| 4 |
|
use Mysql; |
| 5 |
|
use File::Copy; |
| 6 |
|
use Getopt::Long; |
| 7 |
< |
sub printHelp; |
| 7 |
> |
use POSIX; |
| 8 |
|
|
| 9 |
+ |
sub printHelp; |
| 10 |
|
sub outputPset; |
| 11 |
|
sub outputRunList; |
| 12 |
|
sub getLocation; |
| 13 |
|
sub outputCondor; |
| 14 |
|
|
| 15 |
< |
our $db; |
| 15 |
> |
our $db = Mysql->connect ("cmshead", "ntuple", "osuT3User"); |
| 16 |
|
|
| 17 |
|
my %opt; |
| 18 |
|
Getopt::Long::Configure ("bundling"); |
| 33 |
|
exit; |
| 34 |
|
} |
| 35 |
|
mkdir $ARGV[2]; |
| 36 |
< |
outputRunList ($ARGV[0], $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], $ARGV[3]); |
| 40 |
> |
outputCondor ($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 |
| 74 |
|
my $dataset = shift; |
| 75 |
|
my $workingDir = shift; |
| 76 |
|
|
| 77 |
< |
my ($location, $nFiles, $status) = getLocation ($dataset); |
| 78 |
< |
if ($status ne "present") |
| 79 |
< |
{ |
| 80 |
< |
print "This dataset is not marked as present on the Tier 3!\n"; |
| 81 |
< |
exit; |
| 77 |
> |
my $location; |
| 78 |
> |
my $nFiles; |
| 79 |
> |
my $status; |
| 80 |
> |
my $crossSection; |
| 81 |
> |
my $isLocation = 0; |
| 82 |
> |
$isLocation = 1 if -e $dataset; |
| 83 |
> |
if (!$isLocation) |
| 84 |
> |
{ |
| 85 |
> |
($location, $nFiles, $status, $crossSection) = getLocation ($dataset); |
| 86 |
> |
if ($status ne "present") |
| 87 |
> |
{ |
| 88 |
> |
print "This dataset is not marked as present on the Tier 3!\n"; |
| 89 |
> |
print "Continue anyway? (y/N): "; |
| 90 |
> |
my $response = <STDIN>; |
| 91 |
> |
$response =~ s/\n//g; |
| 92 |
> |
exit if !$response || lc ($response) ne "y"; |
| 93 |
> |
} |
| 94 |
> |
if (!(-e $location)) |
| 95 |
> |
{ |
| 96 |
> |
print "The database does not know where this dataset is!\n"; |
| 97 |
> |
exit; |
| 98 |
> |
} |
| 99 |
|
} |
| 100 |
< |
if (!(-e $location)) |
| 100 |
> |
else |
| 101 |
|
{ |
| 102 |
< |
print "The database does not know where this dataset is!\n"; |
| 82 |
< |
exit; |
| 102 |
> |
$location = $dataset; |
| 103 |
|
} |
| 104 |
|
opendir (LOCATION, $location); |
| 105 |
|
my @files = readdir (LOCATION); |
| 106 |
|
closedir (LOCATION); |
| 107 |
< |
if (@files - 2 != $nFiles) |
| 107 |
> |
if (!$isLocation && @files - 2 != $nFiles) |
| 108 |
|
{ |
| 109 |
|
print "Number of files does not match database entry!\n"; |
| 110 |
< |
exit; |
| 110 |
> |
print "Continue anyway? (y/N): "; |
| 111 |
> |
my $response = <STDIN>; |
| 112 |
> |
$response =~ s/\n//g; |
| 113 |
> |
exit if !$response || lc ($response) ne "y"; |
| 114 |
> |
} |
| 115 |
> |
elsif ($isLocation) |
| 116 |
> |
{ |
| 117 |
> |
$nFiles = @files - 2; |
| 118 |
|
} |
| 119 |
|
open (RUNLIST, ">$workingDir/runList.py"); |
| 120 |
|
print RUNLIST "runList = [\n"; |
| 128 |
|
} |
| 129 |
|
print RUNLIST "]"; |
| 130 |
|
close (RUNLIST); |
| 131 |
+ |
if ($crossSection && $crossSection >= 0.0) |
| 132 |
+ |
{ |
| 133 |
+ |
open (CROSS_SECTION, ">$workingDir/crossSectionInPicobarn.txt"); |
| 134 |
+ |
print CROSS_SECTION "$crossSection\n"; |
| 135 |
+ |
close (CROSS_SECTION); |
| 136 |
+ |
} |
| 137 |
+ |
|
| 138 |
+ |
return $nFiles; |
| 139 |
|
} |
| 140 |
|
|
| 141 |
|
sub |
| 144 |
|
my $dataset = shift; |
| 145 |
|
|
| 146 |
|
my $results; |
| 147 |
< |
$db = Mysql->connect ("cmshead", "ntuple", "osuT3User"); |
| 148 |
< |
my $query = "select user,creationTime,location,nFiles,status from ntuple where dataset='$dataset' order by lastUpdateTime"; |
| 147 |
> |
my $queryDataset = $dataset; |
| 148 |
> |
$queryDataset =~ s/\*/%/g; |
| 149 |
> |
$queryDataset =~ s/(.*)/%$1%/g; |
| 150 |
> |
my $query = "select dataset,user,creationTime,location,nFiles,status,crossSectionInPicobarn from ntuple where dataset like '$queryDataset' order by creationTime"; |
| 151 |
|
$db->selectdb ("ntuple"); |
| 152 |
|
$results = $db->query ($query); |
| 153 |
|
if ($results->numrows () == 1) |
| 154 |
|
{ |
| 155 |
|
my @row = $results->fetchrow (); |
| 156 |
< |
return ($row[2], $row[3], $row[4]); |
| 156 |
> |
return ($row[3], $row[4], $row[5], $row[6]); |
| 157 |
|
} |
| 158 |
|
if ($results->numrows () == 0) |
| 159 |
|
{ |
| 166 |
|
for (my $i = 1; $i <= $results->numrows (); $i++) |
| 167 |
|
{ |
| 168 |
|
my @row = $results->fetchrow (); |
| 169 |
< |
$map{"$i"} = [$row[2], $row[3], $row[4]]; |
| 170 |
< |
print "($i) created by $row[0] on $row[1]\n"; |
| 169 |
> |
$map{"$i"} = [$row[3], $row[4], $row[5], $row[6]]; |
| 170 |
> |
printf "(%2d) $row[0]\n", $i; |
| 171 |
> |
print " created by $row[1] on $row[2]\n"; |
| 172 |
|
} |
| 173 |
|
print "\nWhich dataset would you like to use?: "; |
| 174 |
|
my $response = <STDIN>; |
| 179 |
|
exit; |
| 180 |
|
} |
| 181 |
|
|
| 182 |
< |
return ($map{$response}[0], $map{$response}[1], $map{$response}[2]); |
| 182 |
> |
return ($map{$response}[0], $map{$response}[1], $map{$response}[2], $map{$response}[3]); |
| 183 |
|
} |
| 184 |
|
|
| 185 |
|
sub |
| 187 |
|
{ |
| 188 |
|
my $workingDir = shift; |
| 189 |
|
my $nJobs = shift; |
| 190 |
+ |
my $dataset = shift; |
| 191 |
|
|
| 192 |
|
my $cmsRun = `which cmsRun`; |
| 193 |
|
open (SUB, ">$workingDir/condor.sub"); |
| 195 |
|
print SUB "Executable = $cmsRun\n"; |
| 196 |
|
print SUB "Universe = vanilla\n"; |
| 197 |
|
print SUB "Getenv = True\n"; |
| 198 |
< |
print SUB "Arguments = config_cfg.py $nJobs \$(Process)\n"; |
| 198 |
> |
print SUB "Arguments = config_cfg.py True $nJobs \$(Process) $dataset\n"; |
| 199 |
|
print SUB "\n"; |
| 200 |
|
print SUB "Output = condor_\$(Process).out\n"; |
| 201 |
|
print SUB "Error = condor_\$(Process).err\n"; |
| 202 |
|
print SUB "Log = condor_\$(Process).log\n"; |
| 203 |
|
print SUB "\n"; |
| 204 |
+ |
print SUB "+IsLocalJob = true\n"; |
| 205 |
+ |
print SUB "Rank = TARGET.IsLocalSlot\n"; |
| 206 |
+ |
print SUB "\n"; |
| 207 |
|
print SUB "Queue $nJobs\n"; |
| 208 |
|
|
| 209 |
|
close (SUB); |
| 215 |
|
my $exeName = $0; |
| 216 |
|
$exeName =~ s/^.*\/([^\/]*)$/$1/; |
| 217 |
|
|
| 218 |
< |
print "Usage: $exeName [OPTION] DATASET CONFIG DIRECTORY NJOBS\n"; |
| 218 |
> |
print "Usage: $exeName [OPTION]... DATASET CONFIG DIRECTORY NJOBS\n"; |
| 219 |
> |
print " or: $exeName [OPTION]... LOCATION CONFIG DIRECTORY NJOBS\n"; |
| 220 |
|
print "Submits CMSSW jobs to the OSU Tier 3 compute nodes using Condor.\n"; |
| 221 |
|
print "\n"; |
| 222 |
|
printf "%-29s%s\n", " -h, --help", "print this help message"; |
| 223 |
|
print "\n"; |
| 224 |
< |
print "The DATASET must exist in the Tier 3 ntuple database, and CONFIG must be a valid\n"; |
| 225 |
< |
print "CMSSW python configuration which can be used with cmsRun. DIRECTORY is a working\n"; |
| 224 |
> |
print "The first argument must be either a DATASET registered in the Tier 3 ntuple\n"; |
| 225 |
> |
print "database or a LOCATION which exists on disk. CONFIG must be a valid CMSSW\n"; |
| 226 |
> |
print "python configuration which can be used with cmsRun. DIRECTORY is a working\n"; |
| 227 |
|
print "directory that is created and in which all output, both from the CMSSW jobs and\n"; |
| 228 |
|
print "from Condor, is placed. Finally, NJOBS is the number of Condor jobs that will\n"; |
| 229 |
|
print "be created.\n"; |
