Possible Spanning-Tree Issue

[robertotron007]

Hey Guys, I hope someone with better spanning tree knowledge can help me out here. I will show you the picture of the network first and this medium size network is mostly in VLAN 1.

The first thing that I was triggered to is there is excessive unicast broadcast packets. I noticed that some of our wireless APs were rebroadcasting every packet for about 30 seconds or so. "The wireless controller is off of Core B". After more wireshark captures, I can tell these broadcasts were only happening within the RPVST domain and not the PVST domain.

What is triggering these broadcasts are link up and link down. When we have a link that is flapping excessively, the broadcasts storm is large. Again, all in the RPVST domain. What I know I need to do is enable the access ports on all of the switches to host ports (access and portfast). This will stop the STP topology changes and the MAC table is not flushed.

However, I still want to know why the storm only happens in the RPVST domain and not the PVST domain. The Cores and SW 1-3 all support RPVST. Core B is the root switch for vlan 1.

Up until today I have been noticing this issue only on the distribution switches and switches 4-10. Now I think I have isolated it to PVST and RPVST domains because that is the only thing I can tell that could be a problem at this point. But I could be wrong.

Your input is greatly appreciated!

[Reza Sharifi]

Hi,

Since you don't have a link between core-a and b, there is no physical loop.  If you have a link between core-a and b, to further isolate the issue, I would migrate all the switches to RPVST, and keep an eye on the them for a period of time to see if it happens again. Migrating to the same protocol makes management and troubleshooting easier.

HTH

[robertotron007]

My bad Reza, i just drew that up quickly. There is a connection between the two Cores.

That is what I was thinking too, upgrading to RPVST on all the other switches. Do you have any explanation on why I would only see the unicast broadcasting on only the RPVST switches and not the PVST switches?

Thanks

[glen.grant]

  I dont see any reason ,seeing that everyone is in the same broadcast domain which is vlan 1 .  If all access ports do not not have portfast turned on for all switches then you might get excessive broadcast everytime a port goes up and down .   Is the root for vlan 1 hardset on one of the core switches? If not it should be , one should be primary root and the other should be secondary.

[robertotron007]

Yes, the root for vlan 1 is hardset. If the switch fails it will failover to CoreA.

Are there any issues with the Root Switch being PVST while switches below it are using RPVST?

[robertotron007]

I am very curious to know if there are any issues with the root switch running an older version of spanning-tree.

thx

[ALIAOF_]

Correct me if you have rpvst and pvst running then each mode will elect a separate root bridge.  So like mentioned above migrating to rpvst so it is the same accross the whole network might be a good idea.

[robertotron007]

Yeah, I am going to do that. Hopefully that works out.