Hi,

 

I am working on a project to implement Creston NVXs, Q-Sys etc, I work on the network side so don't 100% understand all of it. We are having an issue where ports connected to NVX-D30 at random point during the day and night get flooded with traffic, overwhelming port causing traffic to drop therefore the display flick. We can go hours with out issue, there no real pattern to when and why it occurs. Network consists C9300 connected to C9500 core where the igmp querier is. This AV kit sits on its own vlan. Snooping and immediate leave are enabled. No matter how much QoS I put on the put it still falls over. Has anyone see this kind of this before? or has some example config in which might help? I am lost on what it could be. Any help would much appreciated.

Happy to share more logs and config.

Sounds like something we were kinda running into on some Aruba switches too. I don't know what the Cisco equivalent is, but it could be related to this:

https://www.youtube.com/watch?v=BqlGlY8GjNM

So two things...

L2 Multicast will always go to the querier, best practice in simple configurations IME is to have the querier on the switch hosting the majority of encoders and you may need to make sure the uplinks are sized accordingly...

Second you need to disable flooding on a STP topology change notification (TCN). I'm sitting at a hotel restaurant right now but IIRC it's" no span tcn flood" applied to every interface.

If I weren't on mobile I'd give the dissertation for why this is the default but my thumbs are tired 😁

Get Outlook for Android

Thanks Both! Appreciate the help!

 

 

 

I have "no igmp snooping tcn flood" applied to all the access ports for devices on this vlan, I have a few other devices on other vlans on this switch should I apply it here as well and o I need it on the uplinks as well?

 

Uplink bandwidth seems fine no drops or anything there, issue is with the access ports where these devices connect just seem to receive a burst of traffic at random times filling up the buffers and dropping the packets but could go hours and hours without issue.

 

Did you update the firmware? Older ones had issue with NO TCN FLOOD commands not working properly. Although that will require a topology change, such one of the switches dropping and reconnecting. 

Hi!   I had a similar circumstance.   It turned out that I needed to turn of spanning tree protocol on all NVX ports and the problem was gone.

Hi All,

 

Appreciate all the responses and help. Just thought I would close this off as it seems we have solved the issue. On Friday I did a span of one of the problem ports, analysing it in wireshark I could see around the time the drops occurred I could see the port receiving multicast traffic from sources it shouldn't be. I correlated that this happened when spanning tree reconverging which is strange as we were plugging in any switches or making changes here. However we did have no ip igmp snooping tcn flood enabled on the port but seems it wasn't working - seems we hit this bug Cisco Bug: CSCvy88262 - Multicast is flooded when "no ip igmp snooping tcn flood" is configured even though the version we were running is not listed as effected it must have been. Upgraded to 17.12.04 and the issues seem to be resolved and also our spanning tree is no longer constantly reconverging, so perhaps we hit another underlying bug there that is now resolved.

 

Once again thanks for the help!