> On July 3, 2016, 5:32 p.m., Jie Yu wrote: > > src/tests/gc_tests.cpp, line 963 > > <https://reviews.apache.org/r/49520/diff/3/?file=1435174#file1435174line963> > > > > Is it possible that mountPoint is not set but test failed? Are we > > leaking mounts in that case? > > > > I would suggest that we override `CreateSlaveFlags` and save slave's > > work_dir in the test fixture. > > > > During teardown, you can call fs::unmountAll(workDir) to cleanup the > > mounts. > > Jiang Yan Xu wrote: > Hey Jie I didn't realize this, but it seems like Environment already > takes care of this (you wrote this part): > https://github.com/apache/mesos/blob/ada24cb6b29cc74e30ce31550459a6b14d409c96/src/tests/environment.cpp#L847-L861 > > And `umount -l` does cover this case here: so we don't even need a > separate test fixture anymore. > > Of course we should verify this.
Oh, right! Yeah, we probably can get rid of the test fixture! Can you test it? - Jie ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/49520/#review140560 ----------------------------------------------------------- On July 1, 2016, 6:22 p.m., Megha Sharma wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/49520/ > ----------------------------------------------------------- > > (Updated July 1, 2016, 6:22 p.m.) > > > Review request for mesos and Jiang Yan Xu. > > > Bugs: MESOS-5752 > https://issues.apache.org/jira/browse/MESOS-5752 > > > Repository: mesos > > > Description > ------- > > In ROOT_GarbageCollectorUndeletableFilesTest.BusyMountPoint there is > a race between the task creating a file and the test looking for it > which sometimes leads to assertion failure on the existence of the > file. > > > Diffs > ----- > > src/tests/gc_tests.cpp 2e23f04ea908aadcefb21e11203b95b94ec3c60b > > Diff: https://reviews.apache.org/r/49520/diff/ > > > Testing > ------- > > > Thanks, > > Megha Sharma > >
