INTERNALERROR>     config.hook.pytest_runtestloop(session=session)
INTERNALERROR>   File "/home/travis/miniconda3/envs/pandas/lib/python3.6/site-packages/_pytest/vendored_packages/pluggy.py", line 745, in __call__
INTERNALERROR>     return self._hookexec(self, self._nonwrappers + self._wrappers, kwargs)
INTERNALERROR>   File "/home/travis/miniconda3/envs/pandas/lib/python3.6/site-packages/_pytest/vendored_packages/pluggy.py", line 339, in _hookexec
INTERNALERROR>     return self._inner_hookexec(hook, methods, kwargs)
INTERNALERROR>   File "/home/travis/miniconda3/envs/pandas/lib/python3.6/site-packages/_pytest/vendored_packages/pluggy.py", line 334, in <lambda>
INTERNALERROR>     _MultiCall(methods, kwargs, hook.spec_opts).execute()
INTERNALERROR>   File "/home/travis/miniconda3/envs/pandas/lib/python3.6/site-packages/_pytest/vendored_packages/pluggy.py", line 614, in execute
INTERNALERROR>     res = hook_impl.function(*args)
INTERNALERROR>   File "/home/travis/miniconda3/envs/pandas/lib/python3.6/site-packages/xdist/dsession.py", line 539, in pytest_runtestloop
INTERNALERROR>     self.loop_once()
INTERNALERROR>     call(**kwargs)
INTERNALERROR>   File "/home/travis/miniconda3/envs/pandas/lib/python3.6/site-packages/xdist/dsession.py", line 664, in slave_testreport
INTERNALERROR>     self.sched.mark_test_complete(node, rep.item_index, rep.duration)
INTERNALERROR>   File "/home/travis/miniconda3/envs/pandas/lib/python3.6/site-packages/xdist/dsession.py", line 280, in mark_test_complete
INTERNALERROR>     self.node2pending[node].remove(item_index)
INTERNALERROR> ValueError: list.remove(x): x not in list
sys:1: ResourceWarning: unclosed file <_io.TextIOWrapper name=0 mode='r' encoding='UTF-8'>
sys:1: ResourceWarning: unclosed file <_io.TextIOWrapper name=1 mode='w' encoding='UTF-8'>
==================== 283 passed, 4 skipped in 36.29 seconds ====================

successful with 1.16.0

The following NEW packages will be INSTALLED:
    py:     1.4.33-py36_0 conda-forge
    pytest: 3.1.1-py36_0  conda-forge
py-1.4.33-py36 100% || Time: 0:00:00  18.05 MB/s
pytest-3.1.1-p 100% || Time: 0:00:00  20.72 MB/s
real	0m8.518s
user	0m7.392s
sys	0m0.544s
Collecting pytest-xdist
Requirement already satisfied: py>=1.4.22 in /home/travis/miniconda3/envs/pandas/lib/python3.6/site-packages (from pytest-xdist)
Requirement already satisfied: pytest>=2.7.0 in /home/travis/miniconda3/envs/pandas/lib/python3.6/site-packages (from pytest-xdist)
Collecting execnet>=1.1 (from pytest-xdist)
  Using cached execnet-1.4.1-py2.py3-none-any.whl
Requirement already satisfied: setuptools in /home/travis/miniconda3/envs/pandas/lib/python3.6/site-packages (from pytest>=2.7.0->pytest-xdist)
Collecting apipkg>=1.4 (from execnet>=1.1->pytest-xdist)
  Using cached apipkg-1.4-py2.py3-none-any.whl
Installing collected packages: apipkg, execnet, pytest-xdist
Successfully installed apipkg-1.4 execnet-1.4.1 pytest-xdist-1.16.0

with 1.17
https://travis-ci.org/pandas-dev/pandas/jobs/241382582

Package plan for installation in environment /home/travis/miniconda3/envs/pandas:
The following NEW packages will be INSTALLED:
    py:     1.4.33-py36_0 conda-forge
    pytest: 3.1.1-py36_0  conda-forge
py-1.4.33-py36 100% || Time: 0:00:00  19.42 MB/s
pytest-3.1.1-p 100% || Time: 0:00:00  22.30 MB/s
real	0m9.853s
user	0m7.700s
sys	0m0.596s
Collecting pytest-xdist
Requirement already satisfied: pytest>=2.7.0 in /home/travis/miniconda3/envs/pandas/lib/python3.6/site-packages (from pytest-xdist)
Collecting execnet>=1.1 (from pytest-xdist)
  Using cached execnet-1.4.1-py2.py3-none-any.whl
Requirement already satisfied: py>=1.4.22 in /home/travis/miniconda3/envs/pandas/lib/python3.6/site-packages (from pytest-xdist)
Requirement already satisfied: setuptools in /home/travis/miniconda3/envs/pandas/lib/python3.6/site-packages (from pytest>=2.7.0->pytest-xdist)
Collecting apipkg>=1.4 (from execnet>=1.1->pytest-xdist)
  Using cached apipkg-1.4-py2.py3-none-any.whl
Installing collected packages: apipkg, execnet, pytest-xdist
Successfully installed apipkg-1.4 execnet-1.4.1 pytest-xdist-1.17.0

[msarahan]

One more example: https://travis-ci.org/conda/conda-build/jobs/241427114#L1343

[nicoddemus]

pytest's own test suite also broke: pytest-dev/pytest#2487.

Opened #158 to revert the change and possibly make a new hot-fix release.

[enkore]

Consider retracting 1.17.0 until a fixed version is available; it seems to break most uses of xdist.

[nicoddemus]

1.17.1 is out, sorry everyone for the inconvenience.

Consider retracting 1.17.0 until a fixed version is available; it seems to break most uses of xdist.

Sorry @enkore did know about this option, will take a look at it in the future! 👍

[enkore]

Thanks for resolving it so quickly :)

Retracting releases is often shunned by people ("rolling forward is the only correct way"), and I mostly agree; as a stop-gap measure it can still be useful.

[jreback]

@nicoddemus thanks for the fast fix!