[rrd-users] rrdfetch resolution problem

Matt Pounsett matt at conundrum.com
Wed May 2 22:30:37 CEST 2007

Hash: SHA1

Thanks for the followup, Sam.

On 2007-May-02, at 15:33, Sam Umbach wrote:

> It's my understanding that fetch will use the highest resolution RRA
> containing the entire time interval you request (start time through
> end time).

Since the step value for the RRD is 300, shouldn't the "highest  
resolution RRA" be 300 (or possibly less)?  Based on the rrdinfo  
output I submitted earlier, I think the highest resolution should be  
at least 300 seconds, but I seem to be only able to get 3600 seconds  
out of it.

> If you perform an rrdtool dump, do you see the values
> you're looking for?

Yes, though the start/end restrictions don't seem to work.. The  
AVERAGE rra section will only give me data back to April 25, even  
though I know there's data going back to the beginning of January,  
and the MIN and MAX rra sections give me data from 2005, which I know  
isn't in the RRD.

This is doing:
% rrdtool dump server1.rrd -s 1167627600 -e 1175399700

> We need more detail about your rrd file
> (specifically the RRA definitions) to be helpful.  Please send the
> rrdtool create command you used to create the rrd or a full rrd dump.

The following code is used by a CGI that receives data submissions  
from remote sites to create RRDs that don't exist yet.  All of the  
RRDs were created with this same code.  $start_time is the time stamp  
of the earliest chunk of data in the submission, and the rest should  
be self explanatory.

my @TrackRRTypes = qw( A AAAA ANY CNAME MX NS PTR SOA );
my @args = ( "rrds/$host.rrd",
         qw(--step 300 ),
         "-b", $start_time-300,
         (map { "DS:$_:ABSOLUTE:600:0:10000000" } @TrackRRTypes ),
RRDs::create( @args );

Version: GnuPG v1.4.6 (Darwin)


More information about the rrd-users mailing list