[rrd-users] Re: odd spikes due to early resets

BAARDA, Don don.baarda at baesystems.com
Tue Feb 20 00:32:32 MET 2001


> -----Original Message-----
> From:	Alex van den Bogaerdt [SMTP:alex at slot.hollandcasino.nl]
> Sent:	Tuesday, February 20, 2001 9:15 AM
> To:	don.baarda at baesystems.com
> Cc:	rrd-users at list.ee.ethz.ch
> Subject:	[rrd-users] Re: odd spikes due to early resets
> BAARDA, Don wrote:
> > > 
> > > You can define you archive as DERIVE with minimum value of 0. It
> > > will take care of the problem.
> > > 
> > 	Where did this idea come from? The _best_ solution is to set a
> > proper max.
> Nope.  That is a workaround to ignore problems.
> The best solution is to detect wraps AND resets.
	Which is impossible in all cases, but by setting a max you can get
it right in most cases... see my response to Tobi for formula's and examples
on the probability that using max gets it wrong.

> The next best thing is to ignore wraps and resets. That is the derive
> thingy.

	It depends on what you consider the biggest evil... setting all
legitimate wraps as "Unknown", or setting the occasional reset with an
incorrect (but not unlikely) value. The probability of having an incorrect
value for a reset is actually very low (<1%) for typical 32bit counter
usage, and non-existent for 64bit counters. 

> For interfaces normally not carrying a lot of traffic, ignoring an
> occasional REAL wrap is no big deal.  Incrementing that same counter
> from 2^32-3,750,000,000 to 2^32 may take a long time and thus a large
> window for error.

	But it is precisely these conditions where the probability of reset
resulting in an incorrect value is negligible. Once again, it depends on
what you consider the biggest evil...


Unsubscribe mailto:rrd-users-request at list.ee.ethz.ch?subject=unsubscribe
Help        mailto:rrd-users-request at list.ee.ethz.ch?subject=help
Archive     http://www.ee.ethz.ch/~slist/rrd-users
WebAdmin    http://www.ee.ethz.ch/~slist/lsg2.cgi

More information about the rrd-users mailing list