> On March 26, 2016, 4:24 p.m., Joris Van Remoortere wrote: > > 3rdparty/libprocess/src/subprocess.cpp, lines 419-423 > > <https://reviews.apache.org/r/45232/diff/2/?file=1312472#file1312472line419> > > > > I don't think it's obvious from the code that this does a fork, and > > that the *parent* will actually exit, and the *child* will continue > > execution from this point. > > Let's make this very clear.
Not sure whether this change is sufficient to what you wanted. - Joerg ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/45232/#review125521 ----------------------------------------------------------- On March 28, 2016, 11:57 a.m., Joerg Schad wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/45232/ > ----------------------------------------------------------- > > (Updated March 28, 2016, 11:57 a.m.) > > > Review request for mesos and Joris Van Remoortere. > > > Repository: mesos > > > Description > ------- > > Some newly created processes such as perf should be killed in case the > parent dies. Currently this is achieved by forking a new process from the > child process which serves as a 'watchdog' and kill the child if the parent > dies. This review introduces this as a general behavior into subprocess > (and hence removes the need for the custom setup function). > > > Diffs > ----- > > 3rdparty/libprocess/include/process/subprocess.hpp > e0c306aa5cf5f393abb73768bbd287c45730f076 > 3rdparty/libprocess/src/subprocess.cpp > b99bad04f7251169df3bfcec5dee459977440997 > 3rdparty/libprocess/src/tests/subprocess_tests.cpp > 09e46eb1ce10a6c38cc364077f0b5952579d46e7 > > Diff: https://reviews.apache.org/r/45232/diff/ > > > Testing > ------- > > Tested complete chain (see https://reviews.apache.org/r/45236/). > > > Thanks, > > Joerg Schad > >