IGMP Version 2

Hi @shehreyar_khan,

Something to keep in mind is that IGMP and PIM are two separate protocols, they don’t really interact. IGMP is used on the “host” side so that a router can figure out which devices are interested in what multicast groups.

On the “router” side, we use PIM to get multicast traffic flowing from one router to another. The periodic queries that the router sends because of IGMP version 1 are meant for the hosts and don’t have anything to do with PIM :slight_smile:

The only exception to this is IGMP version 3, it is required for SSM (Source Specific Multicast) because only IGMP version 3 is able to request a certain source.

Hi Amit,
By reading this post I understand that, if multiple hosts are there in a single multicast segment (such as swiched LAN) then all host don’t need to convey membership report to the router. Only one host can send it based on the lower MRT. It greatly scales the entire mechanism and router can avoid extra overheads. Now the question is that, if only one host can send membership report to the router then how does router send multicast traffic to other hosts those who are still alive? My understanding is that, as router is receiving membership report form one host so router knows it should send multicast traffic on specific interface. And whenever the traffic reaches on switch, then switch will forward it to all the host in that group because switch is also running IGMP Snooping.
If I am wrong and something is different there please be kind enough to assist me.

BR
Naser

Hi Rene,

Why the membership report sends to 224.0.0.0 - 239.255.255.255 multicast addresses and the leave membership sends to 224.0.0.2 (all routers) ?? I need to know when the IGMP use group specific address and when use all router multicast address and why ??

Hi Hussein,

Membership reports are sent to the specific multicast group address:

https://www.cloudshark.org/captures/909766aa3ea6

Membership leave group messages are sent to the 224.0.0.2 (all routers) address:

https://www.cloudshark.org/captures/7d789d000734

RFC 2236 explains why:

Leave Group messages are addressed to the all-routers group because other group members have no need to know that a host has left the group, but it does no harm to address the message to the group.

2 Likes

Thank you very much sir that was very helpful for me

Hi Rene,

Can you give a tutorial on IGMP Fast-Leave and IPTV?

Regards,
Siji

Hello Siji

Thanks for the suggestion for a lesson topic. I suggest you add it to the Lesson Ideas page here and have other Networklessons users view and vote for it to be one of the next lessons Rene creates.

I hope this has been helpful!

Laz

dear Agapides
have anice day,

kindly, in igmp v1 since we dont have Querier election process ,so who elect router to be in active Querier , if we have two routers connected on same lan segment & depend one which criteria ?
wait your nice feedback
god safe you

Hello Saif

In IGMPv1, there is no election process for the active querier. However, if there are two IGMPv1 routers on the same subnet, then it is the Designated Router (DR) that will also perform the role of the querier. The designated router is elected exactly for the purpose of avoiding the duplication of multicast traffic for connected hosts.

More information about this can be found here:

I hope this has been helpful!

Laz

thanks & highly appreciated , yes very helpful…

dear Agapides
kindly, just one question , mean hosts will recieved two copy of multicat traffic in IGMPV1 ?

or assert process prevent that

Hello Saif

My previous post was not clear and had some erroneous information. I have since revised it to correct it. You can take a look at it here where your question is also answered:

I hope this has been helpful!

Laz

Hi,

In my setup I have a router R1 with an interface configured with Pim sparse mode. It is sending the IGMP v2 query. I have a LAN segment attached to this interface. There are two hosts H1 and H2. H1 joined 239.1.1.1 and H2 joined 239.1.1.2. Couple of questions

  1. When H2 sends membership report to 239.1.1.2 H1 is also receiving it. Likewise when H1 sends membership report H2 is receiving it. Ideally only the router R1 should be receiving it. Not sure why H1 and H2 are seeing each other’s reports.
  2. When I ping 239.1.1.2 from R1 which H2 is listening, I can see the ICMP ping requests are sent to H1 as well. Ideally the packet should be sent only to H2. Why is H1 receiving the ICMP destined to 239.1.1.2 ?

Thanks
Krishna

1 Like

Hello Krishna

If H1 has successfully registered to 239.1.1.1 and H2 to 239.1.1.2, then membership reports sent by each device should not be visible to the other device. The same goes for pings. Check to see if both H1 and H2 have registered to both groups and verify that this is not the case before testing again.

I hope this has been helpful!

Laz

Hi Rene/lagapides,
I have found the same that Krishna found . Ensure H1 is registered on 239.1.1.1 & H2 is registered on 239.1.1.2 and the two group found in Multicast Routing table in R1 .
My questions …

  1. How Switch segregated 239.1.1.1 group Traffic & 239.1.1.2 group Traffic ?
  2. when Router send traffic for Group 239.1.1.1 The L3 Dst. IP will be 239.1.1.1 & Dst. MAC will be 01-00-5E-01-01-01 . So, How Switch Segregate that traffic will send to H1 only.
  3. Again when Router send traffic for Group 239.1.1.2 The L3 Dst. IP will be 239.1.1.2 & Dst. MAC will be 01-00-5E-01-01-02 . So, How Switch Segregate that traffic will send to H2 only.
    So, H1 & H2 will not see each other group traffic .Thanks

BR//ZAMAN

Hello Mohammad

The mechanism by which a switch will verify that multicast traffic is forwarded only to the ports where multicast group members exist is called IGMP Snooping. This can be configured on L2 switches, and you can find out more about it at this lesson:

I hope this has been helpful!

Laz

Dear Rene,

i have a quick inquiry regarding same point the Membership leave group msg:
in igmpv2 topic i understood that when H2 need to leave the group he will send leave group membership to destination 224.0.0.2 , then Router will send MSQ msg to make sure if there is another host still need otherwise will remove the 239.1.1.1 from the table, then H1 send again MSR msg asking for same stream 239.1.1.1 then again Router will send MSQ for same 239.1.1.1 ,

The question comes here i need to know when the router sends the last MSQ with src:192.168.1.1 and dst:239.1.1.1 will this msg also reach to H2? however H2 already sent leave msg and don’t want to get stream anymore , so if your answer is yes every host will keep receiving the 239.1.1.1 even if they don’t want this stream and they will drop the packet i think , but in this case it will be overload traffic towards the hosts,

in Brief lets say if we have multiple hosts and multiple Multicast group addresses which means multiple streams in the network , and few of hosts sending MSR for different MCGA 239.x.x.x ,so will the other quite hosts keep receiving the MSQ msg which contains all different Multicast group address 239.x.x.x from the router ? or only the hosts which already sent the MSR msg will only receive the MSQ msg 239.x.x.x from the router and the other will not?

Hello Ziad

Remember that a multicast router will only keep track of the interfaces out of which multicast traffic must be sent. It doesn’t keep track of the number of hosts asking for such traffic. So once it receives a leave request on the Gi0/1 interface, it must determine if there are still hosts interested that are on this network segment. To do this, once the router receives the leave request from H2, it will send out the membership query.

Now the question is “who hears this membership query?” As your example describes very well, if we have multiple hosts on this same segment, and we have multiple multicast groups, will everyone receive all multicast traffic and then have to process them and drop them or is there some mechanism that allows these packets to reach only those devices that are still interested in that specific traffic?

The answer is: it depends on the switch.

Although it is not shown in the diagram in the lesson, you would have a switch connected to the Gi0/1 interface, and you would have all the hosts on that subnet connected to that switch. Now a switch, unless configured differently, treats multicast traffic in the same way as it treats broadcast traffic. This means that all traffic with a multicast destination will indeed reach all hosts on that segment. This is the case unless you configure IGMP Snooping. This is a feature that allows switches to “know” which ports specific multicast traffic destinations should be forwarded to.

You can find out more about this feature at the following lesson:

So regardless of whether the packet being sent is a membership query, or if it is simply regular multicast traffic, it is the IGMP Snooping feature in the switch that will be able to avoid having unnecessary multicast traffic reaching hosts on the same segment, that don’t want that traffic.

I hope this has been helpful!

Laz

1 Like

Hi Lagapides,
Thank you for your response i got the ideas now, so the answer is using igmp snooping in the L2 switch :smiley: .

1 Like

found some conflicting statements regarding the Querier election in IGMPv2.

in the train signal multicasting section Chris Bryant says that the “router with the lowest IP address is the querier”

in the CCNP self study book on page 604?: “router with the highest IP address is the designated querier”

in the CCNP official exam cert guide on page 484: “all routers start as queries and transition to non queries if they hear another querier with a higher IP address”

so can anyone tell me which is correct?