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

Re: [Xen-users] Bonding ­ dom0 or dom1 ? and why is vif1.1 missing in DomU?



On 10 Mar 2007 at 20:11, Julian Pawlowski wrote:

> Why not have a short look to Novells article?
> 
>  http://www.novell.com/coolsolutions/feature/17605.html
> 
> It works perfectly for me.

No unexpected messages in /var/log/messages? Which bonding mode do you use, and 
do 
you use physical access to the NICs in DumU?

Ulrich


> 
> 
> Am 10.03.2007 18:59 Uhr schrieb "Mark Williamson" unter
> <mark.williamson@xxxxxxxxxxxx>:
> 
> >> I have a machine with two different nics (sis900 and dmfe).
> >> I am trying bonding in Xen. (on FC6, intel x386, para virtualization).
> >> 
> >> In the past  I did successfully configured and used bonding on non-Xen
> >> Linux machines.
> >> 
> >> I saw some posts in this list and in other sites regarding Xen
> >> bonding, but I am a little bewildered
> >> so I want to verify something:
> >> 
> >> I am using the default /etc/xen/xend-config.sxp config file, which
> >> means I am using
> >> bridging (ifconfig can show xenbr0).
> >> 
> >> When booting into domain 0 both nics are recognized (as I can see from
> >> dmesg), also "ifconfig -a" from  dom0 shows eth0 and eth1.
> > 
> > The default config bridges eth0, so you'll need to tweak things a bit to get
> > both eth0 and eth1 working as a bonding device on that bridge...
> > 
> >> Now when I boot into domain U , "ifconfig -a" shows only eth0.
> >> /etc/modprobe.conf has a line saying "alias eth0 xennet" (but eth1
> >> does not appear in /etc/modprobe.conf).
> > 
> > That's fine - domUs just see a virtual ethernet device.  There's probably 
> > not
> > much point in exposing the two real ethernet devices to guests.  If you
> > configure bonding in dom0 you should be able to hide this detail completely
> > from the guests - they'll just see their virtual network going faster.
> > 
> >> Now I have two questions:
> >> 1)I want to configure bonding through the initscripts
> >> (namely , create /etc/sysconfig/network-scripts/ifcfg-bond0 and add
> >> MASTER/SLAVE entries in /etc/sysconfig/network-scripts/ifcfg-eth0 and
> >> /etc/sysconfig/network-scripts/ifcfg-eth1 ,
> >> etc.) Where should I do it ? In dom 0 ? or in dom U ? I saw somewhere
> >> in xen-user mailing list that it should be done in dom U ; but as I
> >> said, ifconfig -a does not show eth1 in dom U .
> >> Any tips or guidance as to what should I do or pointers to bonding on
> >> Xen that worked for you will be helpful.
> > 
> > If you edit the config you could pass it through, but I think it would be
> > better to try and do it in dom0 so that the guests don't need to be 
> > configured
> > individually.
> > 
> >> 
> >> 2)
> >>  After boot of domain U , I ran ifconfig -a ***in domain 0***, and I
> >> saw only ONE vif1.* instance (vif1.0) I expected to see also vif1.1
> >> for the second nic.
> >> The XenNetworking wiki page says : "If you create multiple network
> >> interfaces for a domU, it's ends will be eth0, eth1, etc, whereas the
> >> dom0 end will be vif<id#>.0, vif<id#>.1, etc."
> >> (http://wiki.xensource.com/xenwiki/XenNetworking). Did I misunderstood
> >> something ?
> >> Am I wrong somewhere?
> > 
> > Unless you explicitly specified two virtual network interfaces in the 
> > domain's
> > config, you won't get two appearing there.  This is probably fine for you;
> > 
> > I'd be inclined to try bonding eth0 and eth1 in the normal way in dom0 then
> > arranging for the bonding device to be added to the bridge instead of either
> > eth0 or eth1...  I'm not sure if there would be any nasty interactions with
> > the bridging and bonding code here.  If you have any problems, you could 
> > also
> > try using a routed rather than bridged setup.
> > 
> > Solutions to this will probably involve some fiddling with the dom0 config
> > and / or the networking setup scripts used to configure vifs.
> > 
> > Hope this clears up some things for you!
> > 
> > Cheers,
> > Mark
> > 
> 
> 
> _______________________________________________
> Xen-users mailing list
> Xen-users@xxxxxxxxxxxxxxxxxxx
> http://lists.xensource.com/xen-users



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


 


Rackspace

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