Spanning Tree Issue

[rhynosmiz]

We had a loop created recently that caused more switches to go down than I expected.  A desktop switch was plugged into itself which I traced back to the IDF and was plugged into a 2650 at the bottom of the rack.  All five switches in the IDF went down when the loop was created.  I checked the logs and the uplink fiber port was blocked by STP which goes back to a 4108 series switch in the MDF whichi then connects to the .1 top level switch in the MDF which is a 5406zl.  The top level switch (.1) has a STP priority of 0 whereas all other switches have a priority of 8.  Is this what should have happened with STP or do I need to edit the configs?

[Vince-Whirlwind]

What is "A desktop switch"?

No, STP should never take down a whole rack of switches.

No, if you have 4 layers of switches, then your STP topology should be designed with 4 different STP priorities, in ascending order out from your Core switch.

All STP does is block local links to prevent loops. If your "desktop switch" didn't block the loop, then that is where your problem lies.

[rhynosmiz]

Thanks for the reply.  By desktop switch, I mean a five port unmanaged switch.  We don't have redundant links between the four switches in the rack.  I know we could use BPDU protection to prevent unautorized switches from getting plugged into the network, but they are unfortunatey necessary in our environment.  Are redundant links required for propper spanning-tree population?

Thank you again.

[Vince-Whirlwind]

If it's an unmanaged switch then your issue is not spanning tree. Spanning tree is for blocking redundant local links. It can't do anything for broadcast storms occurring on remote switches.

You need to configure storm control on your uplinks to the unmanaged switch. Or some kind of rate limiting for broadcast traffic.