> On March 25, 2017, 1 a.m., Anand Mazumdar wrote: > > src/tests/slave_tests.cpp > > Lines 5138 (patched) > > <https://reviews.apache.org/r/57883/diff/4/?file=1674591#file1674591line5138> > > > > hmm, thinking about this more. Can we set the expectations on this > > object first and then hand the `Owned` pointer to the `MockSlave` object? > > > > Currently, it looks possible that the test as well as the agent thread > > might be using the `secretGenerator` leading to a race?
Per our discussion, I'll leave this as-is for now, since it seems to be consistent with other test code; for example, the master authorization tests, which inject the authorizer as a raw pointer. - Greg ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/57883/#review170071 ----------------------------------------------------------- On March 25, 2017, 4:34 a.m., Greg Mann wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/57883/ > ----------------------------------------------------------- > > (Updated March 25, 2017, 4:34 a.m.) > > > Review request for mesos, Anand Mazumdar and Vinod Kone. > > > Bugs: MESOS-6999 > https://issues.apache.org/jira/browse/MESOS-6999 > > > Repository: mesos > > > Description > ------- > > This patch adds new tests, > `SlaveTest.RunTaskGroupFailedSecretGeneration` and > `SlaveTest.RunTaskGroupInvalidExecutorSecret`, to > verify the agent's behavior when generation of the > executor secret fails. > > > Diffs > ----- > > src/tests/slave_tests.cpp c31c670b667240c4876d415aa5cf90cb34273e8a > > > Diff: https://reviews.apache.org/r/57883/diff/5/ > > > Testing > ------- > > `make check` > `bin/mesos-tests.sh > --gtest_filter="*SlaveTest.RunTaskGroupFailedSecretGeneration*:SlaveTest.RunTaskGroupInvalidExecutorSecret*" > --gtest_repeat=-1 --gtest_break_on_failure` was used to verify that the new > tests are not flaky. > > > Thanks, > > Greg Mann > >
