I am attempting to gain a better understanding of how STP converges when certain state changes occur that affect the STP topology. In this scenario (see below) Switch A (SwA) is the root switch with Switch B (SwB) and (SwC) connected by a single link both to SwA and to one another (so a mesh topology). In this case, the root ports for both SwB and SwC or those ports directly connected to SwA. Also in this case, SwB wins the designate port election based on the fact that it has the lower BID. Thus, it configures its port on the link that connects to SwC as a DP in a forwarding state with SwC configuring its corresponding port in state blocking.

enter image description here

My question is this: What exactly happens if SwB initial root fails (see below) and it now needs to select a new root? I understand what occurs in SwB's case. It will go through the STP process again and ultimately settle on using the port currently connected to SwC as its new root (formerly a designated port). What I don't understand it what happens on SwC since it has configured the port on itself that connects to SwB in a blocking state. My CCNA books states that a port in a blocking state receives Hello BPDUs, but discards them and does not process them. The book also states that a port in a blocking state will not send out Hello BPDUs. How, therefore, does SwC know that it now needs to configure this port as a designated port since, on the corresponding side, SwB will be using this link as its root? Would it be because it would cease receiving Hello BPDUs on this port and, after the expiration of the Max Age Timer, begin the process of changing the STP topology?

enter image description here

  • 1
    My CCNA books states that a port in a blocking state receives Hello BPDUs, but discards them and does not process them. The book also states that a port in a blocking state will not send out Hello BPDUs. Burn that book. – Ron Trunk Nov 27 '19 at 19:01
  • Haha. I think the book actually has it right now that I'm looking again. I just wasn't fully understanding what I was reading. – ahelton Nov 27 '19 at 20:19

Do not confuse blocking the VLAN frames with blocking BPDUs. A port in the blocking state is still enabled, and it will drop frames for the VLAN it is blocking, but BPDUs are not blocked because they use a special link-only multicast destination MAC address. A Cisco switch port could be a trunk using PVST, and the port could be blocking for some of the VLANs, but not other VLANs, because it is per VLAN.

The root bridge (the other end of the down link) will send a topology change notification, so SwC will go through the full reconvergence, and it will go through listen/learn on the formerly blocked port.

STP is constantly sending and receiving BPDUs so that it will notice whenever there is a topology change, and the root bridge will send out the TCN to all switches when it is informed of a change, and all the switches will go through the full convergence to determine a new topology.

| improve this answer | |
  • Ok. That's exactly where I was getting confused. I was misinterpreting what my book meant by blocking thinking that it meant it blocked everything. Thank you. – ahelton Nov 27 '19 at 20:18
  • The point of STP is to prevent loops. BPDUs, being link-only, never get sent beyond the link, so they will not create a loop. There are also other link-only protocols. – Ron Maupin Nov 27 '19 at 20:25
  • Gotcha. My CCNA book is good, but I think it skips over some stuff (I'd never heard of a TCN until you mentioned it). It actually doesn't really get into the fact that a notification is sent from the root to all switches when a topology change occurs. It talks about a single switch making changes, as in the case of SwB in my diagram, but that appears to be about as far as it goes on the topic. It kind of left me with the impression that if a switch wasn't impacted by a topology change, then it didn't need to go through convergence again. – ahelton Nov 27 '19 at 20:53
  • Well, even without a TCN, SwB will no longer have superior BPDUs to SwC, so SwB and SwC could make the change on their own. This is a simple case. The TCN is for when you have complex switch topologies because it can inform switches that would not see a change that there is one and they should reconverge because the root and designated ports on them may need to change based on changes in other parts of the switch topology. In other words, it helps coordinate an overall convergence. – Ron Maupin Nov 28 '19 at 1:54
  • Okay. So TCN isn't necessarily required in this case? Is there a way to observe the contents of BPDUs so I can observe all this in action? Wireshark maybe? – ahelton Nov 28 '19 at 2:50

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.