I am trying to do a Symantic Ghost multicast session from a server to a client that is on another subnet, but on the same router. The problem is that when I do a session, the client subnet multicasts the sessions, but the server subnet is flooded with the session, even though there are no clients recieving the m-cast traffic on that subnet. Essentially on the server subnet, the session is sent to an m-cast session that consists of everybody. If I add a client to the session that is in the same subnet as the server, everything works just fine. It only fouls up when there is no client in the server's subnet. This is not cool, but I can't figure out how to fix it. The topology consists of server -> layer 2 switch -> router -> layer 2 switch -> client. The router is Cisco 4800 layer3 switch with IP multicast-routing turned on using PIM-sparse mode. The switches are Allied Tellesyn layer 2 managed switches with IGMP snooping turned on. Please help!!!

I'd suspect something in the router setup. Are you able to communicate properly between both subnets? Are both subnets able to connect to the server properly?

I have communication between the subnets and both subnets can connect to the server. The problem is that the Ghost session is flooded through out the server subnet (it is being sent to a multicast group which everyone on the subnet joins... virtually a broadcast), but the session is multicasting properly on the client subnet.

The only way I have seen around this problem is to put a client in the server subnet. Then it multicasts properly, but this is not the optimal solution. What should happen is that, on the server subnet, the server should send the data only to the router (the multicast group that the router is a part of and no one else)

I know this won't solve your problem, but I have some questions about your Layer-3 switch as I have never been afforded the opportunity to work with one.

So basically it's just a regular switch with routing capabilities correct? I'm sure you assign the switch an IP address and mask, and does that IP address become to default-gateway for clients wired to the switch on the same subnet? How is yours setup to be a router between multiple subnets? Do you designate one of the 10/100 of a fiber interface as being the interface to forward packets when they match quad zero's?

Thanks for the info and sorry I couldn't be of any help.

The problem is that the Ghost session is flooded through out the server subnet

I'm not understanding you completely here. What do you mean by "server subnet"? Is it one of the two you mentioned or a different subnet? Or are you saying that when you enable the ghost session on the server, both subnets are flooded?

How is your multicast group setup? Is it based on computers, subnets, or users? Is it on both subnets or defined on a single subnet? Where is it configured...withing ghost, on the server or in the router?

They are two different subnets off of the same router. The Ghost server is in one subnet, and the Ghost clients are in another. There are no Ghost clients in the server subnet for this particular Ghost session.

When I try to do a multicast Ghost session, the session is multicasted successfully to only the Ghost clients on the client subnet, but when I look at the subnet that the server is on, the session is multicasted to a multicast group that has every host in the subnet on it... which is not cool. It is virtually a broadcast. This should not happen.

What should happen is that the Ghost session should not be seen at all in the subnet that the server is on, becuase there are no Ghost clients in that subnet for that session. The flooding of the Ghost session to hosts that aren't a part of the session slows down the network. It is a virtual broadcast storm and brings the network on that subnet to a crawl.

The only way I have figured out how to fix this, is to make a Ghost client that is on the server subnet a part of this specific Ghost session. Then, it multicasts the session in the server subnet correctly by sending the Ghost session to ONLY the Ghost clients. This solves the problem, but it is not the optimal solution. With this solution, I will have to have a Ghost client on the server subnet dedicated to the specific task of being a part of every single cross-subnet Ghostcast session. I would rather it work the way it is supposed to instead of having to outsmart it and be minus a fully functional computer

It seems to me that the heart of the problem is that since there are no Ghost clients in the server subnet for that session (unless I do my little fix as mentioned up above), there is no destination multicast group in the server subnet, so it just sends the session to the default multicast group... which is everybody. This should not be the case. I dunno if the problem is with Ghost, the switches or the router.

I have a similar problem. I sent out a image via "Multicast" and I basically took down our Cisco 6509, which took everyone down. At the time the Ghost server and the client in question were on the same segment. I have since stop using Multicast and use ONLY Unicast. I have had no issues since. I can send images across segments without problem. I know the Ghost manual says to use Multicast for 3 or more clients, but who can afford to bring down the network. I have recently moved our Ghost Server to its very own segment, where there are no clients on that segment and I can still Unicast images sucessfully to other segments. I wish I knew why Multicasting causes the flooding, but at this time I don't know. Good luck with your issue.