situation: we have an virtual openswitch router, which we are using to communicate from our virtual vm“s to an external network, currently unicast based into the vms. now we wanna switch to multicast.
i was able to setup an multicast router in the netns, so that we are able to join an multicast group from the vms into the right switch over the router. but this runs into an timeout, because the impg queries are not available in the netns.
is there an way to manage this?
10.200.2.0/24 dev qg-552ffdec-c1 proto kernel scope link src 10.200.2.2
10.200.3.0/24 dev qr-f46ee891-ec proto kernel scope link src 10.200.3.1
is see:
17:42:24.710978 IP 10.200.2.2 > 224.0.0.1: igmp query v2
but inside the vm there is no query, only initial 3 three igmp reports
15:53:37.084064 IP 10.200.3.17 > 238.2.48.3: igmp v2 report 238.2.48.3
We see the traffic on the qbr interface, but the queries are blocked somehow. there are not forwarded to the tap device on the virtual machine.
we set the querier for the qbr to 1. and set the rules for libvirt network filter, but still no success.
Edit:
Actually we broke down the problem to the igmp query part. when we send igmp queries from one virtual instance to the network we will not recieve them on the other instances:
we see the igmp queries on the linux bridge device:
tcpdump -v -n -i qbrd4d83fa8-d5 igmp
but not on the tap device:
tcpdump -v -n -i tapd4d83fa8-d5 igmp
we are using kvm.
we added:
echo 1 >> /sys/devices/virtual/net/qbrd4d83fa8-d5/bridge/multicast_querier
on the bridge, and did no changes on the virsh network filter yet.
some system informations:
Linux compute007 3.8.0-29-generic #42~precise1-Ubuntu SMP Wed Aug 14 16:19:23 UTC 2013 x86_64 x86_64 x86_64 GNU/Linux
and
bridge-utils 1.5-2ubuntu7 Utilities for configuring the Linux Ethernet bridge
for any hint, i would be thankful.
