Prev: Interface name vs interface IP address
Next: cisco 831 - won't obtain IP address
From: thecompnerd on 23 Mar 2010 00:33
We're using MS Network Load Balancing for 2 groups of 2 Win2003 TS
servers. Both groups are configured for L2 multicast (no multicast IP
being used) with IGMP enabled. Both groups of servers are on the same
VLAN and subnet as our other servers and users. It appears as though
the multicast packets are flooding the network as I can be on the far
side of the network and receive L2 multicasts intended for those
servers.

MS's documentation states that multicast pruning (IGMP??) is not
enough to keep the packets from flooding the networking, although I
thought that was the whole point of IGMP snooping?

Our switches all support IGMP snooping and have it enabled, although
no other configuration has been applied. I'm curious if I'm
misunderstanding how IGMP snooping works? Can anyone provide some
insight into how multicast is working in this scenario and why I'm not
getting the expected result - multicast session joining by
participating hosts?
From: Martin Gallagher on 23 Mar 2010 05:30
thecompnerd wrote:

> We're using MS Network Load Balancing for 2 groups of 2 Win2003 TS
> servers.

Well there's your mistake then.

Have a look at:

Catalyst Switches for Microsoft Network Load Balancing Configuration Example
http://www.cisco.com/en/US/products/hw/switches/ps708/products_configuration_example09186a0080a07203.shtml

Maybe that will help.
From: thecompnerd on 23 Mar 2010 14:33
>  Well there's your mistake then.
>
> Have a look at:
>
> Catalyst Switches for Microsoft Network Load Balancing Configuration Examplehttp://www.cisco.com/en/US/products/hw/switches/ps708/products_config...
>
>  Maybe that will help.

I've consulted the document before, but still don't have the answers I
was hoping for. Shouldn't IGMP keep the switch from flooding all
switch ports? Otherwise, what's the point?

Back to the scenario... it appears as though the way we should go with
this is to put the Cluster TS's in a VLAN according to Cisco?

>However, since the incoming packets have a unicast destination IP address and multicast destination MAC the Cisco device ignores this entry and process-switches each cluster-bound packets. In order to avoid this process >switching, insert a static mac-address-table entry as given below in order to switch cluster-bound packets in hardware.

>mac-address-table static 0300.5e11.1111 vlan 200 interface fa2/3 fa2/4"

I'm not familiar with what's going on in this command. Is it mapping
the multicast address statically and associating it as part of VLAN
200 on ports 2,3, and 4? If so, will this same entry need to be
applied to each switch trunk in order to keep the packet from being
flooded out the client ports intended for the cluster?

Thanks.
From: Martin Gallagher on 24 Mar 2010 06:21
thecompnerd wrote:

>> Well there's your mistake then.
>>
>> Have a look at:
>>
>> Catalyst Switches for Microsoft Network Load Balancing Configuration
>>
Examplehttp://www.cisco.com/en/US/products/hw/switches/ps708/products_config...
>>
>> Maybe that will help.
>
> I've consulted the document before, but still don't have the answers I
> was hoping for. Shouldn't IGMP keep the switch from flooding all
> switch ports?

Do you have one switch or several? If more than one switch and only one
vlan you might also need to look at:

Multicast Does Not Work in the Same VLAN in Catalyst Switches
https://www.cisco.com/en/US/products/hw/switches/ps708/products_tech_note09186a008059a9df.shtml

If you have multiple switches, even if the MS boxes are sending IGMP
membership reports for their bogus multicast group, the other switches
won't see the IGMP.

 | 
Pages: 1
Prev: Interface name vs interface IP address
Next: cisco 831 - won't obtain IP address