cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
2000
Views
0
Helpful
6
Replies

Multicast across Data Centres OTV & VPC

stephendrkw
Level 3
Level 3

I have 2 Data Centres setup with Virtual Buildings (i.e - 1 host in each building on the same subnet. SVI's are running IP PIM-SM)

4 x Nexus 7710 (2 in each DC)

Each Nexus is running VPC to each other and other, also other switches on the network connect to Nexus A and B by Port-channel VPC

Data Centre A

Is the DR with highest priority Nexus A (sh ip igmp interface vlan xxx)

Nexus A is the Primary VPC AGG role

Nexus B is the Secondary VPC AGG role (show vpc role)

Hosts in Data Centre A send multicast packets to many hosts with the same Multicast Groups and are both Senders and Receivers ASM (if anyone knows TIBCO in a banking environment they know that hosts act as a Publisher and Receiver)

Hosts in Data Centre A register successfully with the DR and the DR forwards to the RP successfully and multicast is working, however any multicast senders or receivers sent across to Data Centre B fails.

Data Centre B

Nexus C is the Primary VPC role

Nexus D is the Secondary VPC role (show vpc role)

Hosts are sending Multicast packets to hosts DC A and DC B no responses seen at all. I can see multicast been sent from hosts as the mroute counters are incrementing. After sniffing an example host I can see IGMP membership report coming from Nexus D which is the Primary role VPC. Why? shouldn't the highest IP address be the DR? which is Nexus D? And more importantly the membership report should be coming back from Nexus A in DC A which is the Primary DR

You should only have 1 DR in a subnet and that DR registers with the RP at the Core n our case. So this clearly says to me that the Multicast is not been sent across the OTV! or been screwed up on VPC. And the IGMP Querier should be coming from the DR, but for some reason in DC B the Primary IGMP Querier is  Nexus C, this means Nexus C acts as the mrouter, again mrouter should be in DC A from the DR.

I also traced Port-channels to see where the source mac example host in DC B was traversing and the path connects to Nexus D and via versa so not flipping between Po physcial links.

I have a TAC open at the moment, there is conference calls going on but still no resolution or suggestions why this is happening.I haven't published any commands as yet as I would like to see if anyone else has had this issue.

Thanks.

6 Replies 6

Hello,

can you post the configs of the four switches ?

This would breach my IT Security and Compliance rules. Sorry.

I could post some of the config and change certain parameters, if you need anything specific. 

Hello Stephen,

since the multicast within Data Centre A works, maybe you can post the part of the config with the link between Data Centre A and Data Centre B, that is, how the two sites are connected...

IPN config also attached

Problem resolved. IGMP version 3 (for SSM) was not running on the IPN interface end of the OTV join interface. Was on a webex with Cisco, they showed a very good document on how multicast works across OTV, that is not for public use. They didn't give me a copy! You can see another recent post I have with Multicast OTV, at the moment Cisco have not resolved this issue (with TAC)

Hi,

Please find attached part of configs, excludes VPC configs. Hope this is ok.

Site A (Nexus A and B)

Site B (Nexus C and D)