HPE Aruba Networking & ProVision-based
1827473 Members
1681 Online
109965 Solutions
New Discussion

E3500yl high port utilization despite igmp enabled

 
Tuncay G
Occasional Contributor

E3500yl high port utilization despite igmp enabled

Hi, I have a question regarding the igmp configuration on the 3500yl switch. Although igmp is enabled on the switch, looking at the port utilization it seems like all the multicast traffic is forwarded to all the ports. This core switch feeds three other HP switches (E2610) which are connected to ports 21-23 of the E3500 switch. Please see the attached image for the port utilization figures.

 

There are 4 multicast sources connected to ports 1-4. What I notice is, although these ports are supposed to only receive traffic, they are also transmitting very intensely. The traffic on ports 21-23 is also very high, considering the traffic on the individual ports of the E2610 switches. 

 

Looking at the "show ip igmp" command output (also attached), groups joined seem consistent with successful igmp operation. However I still cannot justify the high port utilization.

 

Something I noticed, after all the 4 switches are rebooted, the port utilization displays normal values for some time (usually 1-2 days). I can't think of a change that will cause this effect.

 

I can give more information about the configuration if requested.

 

I am looking for help from the more experienced people here in this forum.

 

Thanks,

Tuncay

1 REPLY 1
chomezski
Advisor

Re: E3500yl high port utilization despite igmp enabled

Hi Tuncay,

 

Not sure if you have already resolved your issue. One thing I noticed regarding your configuration is that your queurier is not your core switch (e3500zl) but one of your access layer switches (2610) with IP 10.1.1.100. Not sure if this was planned or not but I am guessing that because it had the lower IP it won as the Querier.

My problem is similar. I have 2 5400zl in the distribution layer. Primary VRRP is the queurier. All the access switches (2610) are configured with IGMP. Even though there are not subscribers to join the multicast group on the access ports it appears like the multicast is being broadcast to all ports.

I don't understand why this multicast is not being pruned...

Attached relevant info.

Any help would be greatly appreciated.