3ap in WDS-triangle?

[Juan Jesús García de Soria Lucena]

Hi, Hiphin...

El s?b, 30-10-2004 a las 13:06, Hiphin escribi?:
> Hi Jouni, respect for every men that contrubute and create ...
> 
> > Isn't that the expected behavior of bridging with redundant links? STP
> > detects a look and marks one of the links down. If you were to bring ap1
> > down, ap2 to ap3 would start using the direct link.
> 
> I'm not shoure what you are meaning, all AP are up, and runing, and I can't 
> imagine way ap3 is blocking ap2, except for reason that he see ap1 himself 
> and throught ap2 (he see same MAC in ap2br0, and ap3br0).

I don't know if the following applies to WDS, but it definitely does for
standard ethernet bridging:

Standard ethernet "transparent bridges" build an auto-configured
expansion tree (logic topology as a subset of the physical topology).

This subset is configured as a tree so that no loops happen in it, so
that the forwarding algorithm is simpler and more robust. It's called
"expansion tree" because it's discovered by means of flooding discovery
frames.

The former statement means that your triangle physical topology would be
reduced to a simple tree with only two links in order to avoid the loop
that would appear if the triangle was fully closed.

Yes, I know that could be suboptimal about performance/throughput, but
it's the standard way for ethernet bridges.


Regards,

   Juan Jes?s.

-- 
Juan Jes?s Garc?a de Soria Lucena <skandalfo at telefonica.net>