Re: [Xen-devel] Save/Restore is not working properly

[Wei Liu <wei.liu2@xxxxxxxxxx>]

On Wed, Aug 10, 2016 at 02:24:09PM +0100, George Dunlap wrote:
> On Wed, Aug 10, 2016 at 12:11 PM, Roger Pau Monné <roger.pau@xxxxxxxxxx> 
> wrote:
> > On Sun, Aug 07, 2016 at 07:51:14PM +0430, Cendrin Sa wrote:
> >> Hi,
> >> I was searching a way to clone a machine using both memory and disk
> >> approach.
> >> I checked xen save/restore but after restoring, I can only work some
> >> seconds with my machine and it will crash with the_kernel_task_hang_up.
> >> using an script* to clone a machine is not working either.
> >> so is it a bug or something or I'm cloning the wrong way?
> >
> > Hello,
> >
> > I've not tried to perform cloning myself, but I have a little script to
> > perform VM checkpoints (so that you can restore the VM to any given point in
> > time). It's based on FreeBSD so it uses ZFS, but it should work with LVM
> > also if you replace it with the appropriate runes. AFAICT it should be quite
> > easy to expand it to also do VM cloning. This is transparent from a VM point
> > of view.
> 
> FWIW on a recent version of Xen-unstable, "xl save -c" appears to be
> broken, at least with me CentOS 6 VM.  If I do "xl save" then "xl
> restore", everything works fine; but if I do "xl save -c", then the
> save appears to work as normal, and after it's done the guest console
> has output similar to the output it has when restoring, but processes
> which access the disk hang, and in 2 minutes I get "hung process"
> output as Cendrin described.
> 
> I do get some warning messages though:
> 
> Using NULL legacy PIC
> WARNING: g.e. still in use!
> WARNING: leaking g.e. and page still in use!
> WARNING: g.e. still in use!
> WARNING: leaking g.e. and page still in use!
> WARNING: g.e. still in use!
> WARNING: leaking g.e. and page still in use!
> Changing capacity of (202, 0) to 4194288 sectors
> 
> This is the stock CentOS 6.6 kernel: 2.6.32-504.16.2.el6.x86_64
> 

It looks like the guest kernel is trying to free up all the grant
references.

In the case of xl save -c my impression is that it shouldn't be doing
that because the suspend is supposed to be canceled from guest's PoV.

See comment in xenctrl.h for xc_domain_resume.

Also related: 8903a7a5f6a47cc40c1c204a1cc28b0030b04486

Wei.

>  -George

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
https://lists.xen.org/xen-devel