Re: CSS 11051 Box to Box redundancy

f.chiesa · ‎12-19-2002

I have a redundant configuration of two CSS 11051 with a crossover cable between them for the failover. The type of redundancy is Box to Box.

All works fine except a situation: if you unplug the crossover cable (simulating a fault of one of the two interface FE), the two CSS become both active and all the system goes down. Obviously the behavior is ok but the customer is not happy. It would like to put a second cable between the css to have more availability. BUT I can't see any info about it on the web doc. Then I already know that that CSS doesn't support Etherchannel so no way in that direction.

At the end, anyone know or have tested situation with two crossover cable with one that probably will be blocket by the SPanning tree until the fault oh the other?

Any comment or help will be very appreciated.

pknoops · ‎12-19-2002

Hi,

The bottomline here is very straight forward. What you describe is a very common issue and quite honestly there is no way around it. Removing the cable or losing an ethernet connection on a redundancy-protocol link will cause a dual MASTER situation. The other bad news is that the use of 2 crossover cables between 2 CSS boxes in a bob/box redundancy setup is not supported. This is probably why you can't find anything on the web regarding it. The dual crossover link unsupportablility revolves around a bridge/spanning tree problem.

Regards

Pete Knoops

Cisco Systems

f.chiesa · ‎12-27-2002

Thanks a lot for your reply (and your confirm of the situation...). I will advise the customer of that and I will suggest the use of the other type of redundancy.

Best Regards and wishes for an happy new year!

Fabio Chiesa

snugteren · ‎12-31-2002

Hi,

we have 2 x-cables running between two css11800, both in the same vlan and so both running the redundancy protocol. Both ports are on both boxes in forwarding state. Vlan has an ACL attached only allowing VRRP over the interfaces.

We tested this and have it running in life operation. Can't explain the spanning tree not blocking though. Maybe something to do with the ACL definition (allthough STP is lyr2 offcourse) or some code in the redundancy protocol?

Serge