[libvirt] [PATCH] Allow use of file images for LXC container filesystems

[Daniel P. Berrange]

On Thu, Aug 04, 2011 at 11:55:21AM -0600, Eric Blake wrote:
> On 08/04/2011 09:41 AM, Daniel P. Berrange wrote:
> >From: "Daniel P. Berrange"<berrange at redhat.com>
> >
> >A previous commit gave the LXC driver the ability to mount
> >block devices for the container filesystem. Through use of
> >the loopback device functionality, we can build on this to
> >support use of plain file images for LXC filesytems.
> >
> >By setting the LO_FLAGS_AUTOCLEAR flag we can ensure that
> >the loop device automatically disappears when the container
> >dies / shuts down
> >
> >* src/lxc/lxc_container.c: Raise error if we see a file
> >   based filesystem, since it should have been turned into
> >   a loopback device already
> >* src/lxc/lxc_controller.c: Rewrite any filesystems of
> >   type=file, into type=block, by binding the file image
> >   to a free loop device
> 
> Neat trick.
> 
> If I understand correctly, 'dumpxml' for an lxc domain using one of
> these loopback files will have the live xml showing the loopback
> device created, while the configured xml still shows the original
> file.  But I don't see anything that reverts type=block back into
> type=file when doing a VIR_DOMAIN_XML_INACTIVE, other than the fact
> that live and configured xml are stored in separate domain def
> objects.
> 
> I'm worried that you might be missing some changes to
> domain_conf.[hc] to correctly manipulate actual vs. configured setup
> all within the context of live xml.  This is a similar problem to
> the recent <actual> for networks - we must track which loopback
> block device was tied to the the filesystem, in case libvirtd gets
> restarted before the lxc domain goes away, but we also don't want to
> tie ourselves to that device when the lxc domain reboots (rather,
> each boot should create a new loopback device, opened on the
> original file).
> 
> On the other hand, VIR_DOMAIN_XML_INACTIVE normally means you will
> only see the configured setup, and is only ever used on active
> configuration by virDomain[Managed]Save, which are currently
> unsupported with lxc. And the LO_FLAGS_AUTOCLEAR ends the loopback
> device as soon as the lxc domain quits, which is a convenient way to
> handle a lot of cleanup.
> 
> So while I didn't find any coding bugs in this patch, I need some
> reassurance that we are handling things correctly from the
> VIR_DOMAIN_XML_INACTIVE aspect before I give an ack.

The live & inactive XML documents both live in the libvirtd daemon
context.

The lxc_controler/lxc_container files both execute in libvirt_lxc
helper process context. So any changes made to the virDomainDefPtr
here are not visible anywhere else.


Daniel
-- 
|: http://berrange.com      -o-    http://www.flickr.com/photos/dberrange/ :|
|: http://libvirt.org              -o-             http://virt-manager.org :|
|: http://autobuild.org       -o-         http://search.cpan.org/~danberr/ :|
|: http://entangle-photo.org       -o-       http://live.gnome.org/gtk-vnc :|