[mrtg] Re: [RE: High Speed If Cisco Mibs Was Re: Help--MRTG Grap

[Graeme Fowler]

Jeff

You wrote:
> Thanks for the response. I changed my polling interval from 5 minutes
> to 2, and this seems to have resolved the problem. Is this apparent
> counter limitation caused by MRTG or by the Cisco MIB? Going forward,
> how do I correct this problem? Or should I just leave it polling every
> 2 minutes?

It's a limitation in the counters on the Cisco, not in the MIB per se.

We saw this recently on a 3Com SSII 3300 switch which runs one port
(uplink to a web cache farm) at close to 100Mbps close to 80% of its'
operational time.

Once we got to 80% utilization we noticed some tremendous oddities in
the traffic graphs... which got us all very worried that the farm
wasn't acting as it should!

In the end I worked it out:

The switch uses 32-bit counters and counts octets. Above 80Mbps we were
exceeding 2^32 octets in the five minute period so the counter rolled
over ans started at 0 once again; hence the delta in that period either
came out as negative or zero. In both cases this meant that the
effective delta is zero so the graphs looked rather odd!

Once I figured that out I changed the polling interval to 2 minutes and
everything looked fine again. Phew!

So - don't blame MRTG - it's the boxes themselves. I guess counters
aren't as important in the end as routing/switching engines so they get
the lower spec hardware :)

Graeme


-- 
Graeme Fowler
Network Officer, Infrastructure & Networks Group
Loughborough University Computing Services
+44 1509 228426

--
Unsubscribe mailto:mrtg-request at list.ee.ethz.ch?subject=unsubscribe
Help        mailto:mrtg-request at list.ee.ethz.ch?subject=help
Archive     http://www.ee.ethz.ch/~slist/mrtg