> On Sept. 13, 2016, 6:35 p.m., Joseph Wu wrote:
> > src/slave/containerizer/mesos/isolators/network/cni/cni.cpp, line 818
> > <https://reviews.apache.org/r/51871/diff/1/?file=1498102#file1498102line818>
> >
> >     Will these symlinks resolve correctly if the child has a different 
> > rootfs than the parent container?
> 
> Avinash sridharan wrote:
>     The symlinks are created in the hostfs, and they are then bind mounted to 
> /etc/hosts, /etc/hostname, and /etc/resolv.conf in the container's rootfs 
> here:
>     
> https://github.com/apache/mesos/blob/3e52a107c4073778de9c14bf5fcdeb6e342821aa/src/slave/containerizer/mesos/isolators/network/cni/cni.cpp#L1583
>     
>     So yeah the symlinks should always resolve correctly, as long as the 
> 'root' parent container is alive.

The isolator will create a directory structure like:
```
/var/run/mesos/isolators/network/cni
  |-- parent-container-id/
      |-- hosts
      |-- hostname
      |-- resolv.conf
  |-- child-container-id/
      |-- hosts --> 
/var/run/mesos/isolators/network/cni/parent-container-id/hosts
      |-- hostname --> 
/var/run/mesos/isolators/network/cni/parent-container-id/hostname
      |-- resolv.conf --> 
/var/run/mesos/isolators/network/cni/parent-container-id/resolve.conf
```

Each file in the `child-container-id` folder is then mounted into the child 
container's rootfs.
After reading this ( http://www.proftpd.org/docs/howto/Chroot.html ), I'm not 
so sure this will work.  

The symlink after chroot-ing will point to an absolute location within the 
child's rootfs; whereas it was originally pointing to the host's rootfs.  A 
relative symlink won't work either, as you're not mounting the entire 
`/var/run/mesos/isolators/network/cni` directory.
The child will simply have a bunch of dangling symlinks:
```
/etc
  |-- hosts --> /var/run/mesos/isolators/network/cni/parent-container-id/hosts
  |-- hostname --> 
/var/run/mesos/isolators/network/cni/parent-container-id/hostname
  |-- resolv.conf --> 
/var/run/mesos/isolators/network/cni/parent-container-id/resolve.conf
```

Based on that, you may need to either hard-link the files or mount the parent's 
files directly.


- Joseph


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/51871/#review148839
-----------------------------------------------------------


On Sept. 14, 2016, 9:25 a.m., Avinash sridharan wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/51871/
> -----------------------------------------------------------
> 
> (Updated Sept. 14, 2016, 9:25 a.m.)
> 
> 
> Review request for mesos, Gilbert Song, Jie Yu, Joseph Wu, and Qian Zhang.
> 
> 
> Bugs: MESOS-6156
>     https://issues.apache.org/jira/browse/MESOS-6156
> 
> 
> Repository: mesos
> 
> 
> Description
> -------
> 
> The network file setup in the `network/cni` isolator is now nesting
> aware. Since the children share the network and UTS namespace with the
> parent, the network files need to be created only for the parent
> container. For the child containers, the network files will be simply
> a symlink to a parents network files.
> 
> 
> Diffs
> -----
> 
>   src/slave/containerizer/mesos/isolators/network/cni/cni.cpp 
> 822f11eab5b00c014563322a8c3b2c14cb440e0b 
> 
> Diff: https://reviews.apache.org/r/51871/diff/
> 
> 
> Testing
> -------
> 
> make
> make check
> sudo ./bin/mesos-tests.sh
> 
> 
> Thanks,
> 
> Avinash sridharan
> 
>

Reply via email to