cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
1678
Views
2
Helpful
11
Replies

c9500 Layer 2 multicast replication issue.

Larry Sullivan
Level 3
Level 3

Things to consider:

1.) I have a TAC open for over a month now with no real movement on resolving the issue.

2.) The description of the issue is short because I've spent countless hours looking into this and narrowing down the issue.

We are running into an issue with layer 2 multicast at one of our co-location facilities.  We just installed these c9500 switches earlier this year.  Essentially whenever a multicast video stream needs to be replicated to more than one egress port on either of our two c9500s, we get continuity count errors on the stream.  These errors indicate either dropped or out of order packets which causes the video stream to break up.  At most we can get several errors every few minutes.  They come in groups of about 5 at a time.  We don’t see the issue with unicast video or a multicast video stream with only one subscriber port on the LANIssue only occurs if multicast stream has to be replicated ot multiple ports.

Topology is primary router (mrouter) ASR 1k, Secondary router (querier) ASR 1k, primary and secondary switch are c9500s with port-channel between them.  Most traffic including streams tested and verified dropping are coming in via the primary router.

The there are no corresponding interface errors or control-plane policy-map drops.  "show controller cpu-interface" does have incrementing "SW Forwarding" drops, but this same category does not drop on the control-plane policy-map.

I'm convinced we are running into a limitation of multicast replication throughput or this is a weird unknown bug.  the uplink to the primary router which has most the traffic on it has around 3 Gb UL and DL on it.  25 Gb SFP port with 10 Gb SFP.

CPU usage is low, around 10%.  I already tried adjusting CoPP policy pps for various classes including SW Forwarding.

11 Replies 11

L1anYu
Level 1
Level 1

I’ve encountered the same issue when using Catalyst 9500 switches, with the exact same continuity errors and pixelation problems you described. However, when I use Nexus 9504, the issue doesn’t occur. Could you share how you eventually resolved this?

We actually haven't resolved this even after opening a TAC case back when I first posted this.  It kind of cleared up on it's own for awhile until recently it reared it's ugly head again.  I'm thinking an upgrade is best next bet to try but we have so much sensitive production traffic on these that it's not easy to do.  What image version are you on?

Hello
I assume you have snooping enabled and checked the MTU in path?


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

Enabling snooping could potentially cause the issue mentioned by the original poster? I’m experiencing the exact same problem as they described.

Hello
disagree - Why would it, without it mc will be flooded to every port on the same lan so snooping will actually decrease the traffic the ISE has to switch 


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

L1anYu_0-1726043189871.png

The issue I’m facing is that every 5 minutes, there are 4 packets with CC errors, and my client has informed me that their monitoring system has detected these error packets. Additionally, the video is experiencing frame drops. The errors seem to be caused by missing frames at the application layer. However, when I use monitoring software to check both incoming and outgoing traffic, I see no packet loss. This issue does not occur when using N9K devices.

Hello
As i have stated , I very much doubt snooping will make things worse, this suggests software to me , may even a bug?-  Have you tried a downgrade/upgrade?


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 am not able to upgrade or downgrade.so I haven't tried it yet.

Correct.  IGMP snooping enabled and MTU in path default 1500.

Hello Larry 
So before you swapped to these new switches  you had no issues with the mc flow correct?
Have you tried increasing the mtu inpath just as a test ?


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

No related issue prior.  Since installing these we have had a substantial increase in MC traffic and services compared to previous equipment.  We have not tried increasing MTU.