[AG-TECH] Weird problem mk.II

[Robert Olson]

Ah, good points. Another good piece of information to include in any 
descriptions of multicast problems is the local network architecture - is 
it a big flat switched net? small routed net? etc. We've got persistent 
problems in our huge flat network here that we weren't able to resolve, so 
we put the multicast-requiring hosts on a smaller network that doesn't have 
the multiple switches involved.

--bob

At 05:43 PM 10/21/2003 +0300, Harri Salminen wrote:

>I've seen something like that where some groups work and others don't...
>In addition to routing, I'd check the IGMP snooping on local
>ethernet switches, some do have bugs and some even have a shared MAC table
>for all VLANs on the switch meaning multicasts leak through and mess up
>routing or IGMP... Can be hard to debug even live, try turning off all IGMP
>snooping if it's there and get the firmware revisions OK. Of course it could
>be routing problem too since the multicast tree is individually for
>each group and it might vary according to number of participants or maybe 
>even traffic volumes. But especially if IGMP messages are lost, the 
>routing will not be set up...
>
>Harri