cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
280
Views
0
Helpful
2
Replies

Multicast groups drop from switch - GigE

K.Garretson
Level 1
Level 1

Just upgraded from FDDI to Gigabit E.

Dual 6500 switches with Cat OS.

Dual 7500 Routers with dual GigE cards port-channelled.

Mixed FastE (using Sun QFE cards) & GigE (SysKonnect Fiber Gig) environment

Sun environment - E250's, Ultra 60's & some Ultra2's, a couple of PC's.

Application is multicast driven.

Bringing the lab up from cold start, all devices properly register for multicast. After about 5 minutes, some of the GigE devices time out (Ultra 60's all time out, some the of the E250's time out). None of the QFE devices time out.

Firing off a snoop on a host brings the connection back up, but it will time out once the snoop stops.

We are about to sniff connections via a Sniffer in pass-through mode to evaluate what's happening with IGMP traffic. We have brought the lab into single string (single router, single switch) configuration for troubleshooting purposes.

We are looking most closely at the registration process between host & switch, but nothing conclusive yet. Anyone bumped into anything similar?

Thanks!

2 Replies 2

dwallwork
Level 1
Level 1

This great Cisco article below details specific issues and an example of how one may encounter issues with PIM and multicast not propagating to partner routers because of HSRP configuration. Maybe it does not have anything to do with your issue, but just in case, here it is ... wild guess....

http://www.cisco.com/en/US/tech/tk648/tk363/technologies_tech_note09186a0080094aab.shtml

Good luck.

PS... even if it's not your issue, the PIM Multicast diagnistics are pretty helpful and well documented.

Thanks for the URL.

We were able to get the sniffer up in pass-through mode Friday afternoon.

The sniffer indicated that the IGMP joins are being sent properly from the switch, so it looks like the NICs are not responding correctly to joins. We have the card vendor looking at the problem.

BTW - we use MSDP for multicast distribution & HSRP for router failover and have had good luck with stream stability during regular ops and during failover. (At least in the FDDI environment - clearly we have some testing ahead of us!)

Multicast is ALWAYS interesting!