> On July 29, 2014, 6:10 p.m., Ian Downes wrote: > > src/slave/containerizer/isolators/network/port_mapping.cpp, lines 1105-1108 > > <https://reviews.apache.org/r/24020/diff/1-2/?file=643992#file643992line1105> > > > > The sentinel file is the bind mount which keeps the namespace active? > > Do we not guarantee the bind mount is present if the forked executor has > > been signaled to exec? If so, then I don't understand how case 2 can occur.
No, the sentinel file is the file that slave write to the meta to indicate an executor has been destroyed. > On July 29, 2014, 6:10 p.m., Ian Downes wrote: > > src/slave/containerizer/isolators/network/port_mapping.cpp, lines 1918-1920 > > <https://reviews.apache.org/r/24020/diff/1-2/?file=643992#file643992line1918> > > > > Why not have deallocate() just return if passed an empty interval? We have some sanity check inside deallocate(). Yeah, I can do this check inside allocator. - Jie ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/24020/#review48999 ----------------------------------------------------------- On July 29, 2014, 5:44 a.m., Jie Yu wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/24020/ > ----------------------------------------------------------- > > (Updated July 29, 2014, 5:44 a.m.) > > > Review request for mesos, Chi Zhang and Ian Downes. > > > Bugs: MESOS-1649 > https://issues.apache.org/jira/browse/MESOS-1649 > > > Repository: mesos-git > > > Description > ------- > > See summary. > > > Diffs > ----- > > src/slave/containerizer/isolators/network/port_mapping.hpp dd1db7f > src/slave/containerizer/isolators/network/port_mapping.cpp 5645e20 > > Diff: https://reviews.apache.org/r/24020/diff/ > > > Testing > ------- > > sudo make check > > > Thanks, > > Jie Yu > >
