----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/25865/#review54169 -----------------------------------------------------------
src/slave/containerizer/isolators/namespaces/pid.cpp <https://reviews.apache.org/r/25865/#comment94166> Might be missing something of how pid namespaces work, but I don't see where we clean up the BIND_MOUNT_ROOT. Why do we have to (or can) assume this holder doesn't exist on launch? - Timothy Chen On Sept. 22, 2014, 6:46 p.m., Ian Downes wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/25865/ > ----------------------------------------------------------- > > (Updated Sept. 22, 2014, 6:46 p.m.) > > > Review request for mesos, Jie Yu and Vinod Kone. > > > Repository: mesos-git > > > Description > ------- > > Add namespaces/pid to --isolation slave flag. Places executor into a pid > namespace so it and all descendants will be contained in the namespace. > Requires the filesystem/shared isolator so /proc and /sys are remounted to > reflect the different namespace. > > > Diffs > ----- > > src/Makefile.am 9b973e5503e30180045e270220987ba647da8038 > src/slave/containerizer/isolators/namespaces/pid.hpp PRE-CREATION > src/slave/containerizer/isolators/namespaces/pid.cpp PRE-CREATION > src/slave/containerizer/linux_launcher.cpp > d5ef1d6aa762cf81a3e8384552d97fe95b9cbd95 > src/slave/containerizer/mesos/containerizer.cpp > 9d083294caa5c5a47ba3ceaa1b57346144cb795c > src/tests/isolator_tests.cpp c38f87632cb6984543cb3767dbd656cde7459610 > > Diff: https://reviews.apache.org/r/25865/diff/ > > > Testing > ------- > > Added test that command in pid namespaced container is in a different > namespace and that the command is 'init' (verifies remount of /proc). > > > Thanks, > > Ian Downes > >
