----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/55612/#review162031 -----------------------------------------------------------
Fix it, then Ship it! It seems like a bit of a shame that the standard agent recovery path involves two calls to `statusUpdateManager->resume`, which are likely to occur in quick succession. However, it seems like we expect frameworks to handle duplicate status updates gracefully, so perhaps it's not a big deal. It's not clear to me yet from looking at the code if there would be an easy way to avoid the duplicates in this case. src/tests/slave_recovery_tests.cpp (line 1204) <https://reviews.apache.org/r/55612/#comment233290> As long as you're cleaning this up, might as well fix the spacing - should be just one space before the comment begins. - Greg Mann On Jan. 17, 2017, 1:06 p.m., Vinod Kone wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/55612/ > ----------------------------------------------------------- > > (Updated Jan. 17, 2017, 1:06 p.m.) > > > Review request for mesos and Greg Mann. > > > Bugs: MESOS-6922 > https://issues.apache.org/jira/browse/MESOS-6922 > > > Repository: mesos > > > Description > ------- > > Just like RecoverHTTPTerminatedExecutor test, the scheduler should > handle duplicate status updates after agent restarts. See the > attached ticket for details. > > > Diffs > ----- > > src/tests/slave_recovery_tests.cpp 9323cbb74dd555fe6c44af1ef6da70e0c2d76dcb > > Diff: https://reviews.apache.org/r/55612/diff/ > > > Testing > ------- > > make check > > Ran in a loop for 20 times. > > > Thanks, > > Vinod Kone > >
