STP Problem

[iaglindro]

I'm on the process of configuring two switches (4,5) to our existing network.

Topology overview as follows:

1 - 6509

2,3 - 3548

4,5 - 3550-24

1-->2-->3-->5-->4-->2 (all connections are trunks)

configuration:

between 5 and 4: spanning-tree cost 500 (configured on both switches)

between 3 and 5: spanning-tree cost 200 (configured on 5)

between 2 and 4: spanning-tree cost 400 (configured on 4)

problem: loop exists because LEDs of corresponding ports are continuously blinking and when i do show spanning-tree on all switches, all are FWD.

Question: How can i remove this loop.

thanks for the time and reply is very much appreciated.

cheers,

-ian

[konigl]

Ian,

First off, disconnect the link between the two 3550-24s (4 and 5). That should stabilize your network.

Then, check the switch-to-switch links, particularly between 3548 (3) and 3550-24 (5), for duplex mismatch. Look for ports which may have been manually configured for full duplex, being connected to ports that are set to autodetect speed and duplex. When a port is hard-coded for duplex, it does not autonegotiate; and a port that's set to autonegotiate will go to half-duplex if it's connected to something that won't negotiate. This can lead to ports that would otherwise be blocked by STP, moving to forwarding state and creating a loop.

Also check the switch-to-switch links also for Spanning Tree PortFast enabled. If it is enabled on either end or both ends of a link, disable it. Loops on PortFast ports should clear when one port should go to blocking, but if traffic's heavy sometimes they don't have the chance to.

Those are some of the easier hints. Would need to know more about your network (for example, how many VLANs do you have and what are they, which switch is the Root Bridge for each VLAN, what kinds of links do you have between switches, how are those links configured, are any STP parameters changed from their default settings, etc.) to go further.

Hope this helps.

[iaglindro]

hi!

what would you recommend? force all ports to full-duplex? right now, all ports involved are autonegotiate fastethernet ports (3548s and 3550s) and half duplex for gigabitethernet ports (3548s and 6509). No configurations for portfast exists.

right now, we have 50+ vlans and the root bridge is the 6509. all stp parameters are by default except for the spanning-tree costs on the ports for switches 3, 4 and 5.

thanks again for your prompt reply.

cheers,

-ian

[konigl]

Since they're direct switch-to-switch connections, with nothing else in the way, there's no point in running any of them half duplex. Unless you're cascading GigaStack GBICs in the 3548s or 3550-24s. (Are you?) Crank them up to full; as long as nobody moves the patch cables off those ports you should be fine.

While we're on the subject of switch-to-switch connections, just how exactly are these things connected together? I'm trying to draw a picture, and want to make sure it matches what you have.

6509 (1) slot__/port__ to 3548 (2) GigabitEthernet 0/port__: TX, SX, or LX/LH GBICs?

3548 (2) GigabitEthernet 0/port__ to 3548 (3) Gigabit Ethernet 0/port__: GS, TX, SX, or LX/LH GBICs?

3548 (3) FastEthernet 0/port__ to 3550-24 (5) FastEthernet 0/port__

3550-24 (5) FastEthernet 0/port__ to 3550-24 (4) Fastethernet 0/port__

3550-24 (4) FastEthernet 0/port__ to 3548 (2) FastEthernet 0/port__

And are they all in the same room? If not, which ones are where? Don't need to be real specific, just trying to understand what your MDF and IDF layout is, how the switches are grouped. Also want to better understand need for redundant connections among the 3548s and 3550-24s, but the solitary connection to the 6509.

What Bridge Priority do you have set for the 6509 acting as Root Bridge? Is 3548 (2) configured to act as a secondary Root Bridge if the link to the 6509 goes down? Are any of the 3548s or 3550-24s running with UplinkFast enabled?

You're not pruning any VLANs off the switch-to-switch VLAN trunk links, are you? And all your switch-to-switch links are configured as VLAN trunk ports, right? Which VLANs have access ports on what switches?

And why the unusual STP costs? Seems to me from the picture I already have (which may be inaccurate), that default costs would have resulted in blocking the same port on 3550-24 (5).

Sorry about all the questions, but I see nothing obvious yet that's causing your network problem, so I've got to ask them.

[iaglindro]

No gigastack except on 2 and 3 which are working fine.

6509 (1) slot1/1 to 3548 (2) GigabitEthernet 0/1:LX GBICs

3548 (2) GigabitEthernet 0/2 to 3548 (3) Gigabit Ethernet 0/1: GS

3548 (3) FastEthernet 0/48 to 3550-24 (5) FastEthernet 0/2

3550-24 (5) FastEthernet 0/1 to 3550-24 (4) Fastethernet 0/1

3550-24 (4) FastEthernet 0/2 to 3548 (2) FastEthernet 0/13

actually i have two redundant 6509 switches but since one is on standby, i just pictured out 1 6509. secondary bridge is 6509

None of the switches are running uplinkfast.

-ian