[lxc-devel] [PATCH] Add support for checkpoint and restore via CRIU

[Serge Hallyn]

Quoting Tycho Andersen (tycho.andersen at canonical.com):
> Hi Serge,
> 
> On Fri, Aug 22, 2014 at 12:09:31AM +0000, Serge Hallyn wrote:
> > Quoting Tycho Andersen (tycho.andersen at canonical.com):
> > > Hi Stéphane,
> > > 
> > > On Wed, Aug 20, 2014 at 05:19:14PM -0500, Stéphane Graber wrote:
> > > > On Wed, Aug 20, 2014 at 02:31:05PM -0500, Tycho Andersen wrote:
> > > > > Hi Stéphane,
> > > > > 
> > > > > On Wed, Aug 20, 2014 at 10:34:55AM -0500, Tycho Andersen wrote:
> > > > > >
> > > > > > Sounds good, I will make the changes.
> > > > > 
> > > > > Below is a revised version of the patch.
> > > > 
> > > > Some more comments below.
> > > 
> > > Here is an updated version with the comments addressed.
> > 
> > Thanks, Tycho.  I love how little was really needed in order
> > to make the monitor work.  A few question/comments below, sorry.
> > But after that I think I'm done.
> 
> Indeed!
> 
> > > +[ "network-unlock" = "$CRTOOLS_SCRIPT_ACTION" ] ||

Ok, but what is this ^^^ line supposed to do?

> > > +[ "network-lock" = "$CRTOOLS_SCRIPT_ACTION" ] || exit 0
> > 
> > What exactly is your intent with the two lines above?
> 
> CRIU has a number of callbacks (not just network-unlock or
> network-lock) that it invokes by setting CRTOOLS_SCRIPT_ACTION
> appropriately and then invoking the script. This is basically just to
> stop the script if the action isn't one of these two. Since we do a
> test below, we could get rid of this condition if we wanted, it just
> seems nicer this way.
> 
> Let me know what to do about the above, and I will make the rest of
> the changes you suggested and repost.
> 
> Tycho
> _______________________________________________
> lxc-devel mailing list
> lxc-devel at lists.linuxcontainers.org
> http://lists.linuxcontainers.org/listinfo/lxc-devel