Re: Multicast forwarding loop at mbone-cern.switch.ch

[Bill Fenner <]

The problem definitely seems to come and go, so it's perfectly reasonable
for you to not see it right now.  I'm working on extending my duplicate
detector to keep history and make it available via the web.

The multicast forwarding loops could easily explain your mrouter
freezes.  The kernel gets too busy forwarding traffic around the loop
and can't respond to anything else.  The loop is probably
mbone-cern->ch-ws->ams-mr->mbone-cern.  The fact that the
ams-mr->mbone-cern link is a backup is irrelevant due to the flooding
problem.

>o patching the Solaris kernels: we use 2.6 everywhere. Is there a
>  patch available somewhere ??? 

Contact Sangeeta.Mukherjee@localhost she made the patch for 2.5.1 .

>o we could temporarily remove the redundant tunnels if this might help

Well, it looks like the only redundant tunnel that could explain what
I was seeing is the ams-mr one.  Since it's running 11.3(2), it's supposed
to have this problem fixed.  You could try removing it, but at the same
time, we're not seeing the problem now so we couldn't tell if it did any
good.

>o what impact does this problem have on the global Mbone (How did you
>  get aware of the problem ?) 

Well, many of the duplicates get dropped in the loop or at the next hop,
since it's such a huge burst of traffic.  But the global MBone does see
some duplicates - I saw about 50 copies of each packet from the source
that was having problems the other day.

  Bill