[rrd-users] Re: rrdtool missed values <1.0

[William Owen]

<--quote-->
On Fri, Sep 01, 2006 at 12:09:40PM +0000, Thomas Peter wrote:

> i only get updates on monday-friday from 6:00-19:00 with a distance of
max. 
> 10 hours (on one of these days)
> but since i don't get any update for whole 3 days, i need a heartbeat
of 
> 3*24*60*60 = 259200s
> 
> what would the drawback be if i use a heartbeat of one week?

The entire interval between last update and current update would be
filled with the rate calculated by rrdtool based on your updates.

If that's not what you want, you need another approach.

> the shortest time between updates is given by the step-size. the
longest by the 
> heartbeat?

No.  You can update multiple times during one step if you want.  The
only restriction is that the updates cannot be at exactly the same time.

<--quote-->

Not to discourage this discussion of the relationship between heartbeat
and step but:
Thomas, you state that you are parsing a log file looking for markers
(start and end) that you then calculate the difference of and log into
RRD?  If I'm reading that right, you are doing a periodic poll of this
log file.  I would do an update to your RRD every time this script is
run.  If you do not find a start marker, update with 0 (or nan depending
how you want it to show up).  If you find a start marker and no end
marker, update with now-start.  If you find a start and end marker,
update with end-start.  This should produce the steadily increasing
slope that drops to 0 when the condition is not met.  In this scenario
you would want a step of <script interval> and a heartbeat of <script
interval + some seconds depending on predictability of script>.

William

--
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://lists.ee.ethz.ch/rrd-users
WebAdmin    http://lists.ee.ethz.ch/lsg2.cgi