[rrd-users] Re: CDEF - creating a increasing series

Alex van den Bogaerdt alex at ergens.op.het.net
Wed Sep 8 15:46:04 MEST 2004


On Wed, Sep 08, 2004 at 02:59:01PM +0200, ciaran deignan wrote:

> humm... would still be "nice" if RRD provided the current effective step
> during the calculation phase.

If you *know* you are going to graph 24 hours, and if you *know* the
image will be 480 pixels wide, you *know* 20 pixels will be one hour.
This also means 10 pixels will be 30 minutes, and so on.

"--end 00:00 --start end-24h --width 480" will do this.
"--end now --start end-24h" will not.

You _also_ know how your database looks.  You know about RRA steps and
such.  RRDtool will do a best match.  5 minutes per bucket won't fly,
because that's (24 * 60 minutes / 5 minutes = ) 288 intervals which
won't fit in 480 pixels (480 / 288 is not an integer).

I'm not sure how the current RRDtool will deal with this, let's just
assume it isn't going to be selected as best choice.  Remember I may
be wrong on this when you see unexpected results.

30 minutes per interval results in (24*60 / 30 = ) 48 intervals.
48 intervals in 480 pixels works nice.  So far so good.

2 hours per interval results in (24*60 / 120 = ) 12 intervals. This
does also fit on 480 pixels.  Another choice.

24 hours per interval will also work.

Next to look at:  boundaries.  The 24 hours per interval RRA will
most likely NOT start and end at midnight localtime.  That does not
qualify as "best match" because there are better choices.

Next to look at:  RRA size.  Suppose the RRA holding 30 minutes per
interval has only 30 rows.  In that case, the 2 hours per interval RRA
will have a more complete set of data and so that RRA is a better match
than the 30-minutes RRA.

If the image would span half the time (12 hours), either the 30-minute
RRA or the 5-minute RRA would be the best choice.  This depends on the
best-match algorithm.  I think this can be found in the docs.  If not,
use the source Luke.

Now suppose "--end 00:05 --start end-24h" would be queried.  That would
influence the best-match algorithm considerably.

It's even worse when "--end 00:03 --start end-24h" is queried, where
"00:03" could have been written as "now".

When using "now", today it could have been "23:59", tomorrow it will be
"00:00", the day after it will be "00:05".  Three different result sets.

As you can see: no Voodoo/Judo.  Just logic.

> Thanks for the help Alex,

Thanks for appreciating it.

Alex
-- 
You want an answer?  You'd better follow the following guidelines!
Linesize well below 80 chars. Reply to the list, not to me. Trim
irrelevant lines. Answer below the remaining paragraph.

--
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