In my experience, running tests sequentially is unsafe. I prefer to use -j1: "Run tests in parallel using 1 child processes"
A new process is spawned for each test file. It reduces the risk of side effects between 2 test files. Victor Le lun. 1 juil. 2019 à 16:32, Paul Monson via Python-Buildbots <python-buildbots@python.org> a écrit : > > Can someone merge this PR please? > https://github.com/python/buildmaster-config/pull/97 > > I suspect that the socket tests are flaky no Windows ARM32 because of a lack > of resource, and I would like to test running one test at time to see if it > improves test reliability. > The tests seems less flaky when not running in a buildbot worker, but it > might just be that the sample size is too small when running manually > > Thanks, > Paul > _______________________________________________ > Python-Buildbots mailing list > Python-Buildbots@python.org > https://mail.python.org/mailman/listinfo/python-buildbots -- Night gathers, and now my watch begins. It shall not end until my death. _______________________________________________ Python-Buildbots mailing list Python-Buildbots@python.org https://mail.python.org/mailman/listinfo/python-buildbots
