Re: [Xen-devel] Live migration bug introduced in 2.6.32.16?

[Ian Campbell <Ian.Campbell@xxxxxxxxxx>]

On Thu, 2011-01-27 at 20:42 +0000, Nathan March wrote: 
> On 1/27/2011 12:36 PM, Konrad Rzeszutek Wilk wrote:
> > Whoa.. nice job figuring out what is wrong. One question: did you
> > try to use the git branch from Jeremy's tree as a DomU?
> >
> 
> No, should I? I've been under the impression that domU support in 
> mainline has been pretty solid for a while now. I'll give it a try if 
> you think it's worth it.

I think it's worth a shot since it would indicate whether or not we have
simply missed getting a patch into the upstream stable kernel or if we
have a totally new issue.

I fixed a bunch of migration issues which were introduced in the
2.6.32.16 time frame around about the 2.6.32.18-19 time frame (IIRC my
memory's a bit fuzzy). I wasn't aware of any further subsequent breakage
though.

Are the guests 32 or 64 bit? How about the hypervisor and dom0 kernel?
What is your kernel's .config?

If you edit /etc/sysconfig/{xencommons,xend} (whichever one you have)
then you can set XENCONSOLED_TRACE=guest which will write guest consoles
to /var/log/xen/domain. Also setting "on_crash=preserve" in your guest
configuration will hopefully cause them to stick around long enough to
see what was on the console.

> > git bisect is a great tool for this. You can fetch Linus's tree and
> > the tools for this - takes only about a day to find the culprit.
> >
> Haven't worked with git much yet but I'll take a look!

Thanks, that would be worthwhile.

Ian.



_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel