[rrd-users] First gauge value isn't stored - what am I doing wrong?

Simon Hobson linux at thehobsons.co.uk
Tue Dec 1 08:52:26 CET 2009

Joachim Schrod wrote:

>  >>     rrdtool create test.rrd --start '00:00 24.11.2009' \
>  >>         --step 86400 DS:count:GAUGE:172800:0:U RRA:MAX:0.9:1:30

>Thanks, thanks! localtime was the culprit.

No it wasn't. Your first sample period was not complete, your first 
data finished 30 minutes before the end of the sample period. You 
should find that future periods will not have a problem - as long as 
you allow another day for the sample periods to complete ! Note that 
all sample periods start at n*S from unix epoch (where s is sample 
period) - so specifying a full day (86400s) means sample periods will 
ALWAYS start/finish at midnight UTC. Local time is never used except 
for display.

You should read Alex's tutorial on normalisation.

>I'm at CET, i.e., test.rrd hat actually a start time of 23:00 UTC
>and not of 00:00 UTC. Thus, the jobs ran at 23:30 UTC, not at
>00:30. If I adapt start time and measurement time to proper UTC
>entries, it works. I.e., I'll have to start my cron job not earlier
>than 03:00 local clock (to care for DST switches) and it should work.

If you want to graph by day in your timezone, then you might have to 
store data by the hour, and aggregate it at graphing time. If having 
your data days one or two hours out from local days isn't an issue 
then you can carry on as you are. Ideally, do your data collection 
just after midnight UTC - that way you minimise normalisation while 
getting your data available to graph as soon as possible.

Simon Hobson

Visit http://www.magpiesnestpublishing.co.uk/ for books by acclaimed
author Gladys Hobson. Novels - poetry - short stories - ideal as
Christmas stocking fillers. Some available as e-books.

More information about the rrd-users mailing list