> > > > That may work, but sounds quite complex to me.
> > >
> > > All of migration is quite complex :)
> >
> > But what you try here is even harder, because you have to save/restore
> > both sides of the connection. But you do not restore the server part
> > with criu, so you need to inject the
On Mon, Jun 22, 2015 at 05:23:05PM +0200, Dietmar Maurer wrote:
>
> > > That may work, but sounds quite complex to me.
> >
> > All of migration is quite complex :)
>
> But what you try here is even harder, because you have to save/restore
> both sides of the connection. But you do not restore th
> > That may work, but sounds quite complex to me.
>
> All of migration is quite complex :)
But what you try here is even harder, because you have to save/restore
both sides of the connection. But you do not restore the server part
with criu, so you need to inject the server state otherwise. I t
On Mon, Jun 22, 2015 at 06:41:16AM +0200, Dietmar Maurer wrote:
> > On dump, this would probably just be having the plugin query LXD to
> > make sure that all requests for the container were flushed so that
> > CRIU would at least dump the socket buffers for the kernel. On
> > restore, the plugin w