[Lxc-users] LLS=LXC+LVM+Snapshots

[Arie Skliarouk]

I abandoned the bind-based mounts as I could not make them work. I went
with direct /dev/ manipulation instead. See my blog entry on this:
http://skliarie.blogspot.com/2011/11/llslxclvmsnapshots.html

Please leave a comment if you find this useful.

--
Arie



On Sun, Nov 13, 2011 at 15:29, Arie Skliarouk <skliarie at gmail.com> wrote:

> Hi,
>
> In our company we deal with massive datasets that needs to be easy to
> copy, modify a fraction of it and scrape. Snapshotting is an ideal solution
> here.
>
> I started writing LLS scripts (will opensource them upon completion) that
> enable users to snapshot existing physical volumes and mount them from
> inside of the container.
>
> For that there is a daemon script on the host and client scripts on the
> vservers. The clients communicate with the server over a named pipe in a
> shared (bind mount) directory /lls (lls - LXC+LVM+Snapshots). The directory
> is actually a small partition that is mounted with --make-rshared flag on
> the host. The partition is mounted on each vserver. This enables a snapshot
> to be created on the host machine, mounted onto some /lls/snapshot_version
> directory and thus become visible to all vservers (security is not an issue
> here).
>
> This works as long as the vservers are up and running.
>
> If some partition is mounted onto /lls/abc, the partition is no longer
> visible after a vserver is started (it appears empty, both from the host
> and from the vserver). I suspect this relates to the way bind mount works.
>
> Please help me to understand the way bind mount work.
>
> PS. If you have a different approach to propose, please share.
>
> --
> Arie
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linuxcontainers.org/pipermail/lxc-users/attachments/20111124/9fb0d8ba/attachment.html>