Yet another STP Question.

Netstudent

OKay so here in the ICND book in the STP chapter there is a little section where it talks about TCN frames and flushing old MAC entries. It actually explaining what happens when a port goes down and a topology change occurs.



In the book it says "When switch2 recieves the TCN, it decides to timeout all MAC table entries based on the max age timer. (default 20seconds)


I dont understand that part. The maxage timer is the amount of time a switch must wait before it decides to change the topology from lack of hearing BPDU's.

In this example switch2 is not the root. So does this mean thats switch 2 will not add entries into the CAM table for 20seconds?

Also in this scenario switch1 is root, so does this mean that all switches will use the root's max age timer? So if switch 3 sends a TCN to switch2 because switch 3's port went down, switch 2 will timeout it's mac entries based on switch1's max age timer?

Someone who is a master of STP please help. ARGG!

dtlokee

Well think about what would happen if the MAC table stayed the same and the root/designated ports changed. If Fa0/1 was the root port and had all the MAC addresses associated with it, then a TCN was received and after listening/learning fa0/10 is now the root but all the MAC addresses are still associated with Fa0/1, the network is not going to forward frames. Normally the dynameic MAC table entries will be in the mac address table for 5 minutes, all they're saying is when a TCN is received, it accelerates the process of aging out the mac address entries for the affected VLANS.

In RSTP the againg process is instant, as soon as a TCN is received it ages out all mac entries associated with the affcted VLANS

EdTheLad

Netstudent wrote:

OKay so here in the ICND book in the STP chapter there is a little section where it talks about TCN frames and flushing old MAC entries. It actually explaining what happens when a port goes down and a topology change occurs.



In the book it says "When switch2 recieves the TCN, it decides to timeout all MAC table entries based on the max age timer. (default 20seconds)


I dont understand that part. The maxage timer is the amount of time a switch must wait before it decides to change the topology from lack of hearing BPDU's.

In this example switch2 is not the root. So does this mean thats switch 2 will not add entries into the CAM table for 20seconds?

Also in this scenario switch1 is root, so does this mean that all switches will use the root's max age timer? So if switch 3 sends a TCN to switch2 because switch 3's port went down, switch 2 will timeout it's mac entries based on switch1's max age timer?

Someone who is a master of STP please help. ARGG!

The book is either wrong or you have misunderstood what is written.When a topology change occurs TCNs are sent to the root bridge.When the root receives a TCN it will send bpdus with the TCA bit set, this bit will be set in bdpus sent by the root bridge for the MaxAge + Fw delay i.e. by default 35sec, when a downstream switch recieves this BPDU with TCA bit set, it will modify its mac table age timer from 300sec to Fw delay i.e. default 15 sec.

Netstudent

dtlokee wrote:

Well think about what would happen if the MAC table stayed the same and the root/designated ports changed. If Fa0/1 was the root port and had all the MAC addresses associated with it, then a TCN was received and after listening/learning fa0/10 is now the root but all the MAC addresses are still associated with Fa0/1, the network is not going to forward frames. Normally the dynameic MAC table entries will be in the mac address table for 5 minutes, all they're saying is when a TCN is received, it accelerates the process of aging out the mac address entries for the affected VLANS.

In RSTP the againg process is instant, as soon as a TCN is received it ages out all mac entries associated with the affcted VLANS

I understand that the MAC entries have to be flushed and changed. I understand that the MAC adresses must be relative to the forwarding topology. I realize that. I'm just not seeing the correlation between the maxage timer and a switch timing out it's MAC entries after a topology change BASED ON THE MAXAGE timer on the root bridge. I learned from a test question that the max age timer is set by the root bridge and all non-root bridges use the root's maxage and hello values.

Ed,

I doubt the book is wrong. I probably just didn't explain the scenario in it's entirety. In this scenario there are 3 switches creating one loop. One side of one of the links automatically blocks to break the loop. Then the author goes on to explain what would happen if one of the forwarding ports on one of the non-root bridges were to fail. Then that switch that failed sends out a TCN to the root bridge.


My main question is why does a switch timout it's mac entries based on the maxage timer. Not why do switches timeout their mac entries. Where is the correlation between the maxage value and timing out mac entries?

dtlokee

First, switches do not age out their mac addresses based on the maxage timer when there's a topology change, it uses the forward delay value (15 s by default). The reason is so the entries in the mac address table can remain consistent with the forwarding topology. If a switch did not accelerate it's aging of mac table entries when a topology change occurs, it would be possible to have mac addresses associated with a port that was designated/root but is now in a blocking state. If they stayed in the mac table for 300s (5 minutes) the frames would be forwarded to an interface that is discarding them until the normal process of aging removes the entries. If the question has to do with why they did it, it makes sense to use the value that is the same as the learning state of STP transistion to forwarding I would guess, but I have never really thought about it, just accepted that it is what it is.


http://www.cisco.com/en/US/docs/switches/lan/catalyst2950/software/release/12.1_22_ea5/configuration/guide/swstp.html#wp1150122

Netstudent

I think I am just thinking too hard about it. Those were my thoughts exactly, that it would be the fwd delay time.

Maybe what the author was trying to say is that when a link on switch3 goes down(max age expired), switch3 sends out a TCN, and as soon as switch2 sees that TCN on it's way to the root switch(switch1), then switch2 will begin the fwd delay process of clearing out old MAC entries.

So maybe what he ment by switch2 timing out its mac entries based on the max age timer was:

As soon as the maxage timer expires on switch3, switch3 goes into the listening state and sends out a TCN, so that switch2 will have timed out it's MAC entries so that when switch3 begins fwding again, switch2 will have relearned the correct MAC entries.

I hope thats comprehendable.

The topology is somthing like this


Switch1

---

Switch2



switch3

switch1 is root. switch2's link to switch3 is the deignated port. Switch3 is blocking between 2 and 3. THEN the link between switch1 and switch3 crashes. SO switch3 sends out a TCN to switch2 on it;s way to the root.