[rrd-users] Another question about MAX value

[Sorrell, Al]

I _thought_ I understood this, but...

My user wants to be able to drill down in detail for information anytime in
the past 13 months, so I create a big RRD with just one AVERAGE and one MAX
consolidation for 400 days' worth of data at 10-minute samples:
    $incr=10;
    $keep_raw = 400*(3600/$incr)*24;    
    $keep_raw_steps=1;
    $hb=2*$incr;          #RRD Heartbeat = 2*increment
    RRDs::create ($rrdfile,
                "--start",$now-1,
                "--step",$incr,
                "DS:busy:GAUGE:$hb:0:U",
                "DS:idle:GAUGE:$hb:0:U",
                "DS:dis:GAUGE:$hb:0:U",
                "RRA:AVERAGE:0.5:$keep_raw_steps:$keep_raw",
                "RRA:MAX:0.5:$keep_raw_steps:$keep_raw"
     );

I put a mess of data into the RRD from earlier .csv file and then plot 4
graphs, using varying start times of -24 hours, -7 days, -31 days and -366
days ago, plot the data and show the MAX value for the period graphed:

  $now=time();            #Nominal sample time
  $now=$now-($now % $incr);# fix minor time offset so time is multiple of
$incr
  @sec_ago=(3600*24, 3600*24*7, 3600*24*31, 3600*24*366);       #1day, 1
week, 1 month, 1 year
  $ngraph=@sec_ago;
  for ($graph=0; $graph < $ngraph; $graph++) {
    $startplot=$now - $sec_ago[$graph];
  ...
    RRDs::graph "$outfile",
        "--title","$title",
        "--start","$startplot",
        "--end","$now",
        "--imgformat","PNG",
        "--width","600",
        "--height","200",
        "--vertical-label","Channels",
        "DEF:dis=$rrdfile:dis:AVERAGE",
        "DEF:busy=$rrdfile:busy:AVERAGE",
        "DEF:idle=$rrdfile:idle:AVERAGE",
        "HRULE:$thresh#0000c8:Warning",
        "AREA:dis#4040ff:Disabled",
        "STACK:busy#fef2a:Busy",
        "STACK:idle#00cc40:Idle\\c",
        "COMMENT:\\s",
        "COMMENT:$comment\\c",
        "GPRINT:busy:MAX:Max %1.0lf circuits busy for range plotted.\\c";
  }

The periods plotted appear correct but the MAX values on the graphs are
inconsistent -
	1 day		7 days	31 days	366 days
MAX	36		56		55		37

I would have expected that since the 31 days and 366 days plots both are a
superset of the 7 days range, that the MAX value should be AT LEAST that
seen during the 7 day plot. Since there is (should be) no aggregation of the
data, I would have expected to see the same value for the last three plots -
I think I'm averaging over 1 sample, which should be the sample value. Not
sure what that would do. 


Thanks,
Al

--
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