STP Questions

[beridor]

Hi Gurus,

I have a question on STP - My understanding is that on a topology change (break) the STP network will drop into the listening and learning states resulting in a recovery time of 30 seconds.

does this mean that the whole network re-calculates a new topology so that even devices that are not affected by the break are prevented from communicating? or is it just the switches that are affected that block communications.

Example below

***********
-------------* SW1 *-------------
| *********** |
*********** | ***********
* SW2 *-------------------------* SW4 *
*********** | ***********
| | | | |
| | *********** | |
| --------------* SW3 *------------- |
| *********** |
| | |
| *********** |
-----------------* SW5 *----------------
***********

In the mesh network "shown" above, if STP finds a route between SW1 and SW5 via SW3, would a failure of SW4 cause communications on all switches to stop whilst the new topology is calculated?

Whilst trialing this in a lab it seemed that a failure of SW2 caused the topology to recalculate and temporarily prevented comms between SW1 and SW5 (even though the link was not directly affected) but a failure of SW4 seemed to have no effect.

Any suggestions would be welcomed.

 

[HungryHouse]

If there is a new root bridge elected, then all the switches will recalculate the oath back to the root and block, or forward as necessary to block loops. If a bridge fails, then blocked links may become forwarding links if less important redundant links were the ones left standing.
In your example, you will need to figure out the root bridge and which links are blocked and forwarding and then the differences in the transitions will make more sense.
Keep in mind that switches in the same STP group will recalculate...switches in a different STP group will be unaffected.

 

[beridor]

Hi HungryHouse,

Thanks for the response I appreciate your help, what controls the STP group? it's the first time I've seen this mentioned.

Also in a config such as below: -

B***********B
-------------* SW1 *-------------
|F *********** |F
***********F F | B ***********
* SW2 *-------------------------* SW4 *
*********** | ***********
F| |B F| |B |F
| | F***********F | |
| --------------* SW3 RB *------------- |
| *********** |
| F| |
| F *********** F |
-----------------* SW5 *----------------
***********

Would a failure of SW2 or SW4 cause STP to recalculate (LAB showed that failure of SW2 had no effect, failure of SW4 caused re-calc of links.)

I'm trying to understand how I can predict (knowing the architecture, Root Bridge, Ports forwarding etc.) what the effect of a failure or the replacement of a failed switch will have on the network.

I was suprised to see a switch fail with no effect on the network, assumed that I would lose the whole network for 30 seconds.

Cheers

Dave

 

[KiscoKid]

The failure of SW2 would have no effect only if SW4 was something called the 'designated bridge' for the segment they connect to.

To predict what your network is going to do, you need to understand which of those devices is the root bridge. Also you would need to understand the bridge priorities and/or MAC addresses of all the bridges there to determine which one(s) will become a designated bridge.

 

[beridor]

I appreciate your input KiscoKid,

Do you have a good reference on STP - the Cisco books don't seem to cover it in as much detals as I would like.

Cheers Dave

 

[HungryHouse]

Here is a pretty good tutorial/reference:

http://www2.rad.com/networks/2005/bridges/main.htm

-HH