IGMP Version 2

IGMP Version 2 was developed as a solution for the first version's weaknesses. One of which was the inability to detect if there are still hosts which are still members of a multicast group. It will take 3 minutes before the router knows that there are no more members in version 1. The enhancements in Version 2 is shown below.

Group-Specific Query messages – enables the router to do a query operation on specific multicast groups

Leave Group messages – hosts sends a Leave message to the router that they are leaving the multicast group.

Querier election process – can elect the router that will do that query without relying on multicast routing protocols.

Maximum Response Time field - permits the Query Router to specify the maximum Query-Response time. This field
permits the tuning of the Query-Response process to control response burstiness and to fine-tune leave latencies.

The first 2 enhancements above, really helps in saving bandwidth. What happens is, if a host leaves the multicast group, it sends a Leave message. When the router receives this Leave message, it will send out a Group Specific query to that group of which the Leave message indicated and within 10 secs (default) if no host replies a Membership Report the router removes the entry in its IGMP membership table. The router doesn't have to wait 3 minutes like in Version 1.

Below is the Message types.

IGMP Version 2 Message Types

Membership Query (Type code = 0x11)

Two Subtypes:
       General Query – operates the same way like in the
       Group-Specific Query - enables the router to do a query operation on specific multicast groups.

Version 1 Membership Report (Type code = 0x12) - This is for backwards compatibility with version 1.

Version 2 Membership Report (Type code = 0x16)

Leave Group (Type code = 0x17)

Now let’s try this in the lab. I have set up the lab like before using Ubuntu and MINT. To start off, let's configure the router the same way in the IGMP Version 1 post. The command "ip multicast-routing" should be configured under global config mode. Commands "ip igmp version 2" and "ip pim sparse-dense-mode" should be configured on the FastEthernet interface. Let’s see the screenshots on Wireshark. Firstly I have generated the traffic in MINT for the Ubuntu VM to join multicast group 239.10.10.10 port 4321.


The screenshot above shows the Membership report which has a Type code of 0x16. Since in the lab that I made, there are no valid multicast sources, the host immediately sends a Leave message.

We see that the host sends the leave message to a multicast group 224.0.0.2 which means “all routers” in the local subnet. Its clear that they Type code is 0x17. Right below the Leave group message, immediately the router sends a Group Specific Query to check if there are still hosts who are still members on the same multicast group. Also it sends a General Query. Check the screenshots below to spot the difference.


Group Specific Query


General Query


The similarities between both is the Type Code which is 0x11. General Queries send it out to a well known multicast group 224.0.0.1 which is all hosts in the segment. Group Specific queries shows that it sends to the group address. In the Multicast address field, General Queries will have 0.0.0.0 which Group Specific will have the group address.

Finally let’s see the router debug output.

Router>
*Nov 18 15:41:32.759: IGMP(0): Received v2 Report on FastEthernet0/0 from 192.168.59.128 for 239.10.10.10
*Nov 18 15:41:32.759: IGMP(0): Received Group record for group 239.10.10.10, mode 2 from 192.168.59.128 for 0 sources
*Nov 18 15:41:32.763: IGMP(0): WAVL Insert group: 239.10.10.10 interface: FastEthernet0/0Successful
*Nov 18 15:41:32.763: IGMP(0): Switching to EXCLUDE mode for 239.10.10.10 on FastEthernet0/0
*Nov 18 15:41:32.763: IGMP(0): Updating EXCLUDE group timer for 239.10.10.10
*Nov 18 15:41:32.767: IGMP(0): MRT Add/Update FastEthernet0/0 for (*,239.10.10.10) by 0
*Nov 18 15:41:33.975: IGMP(0): Received Leave from 192.168.59.128 (FastEthernet0/0) for 239.10.10.10
*Nov 18 15:41:33.975: IGMP(0): Received Group record for group 239.10.10.10, mode 3 from 192.168.59.128 for 0 sources
*Nov 18 15:41:33.979: IGMP(0): Lower expiration timer to 2000 msec for 239.10.10.10 on FastEthernet0/0
*Nov 18 15:41:33.979: IGMP(0): Send v2 Query on FastEthernet0/0 for group 239.10.10.10
*Nov 18 15:41:34.979: IGMP(0): Send v2 Query on FastEthernet0/0 for group 239.10.10.10
*Nov 18 15:41:35.979: IGMP(0): Switching to INCLUDE mode for 239.10.10.10 on FastEthernet0/0
*Nov 18 15:41:35.979: IGMP(0): MRT delete FastEthernet0/0 for (*,239.10.10.10) by 0

Why did the router send 2 Group Specific Queries on the debug above? The router in this case used a “Last Member Query Interval” in the range of 10 -1 second in between the Group Specific queries. What it did in our case, was to send on Group Specific query and wait for 1 second and then sends another so by the 2nd time it sent the queries and there is no reply, we see now that it did remove the group from its MRT as the last line of the debug indicated. It should take around 3 seconds as shown above for the router to stop sending traffic for a group if there are no more members. Should there were other hosts in the same multicast group 239.10.10.10 the router will continue forwarding traffic.