cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
1061
Views
0
Helpful
4
Replies

Cisco Switches dropping fragmented Multicast Traffic over L3 Links

sipoppcco
Level 1
Level 1

Hi,

is anyone there facing the following Problem.

I have the problem that different Cisco MLS switches (3650E,3750) are dropping multicast traffic over L3 links when they are connected over a gigabit(fiber or cooper) and the multicast traffic have a size > 32000 Bytes so the udp packets have to be fragmented. To rebuild the problem you need two Layer3 switches (ipbase or advance) connected over a L3 Link and a multicast sender and receiver (for exampel wsend,wlisten). Each in different networks so the traffic have to be routed over the transfer L3 Link and then in the according vlan where the receiver is connected. If the Sender or the link between the switches are 100 Mbit no problem occurs. But if both links are 1Gig and the payload is >30000Byte udp packets are dropped on the receiver side. There is no counter showing this drops on the switch but you can see it in the wlisten tool or if you trace the packets. I have tested it with different configuration of multicast. Using dense or sparse, pim v1 or v2 the problem still exists. The only solution is to connect the sender with 100Mbit or reduce the Uplink between the switches to 100Mbit. But this could not be the solution.

4 Replies 4

Jeff Horton
Level 1
Level 1

I am getting somewhat of the same issue.. Running IOS XE 16.06.04 on the 3850.. Links on my setup are L2 fiber links... Multicast video streams just fine on a reboot of the 3850 switch stack but gradually starts dropping and day goes on.. I have one 3850 switch stack that consist of 1 3850-24P-S and 1 3850-12S-S... I don't know if the stack cables are playing any part in the issue or not. We have four stations that stream 10 multicast camera feeds from each station and each station has a 2960x that is connected to the switchstack via fiber... Never had this issue when running IOS XE version 3.7.5... 

1. Does anyone think the default CoPP could be dropping the video?

2. How much longer is the 3.7.* version going to be supported/developed?

3. Explain the differences in Fuji, Everest and Denali version....

 

 

Hello

Have you check the similarities between old/new ios-xe versions also for any bugs the new version may contain?

What pim mode are you running? Do you have gimp snooping enabled on the switches ?

 

Fuji, Everest and Denali


Please rate and mark as an accepted solution if you have found any of the information provided useful.
This then could assist others on these forums to find a valuable answer and broadens the community’s global network.

Kind Regards
Paul

I currently have ip pim sparse-dense-mode turned on on  all the VLAN interfaces....

 IGMP snooping is enabled....

 

From what I have read so far is that the recommendation is to run Denali 16.3.* version as it is Cisco's recommended version from a step up from 3.7.5...

 

Going to give this a try and see how this works... Most issues are where I have no access to switches so I have to get other personnel the IOS XE and instructions on how to downgrade to this..

 

I will update this when it is determined either the fix works or not....

Well disabling the system-cpp-police-mulitcast policer has helped.. The video is still dropping on one or two now not 8 or more.... I think there is still some information that I am not getting from the field as to how the links between systems are built.. But for now this is working...

Getting Started

Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community: