Multicast on a 5212

We really need a little bit more information than this.  Specifically, is it the data stream that is being received by the client PCs, or are you receiving some sort of IGMP query to the clients?

I'd also like to know what configuration options are enabled on the 5212, for Multicast Support.

Thanks for replying to this.

The 5212 is in a standalone network with no other switches or routers. When all the client PCs have left the multicast group they still receive the data stream, confirmed by the status LEDs on the switch blinking and Ethereal on the clients. They also receive the IGMP queries as expected but, of course, do not respond.

The IGMP settings on the 5212 are as follows:
IGMP status - Enabled
Act as IGMP querier - Enabled
Query count = 2
Query interval = 125 seconds
Report delay = 10 seconds
Query timeout = 300 seconds
IGMP version = 2

IGMP member port table (after all the clients have left the multicast group):
One VLAN (ID 1)
Only IP Multicast Address is 239.255.255.250 (is this a loopback address?) dynamically set on Port 1. Port 1 is connected to the PC that is viewing the Web Management Interface for the switch.

Multicast router port settings:
None attached.

Hope this helps.

Many thanks
Andy

239.255.255.250 is a multicast address utilized by XP clients and a protocol called Simple Service Discovery Protocol (SSDP).  SSDP is used for multiple services such as - Universal Plug and Play (UPnP) and Internet Gateway Device Discovery and Client Control.   Starting with Windows ME, by default UPnP is disabled, however, Internet Gateway Device Discovery and Client Control are enabled.  Essentially, this allows windows to dynamically discover routers that are UPnP enabled for ease of network management. 

Since the switch is enabled for the option to "Act as IGMP querier", it will periodically flood a multicast query to the network.  The clients will interpret this query as an UPnP router attempting to configure the network, and will reply with the membership report for a multicast group of 239.255.255.250. 

Since the switch is enabled for the option "IGMP Status", this is essentially IGMP Snooping.  The PowerConnect switch will dynamically build the mulitcast group of the host initiating the query and the clients responding with membership reports, and begin flooding traffic to the respective ports. 

What is the multicast address that is being advertised by your server?  What application is initiating the IGMP queries?

I see that the 5212 is enabled for IGMP v2, but is able to undertstand - fast leaves - i.e when a host decides to leave a multicast group and the host sends a IGMPv2 Leave Group message to all routers (224.0.0.2) - does the switch remove the MAC address of Host PC from the multicast group forwarding table on the switch immediately upon seeing that leave.

Just wondered if this contributed to the delay or apparent "latency" in a group data forwarding stopping going to a host......as was the case with IGMPv1 standard......and one of the key reasons v2 was written.

Randy

Many thanks for the info on SSDP and UPnP - now disabled on the hosts so 239.255.255.250 doesn't appear!

The IGMP queries are only being sent by the switch. The host PCs are only ever sending IGMP membership reports.

I've written a number of applications that act as the server and the choice of multicast IP address doesn't seem to make any difference. So far I've used 225.1.1.1, 225.2.2.2, 224.0.252.1 and 225.0.253.2. The first 2 were used to try out multicasting and just "ping" data to clients every second but the last 2 will be used to transmit uncompressed digitised video so 33MB/s of data flooding the network is causing problems!! None of then initiate IGMP queries.

regards
Andy

Andy,

Don't go writing applications in the 224 -239.0.0.0 area - as this is used by registed protocols - OSPF, DVMRP, PIM, HSRP, AutoRP, DOgFight etc etc

This IP address space is controlled via IANA/GLOP......and not to used by just anyone - otherwsie you could interact or flood a valid registered multicast application or protocol.
Class D Address usage

SAP/SDP space 224.0.0.0 - 239.0.0.0
The proper space for user or company mulicast apps is 239.192.0.0 ->239.251.255.255 . which is meant for regional or global apps....and a bit like RFC1918 for IP addresses
Also a note to take care of the 1:32 address overlap in multicast address space - and also the issues known as "address aliasing" - where L3 address aliases to a important L2 address - when aliased via the OUI - 0x01005e - which is the L2 mcast

Hi Sentinel Master,

Thanks for the info. This network is totally standalone at the moment so shouldn't be an issue but I will contact the person who "assigned" me 224.2.252.0 etc and advice them that 239.192.1.1 - 239.251.255.255 would probably be better. They told me to use the original addresses as they were stated as "unassigned - returned 29/4/03" by IANA but I take your point.

However, have also tried the new addresses with my multicast servers and still get flooding of data through the switch once all the clients have left the group!

regards
Andy

Sounds like IGMP snooping process has a bug on the 5212..........and Dell need to comment here

Other issue I have seen before is switches are unable to complete multicast forwarding table is the Multicast server is in the same VLAN (routed subnet) as the intended receivers (hosts)

Can you move the Source to a different subnet to test this theory....depends if you have a router handy ?

Just bumping this back up to the top as I still haven't had a suggestion from Dell as to what is causing this and what the solution is.

regards
Andy

p.s. Thanks for your input Sentinel Master - unfortunately I haven't got a router to try out your suggestion.