testing(run): timeout in run/logging-manual

[tmatsuo]

I saw weird timeout in our py-3.8 builds.

Build

nox > Running session py-3.8
nox > Creating virtual environment (virtualenv) using python3.8 in .nox/py-3-8
nox > pip install -r requirements.txt
nox > pip install -r requirements-test.txt
nox > pytest --junitxml=sponge_log.xml
============================= test session starts ==============================
platform linux -- Python 3.8.3, pytest-6.0.1, py-1.9.0, pluggy-0.13.1 -- /workspace/run/logging-manual/.nox/py-3-8/bin/python
cachedir: .pytest_cache
rootdir: /workspace, configfile: pytest.ini
collecting ... collected 3 items

e2e_test.py::test_end_to_end PASSED                                      [ 33%]
main_test.py::test_no_trace_in_headers PASSED                            [ 66%]
main_test.py::test_with_cloud_headers PASSED                             [100%]

------- generated xml file: /workspace/run/logging-manual/sponge_log.xml -------
======================== 3 passed in 2103.89s (0:35:03) ========================
nox > Session py-3.8 was successful.


err: signal: killed

It seems like the test finishes rather quick, but btlr sent the signal.

[kurtisvg]

@tmatsuo It looks like the runtime was 2103.89s - 35 minutes. Is it possible that the job was hung on a different part - after nox was run?

[tmatsuo]

Oh you're right, so this is expected because now we limit each job to 30m:

python-docs-samples/.kokoro/tests/run_tests.sh

Line 149 in 5df67f0

"--max-cmd-duration=30m"

Let's close this for now and if the timeout continues, I'll do something about it.

[tmatsuo]

I actually ran the test, and the e2e test is taking a long time, way longer than expected. I think we should debug this further.