[rrd-users] Re: Storimg 5 min data for 2 yr

Ketil Oftedahl ketil at oftedahl.no
Tue Feb 4 12:01:29 MET 2003

Stephan Harren wrote:
> I guess this happens because rrd (maybe,
> just guessing) averages the 5-min values to appropriate information fitting
> into the pixel parameters. But anyway, graphs  ! are ! different if you don't
> keep the max or keep the max data.
> Anyone tell me if I'm wrong.

No, you're not.  I was wrong (or maybe just thinking a bit too quick...)

The logic (if any) behind my thinking was that if you store each and
every sampl, the MIN, AVERAGE and MAX RRAs will all contain the same
values (no consolidation will take place when storing the data as long
as the storage interval is equal to the sample interval).  I then
assumed that it would be possible to specify MAX (or MIN) consolidation
when actually graphing the data, and I even tried to verify that by
looking at the src code but I didn't look hard enough or in the right
places.  rrdgraph can use min/average/max consolidation when graphing,
but requires an RRA of the same "kind".  (This is what I did not
discover in the first place.)
If you want to use MAX when graphing, you also need to have a MAX RRA.
If you try to use MAX in a rrdgrapg DEF without having a MAX RRA you
will get the error message :
 ERROR: the RRD does not contain an RRA matching the chosen CF

This means that even though the MIN/AVERAGE/MAX RRAs may contain
identical data you will still need them if you want to use the
corresponding consolidation function in rrdgraph.

This may look like a waste of space but disk space is (relatively) cheap
these days, and I guess that adding the possibility of graphing MIN or
MAX data from a AVERAGE RRA will introduce more challenges than it
solves - or complicate the syntax of rrdgraph...


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