[c-nsp] spanning trees failing on five minute intervals

[Steve Howell]

--- "Church, Chuck" <cchurch at netcogov.com> wrote:
> MAC addresses in the CAM table time out after 5
> minutes by default, I
> believe.  Sounds like an address is timing out,
> forcing a destination to
> be re-learned.  You can try making the timeout
> period 10 minutes for all
> VLANs, and see if the problem occurs every 10
> minutes.  

This is a good suggestion; I will pursue it with my
colleagues.

> If so, you'll
> know for sure that's the cause, and can then work on
> why a destination
> with steady traffic is timing out.  Is it affecting
> just one device? 
> 

So far I've seen it affect only one device first hand.
 We have an application on box A that is sending SYNs
every second to as-of-yet-nonexistent applications on
boxes B1 and B2.  Boxes B1 and B2 are basically
identical, except, of course, for the hardware.  We
only see the regular packet drops en route from box A
to box B1.  Naturally, one would think that there's
something wrong with B1, since packets always reach
B2, but whatever's wrong with B1 is fairly subtle. 
When packets drop, they don't even reach the switch
that is to the left of B1:

  A -> backbone -> 6500 -> switch -> B1

If the problem is with B1, one would think it has
something to do with B1 not properly advertising its
address to the 6500, correct?






		
__________________________________ 
Start your day with Yahoo! - Make it your home page! 
http://www.yahoo.com/r/hs