I agree with Chuck that you don't want to enable portfast on a port that connects a switch. You could disable STP all together, as Chuck mentions, but that is risky unless you have complete control of network equipment installs, good documentation of your logical and physical topology, and a good change control process.
But you could look into Cisco's other features, such as Uplink Fast and Backbone Fast. These speed up STP convergence time. Their proper use is dependent on your topology, which I admit I didn't look at carefully. Hey, my state is on fire, and I'm a little distracted! ;-) Priscilla Chuck's Long Road wrote: > > Some comments below: > > ""dovelet"" wrote in message > [EMAIL PROTECTED]">news:[EMAIL PROTECTED]... > > Hi all, > > > > I have a question about Cisco Catalyst 3500XL switch with > Gigastack in > > Spanning Tree, and I hope someone can help me. I have a > network like this: > > > > [Root] > > SW1 ----- Gigastack (1)------ SW2 > > |----------Gigastack (2) -----x--| > > > > I have 2 Cat.3500XL switches and use 2 gigastack cable to > connect them. It > > work fine and I found that the LED on Gigastack (1) is green > while the > > SW2-Gigastack (2) is in amber. I assume that the port on > SW2-Gigastack (2) > > is spanning tree blocking. Now, I want to install one more > Cat.3500XL > > switch. Then I disconnect the cable SW2-Gigastack (2). > However, I found > that > > the network connection between the two switches are broken > for about > 50sec. > > I think it is due to spanning tree re-learning. > > CL: yes. this is the way spanning tree works. new links > detected. STP runs. > This is intentional, and is to ensure loop free topologies. > this is alos the > bane of network managers everywhere, for a lot of reasons. > > > > >Then I install a new Cat.3500XL (SW3) into the stack. > > > > [Root] > > SW1 ----- Gigastack (1)------ SW2 > > | > > |----------Gigastack (2) -----SW3 > > > > However, I found that the network connection between SW1 and > SW2 was > broken > > again for about 50 sec. The spanning tree re-learn again! > > CL: sounds about right. > > > > > > My questions are: > > 1. Why is the spanning tree re-learn if I just disconnect the > blocking > link > > at SW2-gigastack (2)? > > CL: becasue you have still introduced a change. i.e. the link > in question > disappears > > > 2. Why is the spanning tree re-learn if I install a new > switch? (I am sure > > SW1 is still the root bridge) > > CL: becasue you are still introducing a change. STP runs > because there > exists the possibility that the new switch introduces a loop. > it is also > possible that the new switch has a better priority and is > supposed to become > the root bridge. > > > > 3. I want to enable spanning portfast on all of the gigastack > port. > However, > > I wonder will it cause network loop if I connect the SW2 and > SW3 with > > Gigastack? > > CL: portfast is not a real good idea for ports connected to > other switches. > the whole idea of STP is to ensure loop free topologies. you are > deliberately bypassing the process that ensures loop free when > you enable > portfast. the whole idea of portfast is geared towards > workstations that > time out, thinking they are not connected to a network, during > the time the > STP is running. > > > > 4. As the SW3 is on loan, I need to replace it after several > months. Is > that > > means, the network connection between SW1 and SW2 will be > interrupt again? > > CL: yes. so do the work after hours, if this is an issue. > > > > > Please help! > > > CL: given waht appears to be a small and more or less controlled > environment, you can disable spanning tree, if you think you > can assure loop > free in general. STP is there to protect you, but in some > environments it > may not be necessary. > > > > > > > Thanks, > > Dovelet > > Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=51455&t=51454 -------------------------------------------------- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
