[Lxc-users] Hi all please treat nicely as mother says I am special

[Serge Hallyn]

Quoting Bretton Woods (woods.bretton at yahoo.co.uk):
> Noob LXC user who thinks LXC rocks but is undersold.
> 
> What about Planet-LXC it should be done but then again before I go on Something I only knew about from a look at the Archive

An lxc planet would be neat, and I'd happily put my blog on one.

> Docker! Like wtf where did that come from? I do searches on LXC fairly regular and that one missed my radar.

It came up pretty recently, but I get links relating to docker handed to
me almost daily at this point :)

> I am not a programmer but as a consultant I like to keep up with what the Jones are doing.
> 
> I am sort of stunned but instead of a template name a git repo makes much more sense and someone has fired the lxc starting gun.
> 
> What do the others think my first thoughts are that I don't like it as it has to much marketing behind it, Docker Ha! Containers Yeah!

No no, docker is a cool project based on top of lxc.  Lxc's goal is to
provide flexible but easy-to-use containers, offering much more control
than docker does.  That's perfect for use by things like docker, or for
advanced admins.  Docker on the other hand is awesome for letting
not-so-advanced users make use of containers.

The version control of container contents is not something I'd want to
have in lxc itself.  It belongs exactly where it is, in docker :)

I'm really not quite sure why I see so much "us.vs.you" from some docker
users toward lxc.  <shrug>

> This is where I am confused about docker because with AUFS and union
> mount points it is possible to share a common rootfs? I hope so in the

Yes, aufs and overlayfs mount a writeable 'delta' directory on top of
a read-only lower layer.  You can make 10 such mounts for 10 different
containers.  At first the only disk space will be the single instance
of the read-only lower layer.  As the containers write to files, the
updated copies of those files are stored in the writeable layer.  It's
at a whole-file layer so will take a bit more space than doing it using
btrfs or zfs snapshots, but works with any underlying fs.

(btrfs and zfs snapshots, btw, being supported as backing stores for
containers)

-serge