----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/27461/#review59608 -----------------------------------------------------------
Thinking a little, I believe walking up to pid 1 like this is unnecessary. The test is checking that the process has been reparented. I had incorrectly assumed that all systems would reparent to pid 1. I think we can test reparenting by simply checking that the parent is *no longer* the child that forked it, and that it is still running, i.e., not zombied. I don't actually care who the new parent is, just that it has been reparented. - Ian Downes On Oct. 31, 2014, 8:03 p.m., Joris Van Remoortere wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/27461/ > ----------------------------------------------------------- > > (Updated Oct. 31, 2014, 8:03 p.m.) > > > Review request for mesos and Ian Downes. > > > Bugs: MESOS-2025 > https://issues.apache.org/jira/browse/MESOS-2025 > > > Repository: mesos-git > > > Description > ------- > > Reparenting does not always assign pid 1 (/sbin/init). If there is a user > init such as init --user with some other pid, this will be the new parent. > Modify os_tests to check up the parent tree, and succeed if there is a path > to pid 1 without zombies along the way. > This is not the cleanest fix, but I'm having trouble finding a way to find > the appropriate init to check for. > > > Diffs > ----- > > 3rdparty/libprocess/3rdparty/stout/tests/os_tests.cpp 3f39017 > > Diff: https://reviews.apache.org/r/27461/diff/ > > > Testing > ------- > > make check > > > Thanks, > > Joris Van Remoortere > >