Questions[igmp](server)

I am trying to start a bridge connection and force it to use IGMPV3.

The following command will explain my issue better :

The below is the output of force_igmp_version before issuing brctl commands

/ # cat /proc/sys/net/ipv4/conf/br/force_igmp_version
3
/ # cat /proc/net/igmp
Idx     Device    : Count Querier       Group    Users Timer    Reporter
1       lo        :     1      V3
                        E0000001     1 0:00000000               0
3       eth0      :     1      V3
                        E0000001     1 0:00000000               0
4       eth1      :     1      V3
                        E0000001     1 0:00000000               0
5       eth2      :     1      V3
                        E0000001     1 0:00000000               0

Then I issue a set of brctl command to add a bridge and add my eth0 and eth1 interface to the bridge and do ifconfig on the bridge. But surprisingly bridge created is still using IGMPV2.

/ # cat /proc/net/igmp
Idx     Device    : Count Querier       Group    Users Timer    Reporter
1       lo        :     1      V3
                         E0000001     1 0:00000000               0
3       eth0      :     1      V3
                        E0000001     1 0:00000000               0
4       eth1      :     1      V3
                        E0000001     1 0:00000000               0
5       eth2      :     1      V3
                        E0000001     1 0:00000000               0
6       br        :     1      V2
                        E0000001     1 0:00000000               0

Is there anything I am missing to configure to make the bridge start with IGMPV3?

How can I force the bridge to use IGMPV3?

I am trying to capture multicast traffic via Wireshark (actually TShark), however the switch has IGMP snooping enabled and will only send Multicast traffic on the ports that have an active IGMP subscription.

I am currently getting around this by having a separate application hold the groups open I wish to record, but I am trying to set up a system to start/stop recording data dynamically and this extra complexity is painful.

Is there a way to force Wireshark to send out IGMP Subscriptions for multicast groups it is recording?

I'm experiencing issues related to OSPF (which relies on Multicast) not allowing us to work with enough adjacencies and was told to look at: /proc/sys/net/ipv4/igmp_max_memberships

Now, while it's easy to just increase this number to infinity, is there a way to know how many IGMP memberships are currently held by the OS so I can ensure this is indeed the problem and what number should be used?

I have a server running openSUSE 11.3, with vlan5 and vlan40 configured on eth7.

I want to join two multicast IGMP v2 groups, one on vlan5 and one on vlan40.

When I send out the IGMP v2 Membership Report on vlan40, the multicast datagrams arrive and everything works.

When I send out the IGMP v2 Membership Report on vlan5, the multicast datagrams begin to arrive, but they stop after 100 seconds. Using Wireshark, I've observed that my server is not responding to IGMP v2 Membership Queries from the router on vlan5, so I suppose the router is timing me out of the group.

On vlan40, however, I can see that my server promptly responds to each IGMP v2 Membership Query with an IGMP v2 Membership Report.

Why is my server responding to IGMP v2 Membership Queries on vlan40, but not on vlan5? Where in Linux should I look for configuration options, parameters, or settings that might explain the difference in the behavior of vlan40 and vlan5?

I need to route IGMP/Multicast traffic across two networks. I have a Linux box with two NICs, one public, one private. Need devices on the private network to flow to public. I'm pretty certain I have iptables set up right, been trying to do this with IGMPProxy. It doesn't work on PCs on provate, but from the Linux server I can ping 224.0.0.1.