KNXnet IP and multicast on switches

[Fazil Haneefa]

Hi Everyone,

Recently I was called to troubleshoot an issue concerning the lighting control team who were missing random multicast packets across the network.

The multicast source - the lighting control server multicasts scheduled commands through the LAN to the receivers and it uses Knxnet IP protocol.

So, recently they replaced the old core switch with a new 3850 series switch and from then onwards they noticed random receivers are not receiving the commands. But when they switch back to the old core things work fine.

Now when I compared both the Old and new core configurations the only differences I saw was  on the old core though igmp snooping was enabled no pim configurations were on this particular VLAN.

But in the new switch they had configured ip multicast-routing and enabled ip pim sparse-dense mode on the SVI. Also the subnet on the old switch was /16 and the new switch is /24.

Now when I ping the group 224.1.23.12 which is the default knxnet ip group, all the receivers gives me a reply to the ping, which I suppose that the multicast configurations are OK.

Is there any specific pim mode i should use when considering the multicast config for knx or shall i remove the multicast routing on the new core switch and make it similar to the old core.

If anyone had a similar experience or have any input, it will be much appreciated. 

Thank you,

[Jon Marshall]

I don't have any experience with the specific application but from your description I would not recommend removing the multicast routing without some checks first.

It sounds like with the old core switch you had a flat /16 network so there was no need for any multicast routing but on the new switch it is now a /24 which suggests they have used vlans to create smaller broadcast domains. If they have then they would need to use multicast routing for the multicast to travel between vlans.

So are the receivers in different vlans/IP subnets than the server ?

In terms of which PIM version to use it should not really make any difference if all the vlans are routed on the core switch.

Is there are pattern to the receivers that cannot receive the stream in terms of vlan/IP subnet for example or is it truly random ?

Jon

[Fazil Haneefa]

I will try it without multicast and by making it /16.

But i dont understand the statement /16 help with a broadcast and /24 doesn't, though the number of devices who are joining the group is not more than 54.

It seems to be truly random. When the scheduler runs some lighting devices at a particular area wont receive the signal but some other day those devices receives though.