Re: [Xen-users] VLAN guest networking: near 100% packet loss

[Ian Campbell <ian.campbell@xxxxxxxxxx>]

CC-ing the O/P, as is usually on the Xen lists.

ÂOn Fri, 2016-01-08 at 07:57 +0000, Simon Hobson wrote:
> Pavan Maddamsetti <pavan.maddamsetti@xxxxxxxxx> wrote:
> 
> > I am new to the mailing list
> 
> Welcome
> 
> 
> > My setup consists of Debian 8 on both dom0
> > and domU, and the machine is connected to a VLAN tagged port on the
> > switch meaning that the host must be VLAN-aware.
> > 
> > Dom0 has internet reachability but domU is only able to establish
> > connections to hosts on the LAN; pings sent to external machines come
> > back with virtually 100% packet loss but it is unlikely a routing
> > issue because sometimes one or two packets get through.
> > 
> > Dom0 network configuration is:
> > 
> > auto xenbr0
> > iface xenbr0 inet manual
> > ÂÂÂbridge_ports eth1
> > ÂÂÂbridge_stp off
> > ÂÂÂbridge_waitport 0
> > ÂÂÂbridge_fd 0
> > 
> > auto xenbr0.4
> > iface xenbr0.4 inet static
> > ÂÂÂaddress 192.168.4.13
> > ÂÂÂbroadcast 192.168.4.255
> > ÂÂÂnetmask 255.255.255.0
> > ÂÂÂgateway 192.168.4.1
> > ÂÂÂvlan-raw-device xenbr0
> 
> I'm no expert but I was under the impression that the bridging (not sure
> whether that's the host side or the Xen us of it) doesn't support VLAN
> tagging within the virtual switch.

FWIW I had the same impression. If I was doing this I would make xenbr0.4
use eth0.4 as the underlying device and use xenbr0.4 in my guest config.

Or as Fajar explained you could also expose the VLAN trunk all the way to
the guest and do the tagging there.

>  I have a host using a VLAN trunk like you describe, but what I've done
> is build a separate bridge for each VLAN :

In your naming scheme peth* are what would be default be eth* (e.g. pethint
might have been eth0 at some point) and eth* are the names of your bridge
devices? (Just to help Pavan translate, it's also how sme older versions of
Xen did things by default)

See alsoÂhttp://wiki.xen.org/wiki/HostConfiguration/NetworkingÂfor some
examples.

Ian.

_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxx
http://lists.xen.org/xen-users