> On July 10, 2019, 3:27 p.m., Benno Evers wrote: > > src/tests/containerizer/docker_containerizer_tests.cpp > > Lines 5250 (patched) > > <https://reviews.apache.org/r/71035/diff/1/?file=2153977#file2153977line5250> > > > > The test itself looks good, but is it possible to run it with a stopped > > clock? Given the various timers involved in the body, this feels like a > > prime candidate to become flaky sooner or later.
I ran into issues using a stopped clock due to the use of the Docker runtime. I'm not sure precisely why, but it seems that some Docker containerizer operations require the clock to be running; currently, no tests in this file stop the clock :-/ - Greg ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/71035/#review216483 ----------------------------------------------------------- On July 8, 2019, 6:28 p.m., Greg Mann wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/71035/ > ----------------------------------------------------------- > > (Updated July 8, 2019, 6:28 p.m.) > > > Review request for mesos, Benno Evers, Benjamin Mahler, and Joseph Wu. > > > Bugs: MESOS-9853 > https://issues.apache.org/jira/browse/MESOS-9853 > > > Repository: mesos > > > Description > ------- > > This adds a test which verifies that when a scheduler attemps to > override a task's default kill policy, the Docker executor will > honor that override. > > > Diffs > ----- > > src/internal/evolve.hpp ffbb342a42ec3501383500dee55dff3588725478 > src/internal/evolve.cpp 81de15e192580d2f35237e124c11902e1fc67a1d > src/tests/containerizer/docker_containerizer_tests.cpp > a6217581e20168c5114f733323e927a83ac59fd0 > > > Diff: https://reviews.apache.org/r/71035/diff/1/ > > > Testing > ------- > > `make check` > `sudo bin/mesos-tests.sh --gtest_filter="*ROOT_DOCKER_OverrideKillPolicy*" > --gtest_repeat=-1 --gtest_break_on_failure` > > Verified that this new test fails on current master, and passes when the > preceding patches in this chain are applied. > > > Thanks, > > Greg Mann > >
