From: Marcus Lasarko (mlasarko@baltimorecountymd.gov)
Date: Sat Mar 24 2007 - 10:47:58 ART
Greetings Peyton,
I have a few large-scale Novell/Netware/Suse deployments and deploy VMWare as well.
Many clients connect these servers to 65xx's without any MCast config.
Same for VMWare - Aside from ip helpers for clients/DHCP (outside of the local L3 subnet) there is no relative, non-default, config for BCast or MCast traffic.
What you are describing on the same switch and vlan should be all L2 - ready out of the box.
Two words that may resolve this without any additonal config (or inadvertently breaking something) on your 6513
"packet capture"
Ethereal, Sniffer, etc... Pick your poision.
That should be fairly straight-forward if being performed on a local switch.
The downside is you will have to take time to actually explain the trace to the sys admin, which could take exponentially longer that acquiring the trace itself in my experience, and don't forget the Executuve Summary for management, those take time too.
Seriously, the last couple of times I had to do this to prove some SysAdmin\Programmer did not get the whole picture I would wind up spending 2 days drafting a "whitepaper" for management explaining what the issue was and why it was "not the network" by means of tangible and invariable proof. In other words, just be prepared to draw pretty pictures and get down to the basics here, choo-choo trains, stop signs, all that drawing some of left behind with the crayons (not me, I still like crayons and color as much as I can, whenever I can, especially when I am explaining things to BA's, PM's, the programmers, and so on)
If you want more info on the 6500-specific commands unicast me or post to the Cisco-NSP list (puck.nether.net/mailman/listinfo/cisco-nsp )
Good Luck,
~M
>>> "Peyton Schouest" <pschouest@hotmail.com> 03/22/07 1:44 PM >>>
I have come across a situation at work wondering if any of the more
knowledgable people out there could shed some light on this for me.
From my understanding if a switch does not know what ports a multicast
packet should go to it will broadcast it by default. The issue at work is we
have two servers that basically have a heart beat or stay alive communiction
between them and they use a multicast address. Both servers are on the same
vlan same switch. The heart beat is set for once every 10 seconds the two
servers will send out one multicast packet. This all seems pretty straight
forward except that I have a sys admin telling me it will not work because
cisco switches will only forward 3 packets and then it quits forwarding
multicast by default which I have never in my life heard of. Because of this
"feature" he states that we must enable multicast routing and pim on the
vlan interface for the servers heart beat to work. I have never heard of any
so called feature or bug before concerning a cisco switch. Has anyone dealt
with something like this before or able to stregnthen an argument from
either side? Thank you for your help before hand.
Peyton
This archive was generated by hypermail 2.1.4 : Sun Apr 01 2007 - 06:35:52 ART