Auto-RP with "sparse-dense-mode"

[Andrew Xue Zheng Lay]

Hi,

I'm trying to configure Auto-RP with "sparse-dense-mode", and not "autorp listener", but I'm having trouble getting it to work.

Only R1, R2, & R3 are powered on.

R2 Configuration:
interface Loopback0
 ip address 10.0.0.2 255.255.255.255
 ip pim sparse-dense-mode
!
interface FastEthernet0/0
 ip address 10.2.3.2 255.255.255.0
 ip pim sparse-dense-mode
!
interface FastEthernet2/0
 ip address 10.2.1.2 255.255.255.0
 ip pim sparse-dense-mode
!
ip pim send-rp-discovery Loopback0 scope 255

R1 Configuration:
interface FastEthernet0/0
 ip address 10.1.1.1 255.255.255.0
 ip pim sparse-dense-mode
!
interface FastEthernet2/0
 ip address 10.2.1.1 255.255.255.0
 ip pim sparse-dense-mode
!
interface FastEthernet3/0
 ip address 10.2.2.1 255.255.255.0
 ip pim sparse-dense-mode

R3 Configuration:
interface Loopback0
 ip address 10.0.0.3 255.255.255.255
 ip pim sparse-dense-mode
!
interface FastEthernet1/0
 ip address 10.2.4.3 255.255.255.0
 ip pim sparse-dense-mode
!
interface FastEthernet3/0
 ip address 10.2.2.3 255.255.255.0
 ip pim sparse-dense-mode
!
ip pim send-rp-announce Loopback0 scope 255

R2#show ip pim rp mapping
PIM Group-to-RP Mappings
This system is an RP-mapping agent (Loopback0)

R1(config)#do sh ip pim rp map
PIM Group-to-RP Mappings

R1 & R2 are not learning any RPs.

[Mark Malone]

On the router that's announcing R3 set ip pim send-rp-discovery scope 255 and have you turned on ip multicast routing under each router that should get it going , from what i remember the announce and discovery were always set on the same rp router , you would get multiples rps if you do the same on R2
 

[Peter Paluch]

Hi Mark,

There is absolutely no need to configure the same router both as an RP-candidate and as a mapping agent. Having such requirement would be a major scalability and resiliency drawback. It is perfectly okay for R3 to be just an RP-candidate and for R2 to be just a mapping agent. If it does not work then there is a different issue in this network. I'd like to find out what it is ;)

Best regards,
Peter

[Mark Malone]

Hi

i dont see how its a scalability issue , you can have multiple auto-rp routers in a design , we use msdp and thats how we have it set each router has both the announce and discovery on it , even in Cisco guide below for just 1 rp design for  auto-rp its on 1 router but you can have the option for more to avoid a single point of failure

http://www.cisco.com/c/en/us/support/docs/ip/ip-multicast/9356-48.html#autowithone

[Peter Paluch]

Hi Mark,

First of all, please don't take my disagreeing as picking on you :)

The scalability/resiliency issues I have on my mind are related to the alleged requirement to have both the RP-candidate and the mapping agent configured on the same box. In other words, your suggested solution implied that the RP-candidate and the mapping agent must be configured on the same router. From the viewpoint of the RP autodiscovery mechanism design, if AutoRP worked this way, this would be hampering the scalability and resiliency: Every router intended to be an RP-candidate would be required to be a mapping agent as well, and every mapping agent would be required to be an RP-candidate as well. Moreover, if you had just one router configured for both roles (as it would not be possible to have them separate), its failure would bring the multicast down entirely. It would not be possible to have a separate and independent counts of RP-candidates and mapping agents. The split of the RP-candidate function and the mapping function would not have any meaning in such case.

Surely, the simplest AutoRP deployment is based on a single router configured for both roles. However, these two roles are independent and it is possible to have an arbitrary number of mapping agents (or better said, mapping agent candidates) and RP-candidates in a network.

Best regards,
Peter

[Mark Malone]

Yes true had not though of them as independent features i have always seen them set together but multiple routers providing rps through auto-rp on the network

[Peter Paluch · Accepted solution]

Hi Andrew,

You haven't posted the entire configurations so I am merely guessing, but can you double-check the following?

• Is the ip multicast-routing command activated on all routers in your topology?
• Are you running any internal routing protocol on all your routers and are the loopbacks advertised in this routing protocol? Remember that both the RP candidate and the mapping agent will be sourcing their advertisements from these loopbacks and are flooded using dense mode. If the RPF check fails because the loopbacks are not known to these routers, the advertisements will be dropped. 

Best regards,
Peter

[Andrew Xue Zheng Lay]

Hi,

I've attached the configuration.

[Andrew Xue Zheng Lay]

I've seen a lot of configuration examples on the Internet.

But for most examples, the Mapping Agent & the Rendezvous Point are directly connected.

Which... if they are directly connected, there isn't even a need for "sparse-dense-mode" at all.

Just "sparse-mode" everywhere, with the RP & MA side by side, will work.

[Andrew Xue Zheng Lay]

Lolz!

Resolved!

clear ip mroute *
clear ip pim rp-mapping

I think the old Multicast Route got stuck.