(S,G) route not getting installed on edge in multicast stream

Products

• VMware NSX

Issue/Introduction

• The (S,G) rote is not getting installed on the edge node intermittently, which is resulting in a broken multicast stream.
• The mroute table on the edge shows (*,G) route present while receivers are behind NSX Edge, however, the (S,G) route is not present in the mroute table while RP (present in physical environment) has both (*,G) and (S,G) routes installed:
  
  edge(tier0_sr[4])> get mroute
IP Multicast Routing Table
Flags: S - Sparse, C - Connected, P - Pruned
       R - SGRpt Pruned, F - Register flag, T - SPT-bit for SSM FHR
Source          Group           Flags    Proto  Input            Output           TTL  Uptime
*               239.239.239.233 SC       IGMP   uplink-312       pimreg           1    00:54:57
                                         IGMP                    backplane-308    1


• In the mfib table on edge, both (*,G) and (S,G) routes are present, but (S,G) is pointing to another uplink interface of edge than the one which is receiving multicast streams:
  
  edge(tier0_sr[x])> get mfib
Tue Oct 07 2025 UTC 15:06:18.131
Logical Router
UUID                                   VRF    LR-ID  Name                              Type
aadbb5dc-xxxx-xxxx-xxxx-xxxxxxxc6   4      3      SR-T0-01                  SERVICE_ROUTER_TIER0
IPv4 Multicast Forwarding Table
Source IP          Group IP        Incoming UUID                          Outgoing UUID/VTEP/MTEP/UMG            Gateway IP
*                  239.239.239.233 none                                   105ded35-xxxx-xxxx-xxxx-xxxxxxx28a4
10.10.10.10     239.239.239.233 bae205a4-xxxx-xxxx-xxxx-xxxxxxxxf091   none
*                  *               none 


• If multicast traffic is stopped, the (*,G) route gets removed from both mfib and mroute table, but (S,G) route persists even if there is no multicast traffic flowing.

 

Environment

VMware NSX

Cause

• The multicast traffic was earlier being received on the uplink interface which is present in the mfib table for (S,G).
• This multicast stream was then stopped and another stream was initiated, which is now being handled by another uplink interface on edge.
  Thus, the (S,G) entry should be poiting to this uplink interface which is now handling the multicast stream, but it continues pointing to the uplink interface which handled the previous stream, indicating the stale presence of this entry in mfib table.
• PIM (Protocol Independent Multicast) within FRR sometimes fails to send the necessary delete notifications for (S,G) (Source, Group) routes to the Data Plane via RCPM.
  This results in stale (S,G) entries persisting in the DP's MFIB, even after FRR itself has internally removed or marked the route as uninstalled.

Resolution

This is a known issue impacting VMware NSX.

Workaround:

Put the edge node handling the multicast stream into maintenance mode.