| 1 |
#!/usr/bin/env perl
|
| 2 |
|
| 3 |
use strict;
|
| 4 |
use Mysql;
|
| 5 |
use File::Copy;
|
| 6 |
use Getopt::Long;
|
| 7 |
use POSIX;
|
| 8 |
|
| 9 |
sub printHelp;
|
| 10 |
sub outputPset;
|
| 11 |
sub outputRunList;
|
| 12 |
sub getLocation;
|
| 13 |
sub outputCondor;
|
| 14 |
|
| 15 |
our $db = Mysql->connect ("cmshead", "ntuple", "osuT3User");
|
| 16 |
|
| 17 |
my %opt;
|
| 18 |
Getopt::Long::Configure ("bundling");
|
| 19 |
GetOptions (\%opt, "help|h");
|
| 20 |
my $argc = @ARGV;
|
| 21 |
|
| 22 |
printHelp () if $opt{"help"};
|
| 23 |
printHelp () if $argc != 4;
|
| 24 |
if (!(-e $ARGV[1]))
|
| 25 |
{
|
| 26 |
print "\"$ARGV[1]\" does not exist!\n";
|
| 27 |
exit;
|
| 28 |
}
|
| 29 |
if (-e $ARGV[2])
|
| 30 |
{
|
| 31 |
print "Directory \"$ARGV[2]\" already exists!\n";
|
| 32 |
print "Please delete it or specify another working directory.\n";
|
| 33 |
exit;
|
| 34 |
}
|
| 35 |
mkdir $ARGV[2];
|
| 36 |
my $nFiles = outputRunList ($ARGV[0], $ARGV[2]);
|
| 37 |
my $nJobs = $ARGV[3];
|
| 38 |
my $realNJobs = ceil ($nFiles / ceil ($nFiles / $nJobs));
|
| 39 |
outputPset ($ARGV[2]);
|
| 40 |
outputCondor ($ARGV[2], $realNJobs, $ARGV[0]);
|
| 41 |
copy ($ARGV[1], "$ARGV[2]/userConfig_cfg.py");
|
| 42 |
chdir $ARGV[2];
|
| 43 |
print "Submitting $realNJobs jobs to run on $nFiles files.\n";
|
| 44 |
system ("condor_submit condor.sub");
|
| 45 |
|
| 46 |
sub
|
| 47 |
outputPset
|
| 48 |
{
|
| 49 |
my $workingDir = shift;
|
| 50 |
|
| 51 |
open (PSET, ">$workingDir/config_cfg.py");
|
| 52 |
|
| 53 |
print PSET "import FWCore.ParameterSet.Config as cms\n";
|
| 54 |
print PSET "import OSUT3Analysis.DBTools.osusub_cfg as osusub\n";
|
| 55 |
print PSET "import re\n";
|
| 56 |
print PSET "import userConfig_cfg as pset\n";
|
| 57 |
print PSET "\n";
|
| 58 |
print PSET "fileName = pset.process.TFileService.fileName\n";
|
| 59 |
print PSET "fileName = fileName.pythonValue ()\n";
|
| 60 |
print PSET "fileName = fileName[1:(len (fileName) - 1)]\n";
|
| 61 |
print PSET "fileName = re.sub (r'^(.*)\\.([^\\.]*)\$', r'\\1_' + str (osusub.jobNumber) + r'.\\2', fileName)\n";
|
| 62 |
print PSET "pset.process.TFileService.fileName = fileName\n";
|
| 63 |
print PSET "\n";
|
| 64 |
print PSET "pset.process.source.fileNames = cms.untracked.vstring (osusub.runList)\n";
|
| 65 |
print PSET "pset.process.maxEvents.input = cms.untracked.int32 (-1)\n";
|
| 66 |
print PSET "process = pset.process\n";
|
| 67 |
|
| 68 |
close (PSET);
|
| 69 |
}
|
| 70 |
|
| 71 |
sub
|
| 72 |
outputRunList
|
| 73 |
{
|
| 74 |
my $dataset = shift;
|
| 75 |
my $workingDir = shift;
|
| 76 |
|
| 77 |
my ($location, $nFiles, $status, $crossSection) = getLocation ($dataset);
|
| 78 |
if ($status ne "present")
|
| 79 |
{
|
| 80 |
print "This dataset is not marked as present on the Tier 3!\n";
|
| 81 |
exit;
|
| 82 |
}
|
| 83 |
if (!(-e $location))
|
| 84 |
{
|
| 85 |
print "The database does not know where this dataset is!\n";
|
| 86 |
exit;
|
| 87 |
}
|
| 88 |
opendir (LOCATION, $location);
|
| 89 |
my @files = readdir (LOCATION);
|
| 90 |
closedir (LOCATION);
|
| 91 |
if (@files - 2 != $nFiles)
|
| 92 |
{
|
| 93 |
print "Number of files does not match database entry!\n";
|
| 94 |
exit;
|
| 95 |
}
|
| 96 |
open (RUNLIST, ">$workingDir/runList.py");
|
| 97 |
print RUNLIST "runList = [\n";
|
| 98 |
for (my $i = 0; $i < @files; $i++)
|
| 99 |
{
|
| 100 |
next if $files[$i] eq ".";
|
| 101 |
next if $files[$i] eq "..";
|
| 102 |
print RUNLIST "'file:$location/$files[$i]'";
|
| 103 |
print RUNLIST "," if $i + 1 != @files;
|
| 104 |
print RUNLIST "\n";
|
| 105 |
}
|
| 106 |
print RUNLIST "]";
|
| 107 |
close (RUNLIST);
|
| 108 |
if ($crossSection && $crossSection >= 0.0)
|
| 109 |
{
|
| 110 |
open (CROSS_SECTION, ">$workingDir/crossSectionInPicobarn.txt");
|
| 111 |
print CROSS_SECTION "$crossSection\n";
|
| 112 |
close (CROSS_SECTION);
|
| 113 |
}
|
| 114 |
|
| 115 |
return $nFiles;
|
| 116 |
}
|
| 117 |
|
| 118 |
sub
|
| 119 |
getLocation
|
| 120 |
{
|
| 121 |
my $dataset = shift;
|
| 122 |
|
| 123 |
my $results;
|
| 124 |
my $queryDataset = $dataset;
|
| 125 |
$queryDataset =~ s/\*/%/g;
|
| 126 |
$queryDataset =~ s/(.*)/%$1%/g;
|
| 127 |
my $query = "select dataset,user,creationTime,location,nFiles,status,crossSectionInPicobarn from ntuple where dataset like '$queryDataset' and status='present' order by creationTime";
|
| 128 |
$db->selectdb ("ntuple");
|
| 129 |
$results = $db->query ($query);
|
| 130 |
if ($results->numrows () == 1)
|
| 131 |
{
|
| 132 |
my @row = $results->fetchrow ();
|
| 133 |
return ($row[3], $row[4], $row[5], $row[6]);
|
| 134 |
}
|
| 135 |
if ($results->numrows () == 0)
|
| 136 |
{
|
| 137 |
print "Dataset does not exist on the Tier 3!\n";
|
| 138 |
exit;
|
| 139 |
}
|
| 140 |
my %map;
|
| 141 |
print "Found multiple datasets matching\n";
|
| 142 |
print "\"$dataset\":\n";
|
| 143 |
for (my $i = 1; $i <= $results->numrows (); $i++)
|
| 144 |
{
|
| 145 |
my @row = $results->fetchrow ();
|
| 146 |
$map{"$i"} = [$row[3], $row[4], $row[5], $row[6]];
|
| 147 |
printf "(%2d) $row[0]\n", $i;
|
| 148 |
print " created by $row[1] on $row[2]\n";
|
| 149 |
}
|
| 150 |
print "\nWhich dataset would you like to use?: ";
|
| 151 |
my $response = <STDIN>;
|
| 152 |
$response =~ s/[ \t\n]//g;
|
| 153 |
if (!(exists $map{$response}))
|
| 154 |
{
|
| 155 |
print "Your selection \"$response\" was not a valid option! Quitting.\n";
|
| 156 |
exit;
|
| 157 |
}
|
| 158 |
|
| 159 |
return ($map{$response}[0], $map{$response}[1], $map{$response}[2], $map{$response}[3]);
|
| 160 |
}
|
| 161 |
|
| 162 |
sub
|
| 163 |
outputCondor
|
| 164 |
{
|
| 165 |
my $workingDir = shift;
|
| 166 |
my $nJobs = shift;
|
| 167 |
my $dataset = shift;
|
| 168 |
|
| 169 |
my $cmsRun = `which cmsRun`;
|
| 170 |
open (SUB, ">$workingDir/condor.sub");
|
| 171 |
|
| 172 |
print SUB "Executable = $cmsRun\n";
|
| 173 |
print SUB "Universe = vanilla\n";
|
| 174 |
print SUB "Getenv = True\n";
|
| 175 |
print SUB "Arguments = config_cfg.py $nJobs \$(Process) $dataset\n";
|
| 176 |
print SUB "\n";
|
| 177 |
print SUB "Output = condor_\$(Process).out\n";
|
| 178 |
print SUB "Error = condor_\$(Process).err\n";
|
| 179 |
print SUB "Log = condor_\$(Process).log\n";
|
| 180 |
print SUB "\n";
|
| 181 |
print SUB "+IsLocalJob = true\n";
|
| 182 |
print SUB "Rank = TARGET.IsLocalSlot\n";
|
| 183 |
print SUB "\n";
|
| 184 |
print SUB "Queue $nJobs\n";
|
| 185 |
|
| 186 |
close (SUB);
|
| 187 |
}
|
| 188 |
|
| 189 |
sub
|
| 190 |
printHelp
|
| 191 |
{
|
| 192 |
my $exeName = $0;
|
| 193 |
$exeName =~ s/^.*\/([^\/]*)$/$1/;
|
| 194 |
|
| 195 |
print "Usage: $exeName [OPTION]... DATASET CONFIG DIRECTORY NJOBS\n";
|
| 196 |
print "Submits CMSSW jobs to the OSU Tier 3 compute nodes using Condor.\n";
|
| 197 |
print "\n";
|
| 198 |
printf "%-29s%s\n", " -h, --help", "print this help message";
|
| 199 |
print "\n";
|
| 200 |
print "The DATASET must exist in the Tier 3 ntuple database, and CONFIG must be a valid\n";
|
| 201 |
print "CMSSW python configuration which can be used with cmsRun. DIRECTORY is a working\n";
|
| 202 |
print "directory that is created and in which all output, both from the CMSSW jobs and\n";
|
| 203 |
print "from Condor, is placed. Finally, NJOBS is the number of Condor jobs that will\n";
|
| 204 |
print "be created.\n";
|
| 205 |
|
| 206 |
exit;
|
| 207 |
}
|
