I have little experience with multicast and spent many hours looking through Cisco documentation but still stuck with an issue.
The quick version is that I have a video system using multicast on its own dedicated VLAN, I configured an IGMP snooping querier on this VLAN and everything works correctly if I disable the trunk link to one core switch but not if both links are enabled.
In more detail I have a pair of Cisco Nexus 9k core switches and a number of 3650 access switches connected to both with a port channel. The AV sender/management device is on one of these access switches and the receivers are connected to different access switches, all devices are on a dedicated VLAN and everything seems to work fine for normal connectivity so no problems getting to web interfaces and similar.
Initially the system didn't work so I configured a querier on the first core switch using the command ig igmp snooping querier 192.168.10.4 and immediately a number of devices started working properly. I configured a querier on the other core switch using 192.168.10.5 but it's using the first querier which I think is correct.
However around half the receivers weren't reporting back to the management device so I disabled the trunk link to the first core switch for the management/sender and all the devices reported back and appeared to work normally. I tried swapping the trunk links so it only had a link to the first core switch and nothing worked so although the querier is active on the first core switch, the AV system is only working with the link to the first core switch disabled.
I've tried the no ip igmp snooping mrouter vpc-peer-link command but it doesn't seem to make any difference, from what I can see there are no orphan ports on this VLAN. I've had a look through the various commands to show the igmp snooping status but there's nothing that stands out as pointing to the cause of the problem.
I'm puzzled how this seems to work through one core switch but not both, I'd appreciate any suggestions in where I'm going wrong.
