Spanning tree putting interfaces in err disable

As I am studying for the BCMSN/switching exam this problem has interested me.

Below is a short example of the topology, with multilayer switch 1 as the root bridge. The switches at the top reside in a server rack connected redundantly at 10Gb. The switches to the right reside in another rack and the switch at the bottom is the access layer.

Originally the 10Gb switches were connected to multilayer switch 0 only. When they were connected to the root bridge as well, the 40Gb between the 2 cores went into error disable and spanning tree reconverged across the whole network.

Now I assume the BPDU's met at the 40Gb link and so is where it decided to stop the loop with out taking into consideration path costs to the root bridge and speed of the links?

If the 10Gb switches were connected to the root bridge first and then to the secondary, would the 40Gb link still have gone into error disable?

How would you go about ensuring the 40Gb link never goes into error disable/blocked due to spanning tree, while ensuring you never get a loop with the redundant network? Would BPDU filter on the port group ensure the loop is then blocked in the correct location (as shown in the picture)? The book states you should only enable it on ports that have a single host connected and loops are impossible. Which obviously isn't the case.

Find more posts tagged with

Save $250 on 2025 certification boot camps from Infosec!

Book now with code EOY2025

Button

Comments

Heero

chmorin wrote: »

Any tutorials out there for putting in syslog? Does it cause any downtime in a production environment?

putting syslog in should not cause any issues, but I would always recommend you not make changes to a production environment unless you really know what you are doing. Might consider setting up syslog on a cisco router/switch and setting up a syslog server just to play a round a little bit and learn it. Its a very good thing to know.

Cyanic

I don't get all this spanning tree priority talk. If an election happens then a root bridge will be selected and ports will be put into blocking, but he is seeing ports going into error disabled mode and not a blocking state.

He is going to have to get the ports back into that mode and see what the reason is to fix this.

billscott92787

Cyanic wrote: »

I don't get all this spanning tree priority talk. If an election happens then a root bridge will be selected and ports will be put into blocking, but he is seeing ports going into error disabled mode and not a blocking state.

He is going to have to get the ports back into that mode and see what the reason is to fix this.

I agree because either way no matter which switch M0 or M1 is root, the Etherchannel should still work. Can you post the actual config directly from the other Switch? You are 100% sure that it is the exact same config? I just figured I could put another set of eyes on it and see if I see anything out of the ordinary.

Trifidw

The configuration is identical on both switches. The only thing that differs between them is spanning tree priority and HSRP standby.

Anyway, after lots of head scratching and reading I've come up with the following theory. We plugged in the second 10Gb link, spanning tree then decided to put the etherchannel into blocking mode at M0. Spanning tree then reconverged across the whole network causing an outage. Because communications across the etherchannel were being blocked by spanning tree, LACP messages were no longer being exchanged (and here's the weak part in my plan) M1 put the link in error disable because it thought it was a configuration issue?