condor_plot

#!/usr/bin/perl
use Date::Format;
use Date::Parse;
use Time::Local;

# Messy script for plotting variables from condor history using RRDTool
# Note that you have to read the inline comments to make any sense of it
#
# author: Ian Gable <igable@uvic.ca>
#
#

$year = "2010";
$month = "6";
$day = "1";
$hour = "0";
$minute = "0";

#$condor_start = `condor_history | head -n2 | tail -n1 | awk '{print \$7 " " \$8}'`;
#if ($condor_start =~ m#(\d+)/(\d+) (\d+):(\d+)#) {
	#$month = $1 -1;
	#$day = $2 -1;
#}

# set this variable to be the start time of the graph
$completedsince =  timelocal(0,$minute,$hour,$day,$month,$year);

# creates a rrd database that does no averaging. Every data point is shown.
system "rrdtool create test.rrd  --start=$completedsince --step=86400 DS:jobs:GAUGE:86400:0:10000 RRA:AVERAGE:0.5:1:5000";

# Pull the thing you want to plot out of the condor_history.
# To see a list of available things look at: condor_history -l | head -n 100
system 'condor_history -format "%i\n" CompletionDate > temp';


open TEMP, "<temp" or die "can't read temp file";
%histogram = ();
while($line = <TEMP>){
	# ($date, $jobid) = split(" " ,$line,2);
	$date =$line;
	if($date != "0" && $date > $completedsince){

		#pddrint "$line\n";
		# print "Date:$date JobID:$job\n";
 		#print "Date:".localtime($date)." JobID:$job\n";
 		#($sec,$min,$hour,$mday,$mon,$year,$wday,$yday,$isdst) = localtime($date);
 		#$year = $year+1900;
 		#$mon = $mon+1;
 		#$datestring = $year."-".$mon."-".$mday;
		#$datestring =  time2str("%Y-%m-%d",$date);

		$datestring =  time2str("%Y-%m-%d",$date);

		#this line bins the dates into day wide bins.
 		$histogram{$datestring} = $histogram{$datestring}+1; 

	}	
}

open OUTPUT, ">condor.csv";

#go through the hash (i.e. histogram) in a sorted order
foreach $day ( sort keys %histogram )
{
  # write out the bins of the histogram to a CSV file.
  print OUTPUT "$day, $histogram{$day}\n";

  #update the RRD database
  system 'rrdtool update test.rrd '.str2time($day).':'.$histogram{$day}; 
}

$now = time();

#--x-grid DAY:1:WEEK:1:WEEK:1:0:\%b


# actually make the RRD plot.
system "rrdtool graph jobs.png --title 'Completed Cloud Jobs Per Day'"
      ." --vertical-label 'Jobs' --font AXIS:12 --font TITLE:16 --font UNIT:12"
      ." --width 800 --height 600  --start=$completedsince --end=$now"
      ."  DEF:myjobs=test.rrd:jobs:AVERAGE AREA:myjobs#000AF3";