rapid-pvst+ portfast bpduguard still going to learning state

barberousse · ‎10-04-2012

Hello,

Using packet tracer (v5.3.3.0019) to learn for campus area network (core, distribution and access layer) I simulate topology change to see spanning tree behavior and convergence time.

Only one vlan is set, the default vlan1.

Core switch are two 3560-24PS, distribution four 2960-24TT and five access as 2960-24TT.

Core0 is configuration as root with spanning-tree vlan 1 root primary.

And core1 as secondary.

Access switch port where client are connected is configured with portfast and pbduguard :

Here configuration for one access switch. Port fa0/1 goes to client, fa0/2 for distribution2 and fa0/3 for distribution1

hostname access1

!

spanning-tree mode rapid-pvst

!

interface FastEthernet0/1

switchport mode access

spanning-tree portfast

spanning-tree link-type point-to-point

spanning-tree bpduguard enable

!

interface FastEthernet0/2

spanning-tree link-type point-to-point

!

interface FastEthernet0/3

spanning-tree link-type point-to-point

!

Now, when convergence is done all client can ping each other.

If I shutdown a client and start up later portfast work as expected, port goes from down to forwarding.

If I shutdown a link between the spanning tree root (core0) to a distribution switch (which is connected to access switch that forward client ping test) it activate the backup" link vry fast and ping does not time out. When return the link to up state, distribution switch goes to the "primary" link (to root switch) with, again, no ping time out.

When I shutdown the link between the distribution2 and access1 switch, access1 send traffic to it's backup link to distribution1 and no ping time out.

The problem is here, when this link goes to up access1 re-asign traffic to distribution2 (prefered root path). Ok but the access port to the client goes into block mode for a while, then learning and then forwarding. So traffic is interrupted and ping fail. It make arround 20 seconds !!!

So is it expected behavior ? or the edge port must still in forwarding mode until bpdu (loop) are recived on that port or it goes down.

Thank, ask me question if you don't understand my (french-english) ...

Best regards,

Antonio Knox · ‎10-04-2012

Fabrice,

Remove this from the client port config:

spanning-tree link-type point-to-point

***Please rate helpful posts ***

barberousse · ‎10-06-2012

Hello,

Thank you for your relply Antonio.

But unfortunatly access1 switch still lear on edge port.

According to this document http://www.cisco.com/en/US/docs/switches/datacenter/nexus5000/sw/configuration/guide/cli/RPVSpanningTree.html#wp1205316, it should not. Or am I wrong ?

Here access1 config :

Current configuration : 1287 bytes

!

version 12.2

no service timestamps log datetime msec

no service timestamps debug datetime msec

no service password-encryption

!

hostname access1

!

spanning-tree mode rapid-pvst

spanning-tree portfast default

!

interface FastEthernet0/1

switchport mode access

spanning-tree portfast

spanning-tree bpduguard enable

!

interface FastEthernet0/2

spanning-tree link-type point-to-point

!

interface FastEthernet0/3

spanning-tree link-type point-to-point

!

...

!

interface GigabitEthernet1/2

!

interface Vlan1

no ip address

shutdown

!

line con 0

!

line vty 0 4

login

line vty 5 15

login

!

end

And here DIST 2 (dist1 has the same config)

Current configuration : 1269 bytes

!

version 12.2

no service timestamps log datetime msec

no service timestamps debug datetime msec

no service password-encryption

!

hostname dist1

!

spanning-tree mode rapid-pvst

spanning-tree portfast default

!

interface FastEthernet0/1

spanning-tree link-type point-to-point

!

interface FastEthernet0/2

spanning-tree link-type point-to-point

!

interface FastEthernet0/3

spanning-tree link-type point-to-point

!

interface FastEthernet0/4

spanning-tree link-type point-to-point

!

interface FastEthernet0/5

spanning-tree link-type point-to-point

!

...

interface GigabitEthernet1/2

!

interface Vlan1

no ip address

shutdown

!

line con 0

!

line vty 0 4

login

line vty 5 15

login

!

end

Peter Paluch · ‎10-06-2012

Hello Fabrice,

I am afraid that the Packet Tracer has its limitations in simulating more complex topologies. The STP behavior in Packet Tracer is known to be buggy. As much as I admire and respect the Packet Tracer authors, I do not recommend relying on Packet Tracer to learn about STP.

The problem is here, when this link goes to up access1 re-asign traffic  to distribution2 (prefered root path). Ok but the access port to the  client goes into block mode for a while, then learning and then  forwarding. So traffic is interrupted and ping fail. It make arround 20  seconds !!!

Well, this would suggest, at least in theory, that the access port towards the client is not operating as an edge port. When a new root port is discovered, the switch puts all non-edge designated ports into Discarding state (the result of receiving a Proposal BPDU). These ports in turn start sending proposals themselves. If this happens to an access port towards a PC, though, then the PC will be cut off from the network for 30 seconds as it is unable to respond with an agreement, and the access port must go over the Discarding and Learning states using timers.

This blocking does not apply to edge ports. Edge ports are not influenced by proposal/agreement mechanism.

However, your configuration suggests that you have the PortFast configured on the port towards the client. Therefore, this port should not be affected by the proposal/agreement. If it is then it either received a BPDU from somewhere, or the Packet Tracer does not simulate the RSTP behavior correctly. In a true network, an edge port would never be put into Discarding state just because a new root port is determined.

Best regards,

Peter

barberousse · ‎10-09-2012

Hello Peter,

Thank you for your explanation.

I have got some doubt if Packet Tracer get full fonctionnality implementation.

As internal-BGP is not supported too.

Yes it is a great tool and I have learned lot of things with this. Thank to them.

I have got a 2950 which support rapid-pvst+ but with one switch it's difficult to test this.

I have also got a 5500series and a 2900xl which don't support rapid-pvst+

Best regards,