On Tue, 10 Oct 2000, Steve Meuse wrote:
>
> Beware that some non-cisco switches *may* have issues. Cabletron had issues
> with securefast and HSRP initially, due to some MAC layer caching issue. It
> hit us a few years back.
>
> -Steve
I should clarify/elaborate. Here's an excerpt from the note
I sent internal to our company about this (back in early 1998):
-- Apparently the problem is that the SFS system keeps one database of MAC addresses per "administrative domain". Since HSRP picks its MAC address by taking the standby group number, converting it to hex, and tacking it onto the following prefix: 0000.0c07.ac[0x<standby number>]. Thus, since we use standby group number "1" as our default starting point on a given physical interface, we get the router ARPing with the same MAC address on multiple VLANs. This would cause the Cabletrons to oscillate which port they were delivering packets to. Cabletron indicates that they will fix this problem (really a poor design decision, not a bug) in their 1.8 version of code due out this summer. In the meantime, we have to take care to keep collisions of standby group numbers from happening. --Seems like presuming global MAC address uniqueness (or even w/in one administrative domain) is more problematic than some thought it would be.
Come to think of it, what does this do for IPv6 autoconfiguration using embedded MAC addresses?
If any worms crawl out of this can, please change the Subject: line to something different and more appropriate.
Also, this question shouldn't be construed as meaning I have any interest in IPv6 beyond a morbid highway accident kind of curiosity.
Tony
This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:12:18 EDT