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

Re: [Xen-users] Multicast message handling in VMs - For Upnp server, client or forward in a VM

Thanks for so fast answer; I indeed forgot to check it this time (in my previous attempt on an older system I am sure I checked the obvious and many parameters).
However, all interfaces on host and VM show the "UP BROADCAST RUNNING MULTICAST" statement.

So, as you tell me to check the obvious, do you think that it should propagate the mcast packets to the VMs out of the box?


PS: I attached iptables, ifconfig, and sysctl of host and VM.
(in this example, the inferface is eth1 in the VM and xenbr2/wlan0/vif10.1 on the host
It changed since last try because I used a cleaner VM (fresh installation too) with only two inferfaces in order to simplify the configuration

Le 01/02/2014 10:36, David Sutton a ÃcritÂ:

On Sat, Feb 1, 2014 at 3:10 AM, Calmar <calmar@xxxxxxxxxxx> wrote:

I am using xen 4.1.4 of the Debian Wheezy distribution, with network in bridge mode.
This is a brand new setup, with default setup options on the network point of view.

When a multicast message (here it was Upnp) is sent from the physical network, it can be seen in the eth1 and xenbr1 interfaces, but not in the interfaces inside the VMs.

I have a wlan0 interface on the physical machine that is bridged and seen as 'eth2' interface on a VM.
The wlan0 interface receives the broadcast messages as seen here:
~# tcpdump -ni wlan0Â net
tcpdump: WARNING: wlan0: no IPv4 address assigned
tcpdump: verbose output suppressed, use -v or -vv for full protocol decode
listening on wlan0, link-type EN10MB (Ethernet), capture size 65535 bytes
21:04:30.711511 IP > UDP, length 98
21:04:30.711527 IP > UDP, length 98
21:04:30.794978 IP > UDP, length 98
21:04:30.794992 IP > UDP, length 98

I added a rule to increment the ttl of the broadcast packet that is received on the physical interface:
Âiptables -t mangle -A PREROUTINGÂ -d -i wlan0 -j TTL --ttl-inc 1

However, nothing is seen inside the VM:
Âtcpdump -ni eth2 net
tcpdump: verbose output suppressed, use -v or -vv for full protocol decode
listening on eth2, link-type EN10MB (Ethernet), capture size 65535 bytes
0 packets captured
0 packets received by filter
0 packets dropped by kernel

The next step will be to propagate the packet on the other interfaces, but as soon I got nothing on the original bridged interface,...
In some tests I made a few months ago I got "dropped by interface" packets, but now on this new machine, I get nothing more.

So I am seeking for clues on this subject; I am sure that there should be some parameter to change in order to enable it, but so far I didn't find it...

Greetings all

PS: I posted my previous test here: http://www.xenproject.org/help/questions-and-answers/multicast-message-handling-in-vms-for-upnp-server,-client-or-forward-in-a-vm.html and Russell Pavlicek suggest to put the subject on this mailing list.

Xen-users mailing list
ÂJust to check the obvious; have you made sure that the given interfaces have their multicast flag set? I recently had some issues with a network interface and I noticed it was because the multicast flag hadn't been set; once set the mdns packets were able to be seen by the interface in question. ifconfig should show you the state of the interfaces.



Attachment: mcast.zip
Description: Zip compressed data

Xen-users mailing list



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