Cisco Support Community
Showing results for 
Search instead for 
Did you mean: 
New Member

CSS 11501 Trunking and Spanning Tree

I am running two CSS 11501's (running 7.40) attached to two Catalyst 3550's (running 12.1(19)EA1).

I have converted two of my interfaces from bridging to trunking on my CSS. There is an in-built bridge loop for network resiliency. The CSS's are connected to Catalyst 3550 switches thusly:

css0-e1 --> 3550(0)-fa0/45

css0-e2 --> 3550(1)-fa0/46

css1-e1 --> 3550(1)-fa0/45

css1-e2 --> 3550(0)-fa0/46

When I was bridged, and the 3550 switches were configured to be the designated root and secondary the spanning tree calculated correctly.

There is a note in the CSS configuration guide that talks about this situation, but the last line of the note states:

" When you connect a Cisco Catalyst switch to a CSS over an 802.1Q trunk, the result is that neither switch recognizes the other's BPDUs, and both assume root status. If a spanning-tree loop is detected, the Catalyst switch goes in to blocking mode on one of its looped ports."

This is not happening. When I shut down e2 on each css then, of course, everything works because the STP loop does not exist. When I turn up e2 on both CSS all the ports on the CSS and the 3550 stay in forwarding mode and the HSRP on both 3550 (for vlan1 and vlan10) assume active status with no known standby.

I also received this error at the 3550 when turning up e2 on the CSS:

1d20h: %STANDBY-3-DUPADDR: Duplicate address on Vlan1, sourced by 0013.c3db.c080

Also, is there any possibility that the CSS OS will be changed to work with the Catalyst tagged MAC Address in the future (I checked version 8.10, but it appears that it is the same as 7.40 in this regard).

New Member

Re: CSS 11501 Trunking and Spanning Tree

I don't think you really gain much with the topology you are using, no matter which mode you are running the CSSs in(box-to-box or VIP redundancy).

You may want to consider putting both uplinks out of each CSS into only one switch, and then cross connecting the 3500's instead.

This creates a single point of failure(the link between the switches), so make it a port channel and spread the ports out so they are not coming off the same ASICs.

Then configure the CSS to fail completely in the event that either uplink flakes out.

Make sense?

CreatePlease to create content