Spanning tree - TCN and convergence question

[jschweng]

We have four siwtches all with VLAN 10 running regular perVlan Spt.

They are all cross connected to each other so you can get from one switch to any other switch via only one hop.

max age = 20, forward delay =15,  MAC timtout = 300

The root bridge is on Switch 2.   All ports from switch 1 & 3 are forwarding to the root bridge.  We have traffic going from a host swtich 1 to a host on switch 3.    Then the root bridge fails.

Will the other swtiches get the TCN and reset their MAC timeout to forward delay?

Isn't the root bridge suppose to get the TCN and then tell the otehr swtiches - so since the root failed will that not work and we will have a five minute outgage due to the MAC timeout fo 300s?

Do the other ports on Sw 1 and 2 go immediately into listening and learning once they get a  TCN - or does it wait for MAX Age to timeout.

Does reseting the MAC againg to forward delay occur immediately upon receiing a TCN or after the switch completes it SPT timers?

[Giuseppe Larosa]

Hello Martin,

>>> Then the root bridge fails.

>>>> Will the other swtiches get the TCN and reset their MAC timeout to forward delay?

if the root bridge fails a new root bridge election occurs during re-election no ports will be in forwarding state in surviving switches so second question has no answer if no port is in forwarding state no user traffic is moved

depending on the way that previous root bridge fails the other switches have to wait for max age (if the failed switch keeps up the links to the others) or it will start as soon as the root port fails down

even if it may not apply in your scenario:

>> Does reseting the MAC againg to forward delay occur immediately upon receiing a TCN or after the switch completes it SPT timers?

immediately as it a way to flush CAM table as changes are still happening after all STP timers are expired that would the good time to revert to default values.

Hope to help

Giuseppe

[jschweng]

Hi Giuseppe - thank you for your  response.

teh root bridge failed powered supply so all ports went down.  I guess what I'm trying to find out is we had a 5minute plus outage or a 50 second outage due to the CAM aging time?.

It seems that the CAM aging gets set to Forward delay only via the TCN which would of been propogaged by teh Rootbridge to the otehr swithces.  If the root went down then no TCN?

So what happens on teh otehr siwthces - the root ports continue to stay forwarding and teh other ports are blocking until Max Age. Then listening, and learning until a new root is elected.

Does the CAM table timeout get reset

[Giuseppe Larosa]

Hello Martin,

>> teh root bridge failed powered supply so all ports went down

this is detected by the other switches and trigger a new root bridge election as noted in my previous post

detection time depends on technology used with fiber based links that can be very fast and 100 Mbps RJ45 links that take up to 750 msec to detect fault.

After a new root bridge is elected, new root ports on non root switches will be chosen and designated ports in each lan segment, after all this traffic is served again

Hope to help

Giuseppe

[jschweng]

ok thanks - is it safe to say that the time relect a new bridge is at least Max Age + 2 x forward delay.

is this correct?

After the root fails - the surviving switches will wait for Max Age, then Listening/Learning on every port of that VLAN.

The result will be the new root bridge and every switch will have  a new path to it?

Dosn't the MAC age timer get reset to forward delay at any time on the surving swithces?

[Tharak Abraham]

Martin,

Since the Root switch failed, spanning tree protocol will detect a direct failure on the non-root bridges.

The root ports of all the other three switches (assuming all three have connection to the Root Bridge) goes down.

The scene now calls for relection IMMEDIATELY.

All the three switches agrees on the superior BPDU and elects one of the three remaining bridges as the Root.

This could eventually be equal to the forwarding delay which is 30 Secs.

>>Dosn't the MAC age timer get reset to forward delay at any time on the surving swithces?

The Max age timer kicks in if its an indirect failure, but in your question the other three switches have to transition through the listening and learning.

Thats how STP avoids loops..