[ILUG] Delay between link up and forwarding frames?

[Kenn Humborg]

> Kenn Humborg wrote:
> [snip]
> > Turning on 'portfast' works nicely.  Thanks!
> > 
> >> If I understanding it correctly, spanning tree is only needed on
> >> switch-to-switch links, so I need to remember to turn it back on
> >> if I ever connect a switch to one of these ports.  Correct?
> > 
> > Misunderstanding on my part there.  Since I'm not turning off
> > spanning-tree, I don't really need to worry.
> [snip]
> 
> 	Hi,
> 
> 	On the contrary, you seem to have exactly the right understanding.

You misunderstood my misunderstanding :-)  I initially read 'portfast'
as meaning that spanning-tree would be turned off.  It actually
means "forward frames immediately, do STP stuff, and stop forwarding 
if STP detects a loop".  

> Spanning-tree is a loop-avoidance technology, and since computers
> typically only have one connection to the network, it's unlikely they
> can cause a loop. Hence, the 'portfast' directive on Cisco to state that
> you are plugging a single, end-point device into a port, so the switch
> should not worry about spanning tree for that port. (In the case of
> servers with more than one interface, it's at least likely that traffic
> will not be switched between the two NICs, so portfast is probably still
> safe.)
> 
> 	If you are plugging a switch into a port, you should ensure that
> portfast is NOT configured for that port. You definitely want spanning
> tree to be running on any ports that are connecting switches in your
> network, to ensure that it can protect you from loops.

I actually turned on portfast by accident on the port to the upstream 
switch and the firmware just warned that it got STP frames on a portfast
port, but it worked fine.

Not the way I'd leave it configured, though :-)

Later,
Kenn