----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/53031/#review153319 -----------------------------------------------------------
src/linux/cgroups.cpp (lines 329 - 335) <https://reviews.apache.org/r/53031/#comment222607> `remove()` is called from two places: `destroy()` and `prepare()`. We actually only need to address the first. How about we move this logic to `destroy()`? In `prepare()` we are removing a subcgroup created by ourselves so I don't think we need to handle that. This way we keep the semantics of `remove()` intact (otherwise we should call out this behavior in the documentation). src/linux/cgroups.cpp (lines 1667 - 1670) <https://reviews.apache.org/r/53031/#comment222608> Instead of doing this in many individual places, I feel like we can just do it here: Basically if the `killTasks()` has finished with a failure but the cgroup is gone, we set the future as ready. src/linux/cgroups.cpp (line 1676) <https://reviews.apache.org/r/53031/#comment222609> Quote cgroup with backticks. - Jiang Yan Xu On Oct. 19, 2016, 12:44 p.m., Anindya Sinha wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/53031/ > ----------------------------------------------------------- > > (Updated Oct. 19, 2016, 12:44 p.m.) > > > Review request for mesos, Jie Yu and Jiang Yan Xu. > > > Bugs: MESOS-6414 > https://issues.apache.org/jira/browse/MESOS-6414 > > > Repository: mesos > > > Description > ------- > > Failure in cleanup of non-existing cgroups is treated as success. > > > Diffs > ----- > > src/linux/cgroups.cpp 1475c8eb848ece2a093e8243b51e9ce08981dd7d > > Diff: https://reviews.apache.org/r/53031/diff/ > > > Testing > ------- > > All existing tests passed. > > > Thanks, > > Anindya Sinha > >
