> On Jan. 23, 2018, 4:24 a.m., Vinod Kone wrote: > > Ship It! > > Vinod Kone wrote: > To confirm, would this new test have failed without the fix for > https://issues.apache.org/jira/browse/MESOS-8460 ?
Actually no. The root cause of MESOS-8460 is we used a framework pointer after that framework object is deleted. But when we remove a framework in agent code, we actually put the framework pointer into `BoundedHashMap<FrameworkID, process::Owned<Framework>> completedFrameworks;`, so the framework object will be deleted when the capacity of `completedFrameworks` is reached. The capacity of `completedFrameworks` is 50 (`MAX_COMPLETED_FRAMEWORKS`, there is no agent flag to modify it), so launching one framework in the test will not cause that segfault issue. - Qian ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/65263/#review195932 ----------------------------------------------------------- On Jan. 22, 2018, 9:21 p.m., Qian Zhang wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/65263/ > ----------------------------------------------------------- > > (Updated Jan. 22, 2018, 9:21 p.m.) > > > Review request for mesos and Vinod Kone. > > > Bugs: MESOS-8462 > https://issues.apache.org/jira/browse/MESOS-8462 > > > Repository: mesos > > > Description > ------- > > In the test `SlaveRecoveryTest.RecoverCompletedExecutor`, when the > completed executor is recovered, verify its work and meta directories > gc'ed successfully. > > > Diffs > ----- > > src/tests/slave_recovery_tests.cpp 6d8a57138903637d3cf9e5b19a466961d2c120d7 > > > Diff: https://reviews.apache.org/r/65263/diff/1/ > > > Testing > ------- > > Manually ran this test repeatedly (100 times). > > > Thanks, > > Qian Zhang > >
