What does that even mean?? Let's see..
In the topology, SW-A is the Root.
So E0/1 of SW-B & E0/2 of SW-C are Root Ports..
E0/3 of SW-C is the Alternate Port..
All are Fast Ethernet ports..
In the above topology, because SW-B has all ports in Forwarding state, an indirect failure is a failure of the link between SW-A & SW-B.. In other words it is a indirect link from SW-C's point of view which maintains the Alternate Port..
Let's see what happens when a Root Port is down..
I have enabled debug spanning-tree events on SW-C..
Removing the SW-A to SW-B link..
(click on images to see in full size)
What happened is that when SW-B loses the link to the Root Bridge, it thinks it is the Root Bridge and sends Inferior BPDUs towards SW-C's Alternate port..
SW-C's Alternate Port will go from Blocking > Listening > Learning > Forwarding in the normal manner which will take 50 seconds by default.
Now let's connect the cable again and configure BackboneFast..
This is a global command and should be enabled on all switches in order to work..
SW-A(config)#spanning-tree backbonefast
SW-B(config)#spanning-tree backbonefast
SW-C(config)#spanning-tree backbonefast
Removing the SW-A to SW-B link..
(click on images to see in full size)
Now we can see SW-B is sending one Inferior BPDU only.. And then SW-C did not wait for the MaxAge time of 20s to go to Listening state..
UplinkFast for direct Root Port failure was a lightning fast feature. But BackboneFast can only reduce 50s of default convergence time to 30s by eliminating the MaxAge time.
When reverted back the time taken to re-converge was same (30s).
What really happened was when the BackboneFast is enabled, SW-C sends a special message called RLQ (Root Link Query) to SW-A; the Root to find whether the Root is active or not. When SW-C got reply from SW-A, it assumes SW-B has lost the connection to the Root and immediately moves it's Blocking port to Listening > Learning > Forwarding states ignoring the 20s default MaxAge time.
No comments:
Post a Comment