> On May 11, 2018, 11:46 p.m., Jie Yu wrote: > > high level comments: should we just create `<rootfs>/dev` in the linux > > devices isolator "prepare" phase? The linux devices isolator should also > > prepare those standard devices. Logically, this makes sense because `/dev` > > should just be owned by linux devices isolator?
Yes, eventually I think we can standardize on creating all the devices in this isolator. If we do that, then we can also mount `dev` as a single mount rather than bind mounting all the devices separately. However, I'd prefer to to leave that for future work in a separate patch series. - James ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/67098/#review202977 ----------------------------------------------------------- On May 11, 2018, 6:32 p.m., James Peach wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/67098/ > ----------------------------------------------------------- > > (Updated May 11, 2018, 6:32 p.m.) > > > Review request for mesos, Gilbert Song, Jason Lai, Jie Yu, and Zhitao Li. > > > Bugs: MESOS-8792 > https://issues.apache.org/jira/browse/MESOS-8792 > > > Repository: mesos > > > Description > ------- > > The `linux/devices` isolator needs to make bind mounts into > the `/dev` directory of the container. However, the container > mounts are made before the container `/dev` is mounted as part > of the chroot preparation. We need to prepare the chroot, > then make any necessary container mounts, and finally enter > the chroot. This sequence of operations also requires us to > touch the target mount point, since we can't do it from the > isolator because the '/dev' directory doesn't exist yet. > > > Diffs > ----- > > src/slave/containerizer/mesos/launch.cpp > f25d90651ef32495c9161c3eaed8a327d1b2b926 > > > Diff: https://reviews.apache.org/r/67098/diff/1/ > > > Testing > ------- > > manual > > > Thanks, > > James Peach > >
