> On June 22, 2016, 6:19 p.m., Jie Yu wrote: > > src/slave/containerizer/mesos/isolators/network/port_mapping.cpp, line 1894 > > <https://reviews.apache.org/r/48994/diff/1/?file=1425967#file1425967line1894> > > > > Hum, just realized an issue. If you do that, bindMountRoot is no longer > > guaranteed to be realpath, right? Can you just move mdkir to the top? > > Gilbert Song wrote: > I dont see why it is no longer a realpath. Currently > `PORT_MAPPING_BIND_MOUNT_ROOT()` is hard coded as `/var/run/netns`. If it > does not exist, we mkdir. the newly created dir should be a realpath as well. > > But yes, moving mkdir above should be a cleaner solution and make more > sense.
/var/run might be a symlink to /run on some systems. Therefore /var/run/netns you just created might not be a realpath. - Jie ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/48994/#review139103 ----------------------------------------------------------- On June 21, 2016, 2:46 a.m., Gilbert Song wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/48994/ > ----------------------------------------------------------- > > (Updated June 21, 2016, 2:46 a.m.) > > > Review request for mesos, haosdent huang, Artem Harutyunyan, Jie Yu, and Cong > Wang. > > > Bugs: MESOS-5673 > https://issues.apache.org/jira/browse/MESOS-5673 > > > Repository: mesos > > > Description > ------- > > Fixed portmapping isolator bind mount root non-existed case. > > > Diffs > ----- > > src/slave/containerizer/mesos/isolators/network/port_mapping.cpp > 5df591893f3015926a7cfe07e9b57749c46f6367 > > Diff: https://reviews.apache.org/r/48994/diff/ > > > Testing > ------- > > make check > > > Thanks, > > Gilbert Song > >
