[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] Xend loads slowly

On Thu, Jan 25, 2007 at 03:38:25PM +0000, Keir Fraser wrote:
> On 25/1/07 14:13, "Daniel P. Berrange" <berrange@xxxxxxxxxx> wrote:
> > Is there any work scheduled on the Xen roadmap yet to remove the need to
> > use netloop driver in Dom0 ? It would be very desirable to eliminate the
> > complicated peth0/eth0/vif0 interface renaming/duplication for Dom0 so that
> > the regular OS distro provided networking scripts 'just work' with Xen. So
> > if people wanted a bridged setup they can simply create their regular
> > /etc/sysconfig/network-scripts/ifcfg-br0 (or equivalant) - so XenD would
> > only need to worry about the vif-* scripts, and we can kill off network-*
> > scripts.
> You can set things up today without the loopback. But if domU<->dom0 traffic
> is allowed then you can end up with all domU tx slots tied up in dom0 socket
> buffers for unbounded time. If domU<->dom0 traffic is disallowed, or just
> not done, then the loopback serves no other purpose.

Yes, I was specifically thinking about the ideas that have been brought
up every now & then about fixing the netback side of things to avoid the 
dom0 socket buffers being pinned indefinitely. So we can safely have the
DomU <-> Dom0 traffic without needing netloop.

|=- Red Hat, Engineering, Emerging Technologies, Boston.  +1 978 392 2496 -=|
|=-           Perl modules: http://search.cpan.org/~danberr/              -=|
|=-               Projects: http://freshmeat.net/~danielpb/               -=|
|=-  GnuPG: 7D3B9505   F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505  -=| 

Xen-devel mailing list



Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.