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

[Steve Howell]

We have an otherwise healthy network that is dropping
packets on precise five minute intervals.  The problem
originally came to our attention when we started
seeing SYNs dropping from one of our applications, but
we can also reproduce it with traceroute.  We narrowed
down the problem to where we see our 6500 get a
packet, but it does not deliver it to the neighboring
switch.

There is no evidence of bursty traffic at the time of
the packet drops.

I tried looking for something on the switch that could
have an aging interval of five minutes, and it turns
out that spanning trees typically age out after five
minutes.  I am wondering if it's plausible to have
some kind of spanning tree misconfiguration that could
cause a very transient network failure.  Most of what
I've read so far suggests that spanning tree problems
take significantly longer to resolve themselves than a
second or two, so I could be chasing down a complete
red herring.

Just to complicate things, I'm a software developer,
not a network engineer, and I don't have access to the
Cisco box.  What I'm looking for here is a plausible
theory that would explain short-lived failures to
deliver packets from a 6500 that happen on precise
five minute intervals.  Once I have a decent theory, I
can try to have a somewhat intelligent discussion with
my network engineer colleagues, even though I'm a bit
out of my depth here.

Thanks,

Steve



		
__________________________________ 
Yahoo! FareChase: Search multiple travel sites in one click.
http://farechase.yahoo.com