Hi Ashvin
thanks for your reply.
unfortunatly i dont have the details of the overall setup as we just manage the client side part of the newtork, being this basically a VC juniper configured with an extended vlan for the client servers and the upstream switches facing the headend (ae0 if).
what i know is that there's no L3 Mcast router between the headend and the clients but the L2 equipment close to the headend is configured with the ip igmp snooping querier and the immediate-leave IGMP for our ingress vlan.
i confirmed that all the leave group meesages sourced from the clients were leaving the juniper switch towards the upstream ones, and because of some loss of data on the clients (incident flag) my guess was that the upstream nodes were configured with the immediate-leave, which has been later confirmed.
On the other hand, my concern or question was about the capabilities of the juniper switch to kind of work more L3 oriented and handle locally this leave group meesages meaning that only when the last client sends the leave group message, the local switch would then forward this message up to the upstream ones to cut the feed. This capability is what i could not find info about regarding the IGMP in the juniper platforms Also, if the juniper works purely like a L2 device, i guess that all the mcast traffic must be inspected by the switch CPU as the IGMP reports are directed to the Mcast groups mac address, being this the same one as the traffic itself, please correctme if i'm wrong. Altough my 1) and main question/concern was to confirm that the leav group message is forwarded up to the ae0 if.
Thanks again
/Gab