Spanning Tree

Answered Question
Aug 14th, 2007
User Badges:

Hi,


If a spanning tree environment looses its ROOT bridge, a new bridge will be elected.

My question is if during this time the switch ports will go through the lis-lrn-fwd states or if the new election is done transparent?

(in this case the ROOT bridge that disappears has no links in a loop, it is a ?switch on a sitck?)


/Regards Rikard


Correct Answer by Francois Tallet about 9 years 8 months ago

It is very difficult to predict and depends on your topology. Basically, when the root bridge disappears, the remaining bridges in the network compete for a new root bridge election. The location of the blocked ports will depend on the location of the root bridge. It is possible that a subtree computes a blocked port based on a "temporary" root bridge (a bridge that could only claim being the root for a short period of time, before the information from a better root is received on this subtree). The the temporary root is replaced by a better root, the blocked port may move again. This is done slowly in STP and you will have two ports (instead of one) blocking for 30 seconds in this situation, with STP. With RSTP, and MST, the same thing may happen, but the ports can unblock without waiting on a timer so the traffic interruption should be very short.

Now, applying this to the particular case of your ring. Your final root bridge may lead to blocking a different port than the previous root bridge was blocking. This means that you could have two ports blocking for about 30 seconds, necessarily leading to loss of connectivity for some station. The solution is to put a secondary root (i.e. pre-configure a bridge so that you are sure that it will win the root election when the original root has disappeared) that will result in the same blocked port to be computed in the final topology. Putting your secondary root on the ring bridge that is on the way to the primary root should achieve the desired result. Hopefully, you will not suffer from the "temporary root" effect I was describing earlier in a ring.

As far as RSTP and MST are concerned, you may still get some "syncs" (port reverting to blocking temporarily), but again, it should only result in a very short outage.

Regards,

Francois


  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 4 (1 ratings)
Loading.
mahmoodmkl Tue, 08/14/2007 - 03:06
User Badges:
  • Gold, 750 points or more

Hi


If u dnot have any redundant links between u r switches its like turning off spanning-tree in the network.I dnot think will there be any issue on the switches,as all the switches will not have any ports blocked by the spanning-tree.


Thanks

Mahmood

rwedham Tue, 08/14/2007 - 03:20
User Badges:

I have redundant links, but the STP root bridge doesn?t have any links there.

It is like a ring of switches, and then the root bridge is connected to a switch in the ring but is not taking part in the ring.


/Regards Rikard




Correct Answer
Francois Tallet Tue, 08/14/2007 - 08:56
User Badges:
  • Gold, 750 points or more

It is very difficult to predict and depends on your topology. Basically, when the root bridge disappears, the remaining bridges in the network compete for a new root bridge election. The location of the blocked ports will depend on the location of the root bridge. It is possible that a subtree computes a blocked port based on a "temporary" root bridge (a bridge that could only claim being the root for a short period of time, before the information from a better root is received on this subtree). The the temporary root is replaced by a better root, the blocked port may move again. This is done slowly in STP and you will have two ports (instead of one) blocking for 30 seconds in this situation, with STP. With RSTP, and MST, the same thing may happen, but the ports can unblock without waiting on a timer so the traffic interruption should be very short.

Now, applying this to the particular case of your ring. Your final root bridge may lead to blocking a different port than the previous root bridge was blocking. This means that you could have two ports blocking for about 30 seconds, necessarily leading to loss of connectivity for some station. The solution is to put a secondary root (i.e. pre-configure a bridge so that you are sure that it will win the root election when the original root has disappeared) that will result in the same blocked port to be computed in the final topology. Putting your secondary root on the ring bridge that is on the way to the primary root should achieve the desired result. Hopefully, you will not suffer from the "temporary root" effect I was describing earlier in a ring.

As far as RSTP and MST are concerned, you may still get some "syncs" (port reverting to blocking temporarily), but again, it should only result in a very short outage.

Regards,

Francois


Actions

This Discussion