Spanning-Tree Issue

[yuchenglai]

Does anyone know if switches will go into STP learning mode during the time when the root bridge sends Configuration BPDUs with the Topology Change bit set to 1? I understand that the Topology Change process is used to age out the mac-address tables of switches but do these switches go in to the learning state instead of remaining in the forwarding state? Any advice is greatly appreciated!

[izackvail]

The switch that has the topology change will go through the listening and learning phases on that particular interface but the others will just age out their mac addresses a little faster. If they all went into the learning phase your whole network would go down everytime any switch had a topology change.

[Giuseppe Larosa]

Hello Yu-Cheng,

when a topology change happens th whole tree needs to be calculated again so yes all switches will go via the known states sequence for their ports.

It is the only safe way to work: STP tries to prevent bridging loops.

Most of the ports will come back to same state but a comparison to elect designated port on each segment is needed.

To be noted that user ports can take advantage of spanning-tree portfast: when a portfast port changes state no TC flag is set and the whole tree is not recalculated.

So use spanning-tree portfast on user ports together with spanning-tree bpduguard to protect from unwanted switches

Hope to help

Giuseppe

[yuchenglai]

Giuseppe,

This definitely helps. Are you saying that for switches running 802.1D STP and don't have portfast configured on the user ports will recalculate spanning tree topology if a users plug their devices into the access ports?

[Giuseppe Larosa]

Hello Yu-Cheng,

yes it is seen as a topology change a TCN BPDU is sent upstream towards the root bridge and a recalculation is triggered

To be noted: the same command makes the ports edge ports in 802.1W rapid STP and port classification is important to achieve real fast convergence also with rapid STP

Hope to help

Giuseppe

[izackvail]

"An important point to consider here is that a TCN does not start a STP recalculation. This fear comes from the fact that TCNs are often associated with unstable STP environments; TCNs are a consequence of this, not a cause. The TCN only has an impact on the aging time. It does not change the topology nor create a loop. "

http://www.cisco.com/en/US/tech/tk389/tk621/technologies_tech_note09186a0080094797.shtml

Take a look at this document

[yuchenglai]

izac,

I believe you are correct that TCN does not start an STP recalculation but, rather, are the consequence of recalculation. According to the STP chapters of the CCIE LAN Switching book, a switch generates TCN if it has a port that transitions to either the forwarding state or the blocking state. Can we agree that these state transitions are caused by either loss of BPDUs (past the max-age time) or introducing a switch that generates more superior BPDUs into the network?