CSS Box-to-Box Redundancy in a flat environment, Spanning-Tree

[s.steiner]

Hi,

our customer wants to set up the following configuration :

switch----------switch

| |

CSS1----FE-C/O----CSS2

| |

switch----------switch

All interfaces, except the FE-Crossoverlink for redundancy-protocol, are in the same VLAN.

I think we have a spanning-tree issue here.

- How can we guarantee, that this setup will work with spanning-tree configured ?

- Should we disable spanning-tree on the CSSs to allow a transperant handling of the BPDUs (so that the switches process the BPDUs and not the CSSs) or do the CSSs drop the BPDUs ?

- Should the CSSs participate in the spanning-tree process ?

- What about portfast or uplinkfast on the Cisco-Switches in this szenario to enhance the convergence in case of a switch-over ?

- Are there any conceivable cases, in which spanning-tree cannot determine, that a CSS does not work correctely ?

I couldn't find any in the docs regarding this design. If you have a link or something for me, please post it here.

Thanks a lot in advance,

Stephan

[Gilles Dufour]

The CSS drops BPDUs when spanning-tree is disabled.

http://www.cisco.com/en/US/products/hw/contnetw/ps792/products_configuration_guide_chapter09186a008028ff8c.html

I have seen such design in the past and it was working with spanning-tree enabled.

You can't use portfast with spanning-tree enabled.

The Catalyst will go errdisable if it receives BPDUs from such port.

Not sure about your last question.

Regards,

Gilles.

[s.steiner]

First of all, thanks for your reply.

Ehem, portfast is not really a good idea, yeah !

My last Question :

forget it ! ;-).

Thanks for your help !