----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/32420/#review77507 -----------------------------------------------------------
Looks good! Just hoping we can avoid the process leak. src/tests/perf_tests.cpp <https://reviews.apache.org/r/32420/#comment125603> Can we guarantee that we don't leak a long-lived process? Since you're using linux, we could manually use PR_SET_PDEATHSIG here (I assume this works when the parent is already dead..?). Otherwsie, we can make this loop for a limited duration of time. Just something to prevent this from leaking if the test fails, crashes, or is killed before we get a change to SIGKILL the child. - Ben Mahler On March 24, 2015, 12:17 a.m., Ian Downes wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/32420/ > ----------------------------------------------------------- > > (Updated March 24, 2015, 12:17 a.m.) > > > Review request for mesos and Ben Mahler. > > > Bugs: MESOS-2534 > https://issues.apache.org/jira/browse/MESOS-2534 > > > Repository: mesos > > > Description > ------- > > init/systemd/launchd doesn't always run within a reasonable length interval. > > Changed test to fork and sample a child that spins. > > Reduced the test time too. > > > Diffs > ----- > > src/tests/perf_tests.cpp 16f32267f694ccaae9d10c2629831e2cca23e0c8 > > Diff: https://reviews.apache.org/r/32420/diff/ > > > Testing > ------- > > ./bin/mesos-tests.sh --gtest_filter="PerfTest.ROOT_SamplePid" > --gtest_repeat=100 > > > Thanks, > > Ian Downes > >
