[rrd-users] setting last_update

[Alex van den Bogaerdt]

On Thu, Feb 15, 2007 at 09:18:05AM -0500, Dan Langille wrote:

> FWIW, here is a test case I created by looking at what the symon code 
> was doing:
> 
> rrdtool create cpu0.rrd --step=5 --start=0 DS:user:GAUGE:5:0:100 
> DS:nice:GAUGE:5:0:100 DS:system:GAUGE:5:0:100 
> DS:interrupt:GAUGE:5:0:100 DS:idle:GAUGE:5:0:100 
> RRA:AVERAGE:0.5:1:34560 RRA:AVERAGE:0.5:360:672 
> RRA:AVERAGE:0.5:1440:600 RRA:AVERAGE:0.5:17280:600 
> RRA:MAX:0.5:1:34560 RRA:MAX:0.5:360:672 RRA:MAX:0.5:1440:600 
> RRA:MAX:0.5:17280:600 RRA:MIN:0.5:1:34560 RRA:MIN:0.5:360:672 
> RRA:MIN:0.5:1440:600 RRA:MIN:0.5:17280:600

Looking reasonable.  I only dislike "--start=0", which is
probably harmless.

You could try a much more simple case as well:

rrdtool create test.rrd DS:x:GAUGE:600:U:U RRA:AVERAGE:0:1:5
rrdtool dump test.rrd


> I didn't supply the whole dump as it is 109,870 lines....

Heh... you thought "head -7" was to show off ?  :)

> The last date is 31900-02-14 16:00:00 PST / -38934592 and the first 
> date is 31900-02-13 00:00:05 PST / -38948991

Negative numbers mixed with numbers larger than 32 bit.  Sounds
like a signed/unsigned integer bug, undetected so far, and now
the bug starts biting.

I could be totally wrong here.  I'm no programmer.

-- 
Alex van den Bogaerdt
http://www.vandenbogaerdt.nl/rrdtool/