This article describes the issue of multicast streams not being registered with the RP.

Setup:

Local Network ==> CE (SRX) <==> PE (Router)

Multicast group = 239.28.92.145

To replicate the issue, perform the following procedure:

1. Turn on pim traceoptions to see the log messages with flag all.


2. In this case study, it was clearly noticed that CE (SRX) was not receiving any PIM join messages from the PE, once the initial setup messages were received for the multicast group (that is, 239.28.92.145 in this example).


3. However, the PE device log from (provider) showed that they were sending PIM messages with the 239.28.42.145 join being refreshed at intervals of 1 minute.


4. A subsequent test was performed, after this one showed that the PE device continues to send PIM join messages at about 60 seconds intervals, until the receiver on the customer's device (SRX) eventually times out and issues an IGMP leave for the group; which triggers a cascading PIM prune message down into the provider's VPN.


5. This is immediately followed by a new IGMP group join, which is followed by the CE device seeing a new PIM join message for the group and the routing state was setup again; only to time out again in 210 seconds, as the CE still reported never seeing a PIM join from the PE after the initial messages.


6. The CE logs and the PE logs are contrary to each other. The PE device says that it is consistently sending the PIM joins to refresh the routing state, when the CE device never reports seeing them. One of the devices is reporting incorrectly.

•  Packet capture was performed at the CE and it validated the logs that only PIM Hello messages are seen at the CE, when the time out occurs.


• A similar capture, which was performed at the PE, revealed a MTU mismatch between the PE and CE.


• A MTU mismatch between the CE and the PE devices was causing the PIM refresh messages to come in as jumbo frames and getting dropped at the CE (SRX) input as a jumbo frame.


• The initial joins were working, as there was only 1to 10 joins in a single frame, which kept it under the MTU limit.


• The refresh joins were over 100 and consolidated into a single pim join message, which made the packet exceed the MTU; so they were dropped by the CE upon receipt.

To resolve this issue, match the MTU settings on both the CE and PE.