[ViewVC] Annotation of: cvsroot/UserCode/OSUT3Analysis/DBTools/scripts/osusub

#!/usr/bin/env perl

use strict;
use Mysql;
use File::Copy;
use Getopt::Long;
use POSIX;
use Term::ANSIColor;

sub printHelp;
sub outputPset;
sub outputRunList;
sub getLocation;
sub outputCondor;

our $db = Mysql->connect ("cmshead", "ntuple", "osuT3User");

my %opt;
Getopt::Long::Configure ("bundling");
GetOptions (\%opt, "label|l=s", "maxEvents|m=s", "help|h");
my $argc = @ARGV;

printHelp () if $opt{"help"};
printHelp () if $argc != 3 && $argc != 4;
my $dataset;
my $config;
my $directory;
my $nJobs;
if ($argc == 3)
  {
    $dataset = "";
    $config = $ARGV[0];
    $directory = $ARGV[1];
    $nJobs = $ARGV[2];
  }
if ($argc == 4)
  {
    $dataset = $ARGV[0];
    $config = $ARGV[1];
    $directory = $ARGV[2];
    $nJobs = $ARGV[3];
  }
if (!(-e $config))
  {
    print "\"$config\" does not exist!\n";
    exit;
  }
if (-e $directory)
  {
    print "Directory \"$directory\" already exists!\n";
    print "Please delete it or specify another working directory.\n";
    exit;
  }
mkdir $directory;
my $nFiles = outputRunList ($dataset, $directory);
my $realNJobs = $nJobs;
$realNJobs = ceil ($nFiles / ceil ($nFiles / $realNJobs)) if $nFiles;
my $eventsPerJob = -1;
$eventsPerJob = ceil ($opt{"maxEvents"} / $realNJobs) if $opt{"maxEvents"} && $opt{"maxEvents"} >= 0;
my $realMaxEvents = $eventsPerJob * $realNJobs;
outputPset ($directory, $dataset, $opt{"maxEvents"}, $eventsPerJob);
outputCondor ("$ENV{'CMSSW_BASE'}/src/OSUT3Analysis/DBTools/data/condor.sub", $directory, $realNJobs, $dataset, $opt{"label"});
copy ($config, "$directory/userConfig_cfg.py");
chdir $directory;
print "Submitting $realNJobs jobs to run on $realMaxEvents events in $nFiles files.\n" if $realMaxEvents >= 0;
print "Submitting $realNJobs jobs to run on all events in $nFiles files.\n" if $realMaxEvents < 0;
system ("LD_LIBRARY_PATH=/usr/lib64/condor:\$LD_LIBRARY_PATH condor_submit condor.sub");

sub
outputPset
{
  my $workingDir = shift;
  my $dataset = shift;
  my $nEvents = shift;
  my $eventsPerJob = shift;

  open (PSET, ">$workingDir/config_cfg.py");

  print PSET "import FWCore.ParameterSet.Config as cms\n";
  print PSET "import OSUT3Analysis.DBTools.osusub_cfg as osusub\n";
  print PSET "import re\n";
  print PSET "import userConfig_cfg as pset\n";
  print PSET "\n";
  print PSET "fileName = pset.process.TFileService.fileName\n";
  print PSET "fileName = fileName.pythonValue ()\n";
  print PSET "fileName = fileName[1:(len (fileName) - 1)]\n";
  print PSET "fileName = re.sub (r'^(.*)\\.([^\\.]*)\$', r'\\1_' + str (osusub.jobNumber) + r'.\\2', fileName)\n";
  print PSET "pset.process.TFileService.fileName = fileName\n";
  print PSET "\n";
  if ($dataset)
    {
      print PSET "pset.process.source.fileNames = cms.untracked.vstring (osusub.runList)\n";
      print PSET "pset.process.maxEvents.input = cms.untracked.int32 ($eventsPerJob)\n" if $eventsPerJob < 0;
    }
  print PSET "pset.process.maxEvents.input = cms.untracked.int32 ($eventsPerJob)\n" if $eventsPerJob >= 0;
  print PSET "process = pset.process\n";

  close (PSET);
}

sub
outputRunList
{
  my $dataset = shift;
  my $workingDir = shift;

  return 0 if !$dataset;
  my $location;
  my $nFiles;
  my $status;
  my $crossSection;
  my $isLocation = 0;
  $isLocation = 1 if -e $dataset;
  if (!$isLocation)
    {
      ($location, $nFiles, $status, $crossSection) = getLocation ($dataset);
      if ($status ne "present")
        {
          print "This dataset is not marked as present on the Tier 3!\n";
          print "Continue anyway? (y/N): ";
          my $response = <STDIN>;
          $response =~ s/\n//g;
          exit if !$response || lc ($response) ne "y";
        }
      if (!(-e $location))
        {
          print "The database does not know where this dataset is!\n";
          exit;
        }
    }
  else
    {
      $location = $dataset;
    }
  opendir (LOCATION, $location);
  my @files = readdir (LOCATION);
  closedir (LOCATION);
  if (!$isLocation && @files - 2 != $nFiles)
    {
      print "Number of files does not match database entry!\n";
      print "Continue anyway? (y/N): ";
      my $response = <STDIN>;
      $response =~ s/\n//g;
      exit if !$response || lc ($response) ne "y";
    }
  elsif ($isLocation)
    {
      $nFiles = @files - 2;
    }
  open (RUNLIST, ">$workingDir/runList.py");
  print RUNLIST "runList = [\n";
  for (my $i = 0; $i < @files; $i++)
    {
      next if $files[$i] eq ".";
      next if $files[$i] eq "..";
      print RUNLIST "'file:$location/$files[$i]'";
      print RUNLIST "," if $i + 1 != @files;
      print RUNLIST "\n";
    }
  print RUNLIST "]";
  close (RUNLIST);
  if ($crossSection && $crossSection >= 0.0)
    {
      open (CROSS_SECTION, ">$workingDir/crossSectionInPicobarn.txt");
      print CROSS_SECTION "$crossSection\n";
      close (CROSS_SECTION);
    }

  return $nFiles;
}

sub
getLocation
{
  my $dataset = shift;

  my $results;
  my $queryDataset = $dataset;
  $queryDataset =~ s/\*/%/g;
  $queryDataset =~ s/(.*)/%$1%/g;
  my $query = "select dataset,user,creationTime,location,nFiles,status,crossSectionInPicobarn,higherOrderCrossSectionInPicobarn from ntuple where dataset like '$queryDataset' order by creationTime";
  $db->selectdb ("ntuple");
  $results = $db->query ($query);
  if ($results->numrows () == 1)
    {
      my @row = $results->fetchrow ();
      return ($row[3], $row[4], $row[5], $row[7]) if $row[7];
      return ($row[3], $row[4], $row[5], $row[6]) if !$row[7];
    }
  if ($results->numrows () == 0)
    {
      print "Dataset does not exist on the Tier 3!\n";
      exit;
    }
  my %map;
  print "Found multiple datasets matching\n";
  print "\"$dataset\":\n";
  for (my $i = 1; $i <= $results->numrows (); $i++)
    {
      my @row = $results->fetchrow ();
      $row[2] =~ s/([^ ]*) [^ ]*/$1/g;
      $map{"$i"} = [$row[3], $row[4], $row[5], $row[7]] if $row[7];
      $map{"$i"} = [$row[3], $row[4], $row[5], $row[6]] if !$row[7];
      printf "(%2d) $row[0]\n", $i;
      print "     (";
      print color "green" if $row[5] eq "present";
      print color "bold yellow" if $row[5] eq "submitted";
      print color "bold red" if $row[5] eq "created" or $row[5] eq "cancelled" or $row[5] eq "deprecated";
      print $row[5];
      print color "reset";
      print ") created by $row[1] on $row[2]\n";
    }
  print "\nWhich dataset would you like to use?: ";
  my $response = <STDIN>;
  $response =~ s/[ \t\n]//g;
  if (!(exists $map{$response}))
    {
      print "Your selection \"$response\" was not a valid option! Quitting.\n";
      exit;
    }

  return ($map{$response}[0], $map{$response}[1], $map{$response}[2], $map{$response}[3]);
}

sub
outputCondor
{
  my $condorFileName = shift;
  my $workingDir = shift;
  my $nJobs = shift;
  my $dataset = shift;
  my $label = shift;

  my $cmsRun = `which cmsRun`;
  my $condorFile = "";

  if (!(-e $condorFileName))
    {
      my $arguments = "Arguments               = config_cfg.py True $nJobs \$(Process)";
      $arguments .= " $dataset" if $dataset;
      $arguments .= " NULL" if !$dataset;
      $arguments .= " $label" if $label;
      $arguments .= " NULL" if !$label;
      $arguments .= "\n";

      $condorFile .= "Executable              = $cmsRun\n";
      $condorFile .= "Universe                = vanilla\n";
      $condorFile .= "Getenv                  = True\n";
      $condorFile .= $arguments;
      $condorFile .= "\n";
      $condorFile .= "Output                  = condor_\$(Process).out\n";
      $condorFile .= "Error                   = condor_\$(Process).err\n";
      $condorFile .= "Log                     = condor_\$(Process).log\n";
      $condorFile .= "\n";
      $condorFile .= "+IsLocalJob             = true\n";
      $condorFile .= "Rank                    = TARGET.IsLocalSlot\n";
      $condorFile .= "\n";
      $condorFile .= "Queue $nJobs\n";
    }
  else
    {
      open (SUB, "<$condorFileName");
      my @condorFile = <SUB>;
      close (SUB);
      $condorFile = join ("", @condorFile);
      $condorFile =~ s/\$cmsRun/$cmsRun/g;
      $condorFile =~ s/\$nJobs/$nJobs/g;
      $condorFile =~ s/\$dataset/$dataset/g if $dataset;
      $condorFile =~ s/\$dataset/NULL/g if !$dataset;
      $condorFile =~ s/\$label/$label/g if $label;
      $condorFile =~ s/\$label/NULL/g if !$label;
    }

  open (SUB, ">$workingDir/condor.sub");
  print SUB $condorFile;
  close (SUB);
}

sub
printHelp
{
  my $exeName = $0;
  $exeName =~ s/^.*\/([^\/]*)$/$1/;

  print "Usage: $exeName [OPTION]... [DATASET | LOCATION] CONFIG DIRECTORY NJOBS\n";
  print "Submits CMSSW jobs to the OSU Tier 3 compute nodes using Condor.\n";
  print "\n";
  printf "%-29s%s\n", "  -h, --help", "print this help message";
  printf "%-29s%s\n", "  -l, --label LABEL", "give the dataset a short label";
  printf "%-29s%s\n", "  -m, --maxEvents N", "only run over N events in the dataset; default is";
  printf "%-29s%s\n", " ", "to run over all events";
  print "\n";
  print "The optional first argument must be either a DATASET registered in the Tier 3\n";
  print "ntuple database or a LOCATION which exists on disk. CONFIG must be a valid\n";
  print "CMSSW python configuration which can be used with cmsRun. DIRECTORY is a\n";
  print "working directory that is created and in which all output, both from the CMSSW\n";
  print "jobs and from Condor, is placed. Finally, NJOBS is the number of Condor jobs\n";
  print "that will be created.\n";

  exit;
}
Revision:	1.19
Committed:	Thu Jun 6 20:53:32 2013 UTC (11 years, 11 months ago) by ahart
Branch:	MAIN
Changes since 1.18:	+1 -1 lines
Log Message:	Got rid of the "no version information available" warning from condor_submit.
#	User	Rev	Content
1	ahart	1.1	#!/usr/bin/env perl
2
3			use strict;
4			use Mysql;
5			use File::Copy;
6			use Getopt::Long;
7	ahart	1.5	use POSIX;
8	ahart	1.18	use Term::ANSIColor;
9	ahart	1.2
10	ahart	1.1	sub printHelp;
11			sub outputPset;
12			sub outputRunList;
13			sub getLocation;
14			sub outputCondor;
15
16	ahart	1.3	our $db = Mysql->connect ("cmshead", "ntuple", "osuT3User");
17	ahart	1.1
18			my %opt;
19			Getopt::Long::Configure ("bundling");
20	ahart	1.16	GetOptions (\%opt, "label\|l=s", "maxEvents\|m=s", "help\|h");
21	ahart	1.1	my $argc = @ARGV;
22
23			printHelp () if $opt{"help"};
24	ahart	1.12	printHelp () if $argc != 3 && $argc != 4;
25			my $dataset;
26			my $config;
27			my $directory;
28			my $nJobs;
29			if ($argc == 3)
30	ahart	1.1	{
31	ahart	1.12	$dataset = "";
32			$config = $ARGV[0];
33			$directory = $ARGV[1];
34			$nJobs = $ARGV[2];
35			}
36			if ($argc == 4)
37			{
38			$dataset = $ARGV[0];
39			$config = $ARGV[1];
40			$directory = $ARGV[2];
41			$nJobs = $ARGV[3];
42			}
43			if (!(-e $config))
44			{
45			print "\"$config\" does not exist!\n";
46	ahart	1.1	exit;
47			}
48	ahart	1.12	if (-e $directory)
49	ahart	1.1	{
50	ahart	1.12	print "Directory \"$directory\" already exists!\n";
51	ahart	1.1	print "Please delete it or specify another working directory.\n";
52			exit;
53			}
54	ahart	1.12	mkdir $directory;
55			my $nFiles = outputRunList ($dataset, $directory);
56	ahart	1.13	my $realNJobs = $nJobs;
57			$realNJobs = ceil ($nFiles / ceil ($nFiles / $realNJobs)) if $nFiles;
58			my $eventsPerJob = -1;
59			$eventsPerJob = ceil ($opt{"maxEvents"} / $realNJobs) if $opt{"maxEvents"} && $opt{"maxEvents"} >= 0;
60			my $realMaxEvents = $eventsPerJob * $realNJobs;
61			outputPset ($directory, $dataset, $opt{"maxEvents"}, $eventsPerJob);
62	ahart	1.16	outputCondor ("$ENV{'CMSSW_BASE'}/src/OSUT3Analysis/DBTools/data/condor.sub", $directory, $realNJobs, $dataset, $opt{"label"});
63	ahart	1.12	copy ($config, "$directory/userConfig_cfg.py");
64			chdir $directory;
65	ahart	1.13	print "Submitting $realNJobs jobs to run on $realMaxEvents events in $nFiles files.\n" if $realMaxEvents >= 0;
66			print "Submitting $realNJobs jobs to run on all events in $nFiles files.\n" if $realMaxEvents < 0;
67	ahart	1.19	system ("LD_LIBRARY_PATH=/usr/lib64/condor:\$LD_LIBRARY_PATH condor_submit condor.sub");
68	ahart	1.1
69			sub
70			outputPset
71			{
72			my $workingDir = shift;
73	ahart	1.12	my $dataset = shift;
74	ahart	1.13	my $nEvents = shift;
75			my $eventsPerJob = shift;
76	ahart	1.1
77			open (PSET, ">$workingDir/config_cfg.py");
78
79			print PSET "import FWCore.ParameterSet.Config as cms\n";
80			print PSET "import OSUT3Analysis.DBTools.osusub_cfg as osusub\n";
81			print PSET "import re\n";
82			print PSET "import userConfig_cfg as pset\n";
83			print PSET "\n";
84			print PSET "fileName = pset.process.TFileService.fileName\n";
85			print PSET "fileName = fileName.pythonValue ()\n";
86			print PSET "fileName = fileName[1:(len (fileName) - 1)]\n";
87			print PSET "fileName = re.sub (r'^(.)\\.([^\\.])\$', r'\\1_' + str (osusub.jobNumber) + r'.\\2', fileName)\n";
88			print PSET "pset.process.TFileService.fileName = fileName\n";
89			print PSET "\n";
90	ahart	1.12	if ($dataset)
91			{
92			print PSET "pset.process.source.fileNames = cms.untracked.vstring (osusub.runList)\n";
93	ahart	1.13	print PSET "pset.process.maxEvents.input = cms.untracked.int32 ($eventsPerJob)\n" if $eventsPerJob < 0;
94	ahart	1.12	}
95	ahart	1.13	print PSET "pset.process.maxEvents.input = cms.untracked.int32 ($eventsPerJob)\n" if $eventsPerJob >= 0;
96	ahart	1.1	print PSET "process = pset.process\n";
97
98			close (PSET);
99			}
100
101			sub
102			outputRunList
103			{
104			my $dataset = shift;
105			my $workingDir = shift;
106
107	ahart	1.12	return 0 if !$dataset;
108	ahart	1.11	my $location;
109			my $nFiles;
110			my $status;
111			my $crossSection;
112			my $isLocation = 0;
113			$isLocation = 1 if -e $dataset;
114			if (!$isLocation)
115			{
116			($location, $nFiles, $status, $crossSection) = getLocation ($dataset);
117			if ($status ne "present")
118			{
119			print "This dataset is not marked as present on the Tier 3!\n";
120			print "Continue anyway? (y/N): ";
121			my $response = <STDIN>;
122			$response =~ s/\n//g;
123			exit if !$response \|\| lc ($response) ne "y";
124			}
125			if (!(-e $location))
126			{
127			print "The database does not know where this dataset is!\n";
128			exit;
129			}
130	ahart	1.1	}
131	ahart	1.11	else
132	ahart	1.1	{
133	ahart	1.11	$location = $dataset;
134	ahart	1.1	}
135			opendir (LOCATION, $location);
136			my @files = readdir (LOCATION);
137			closedir (LOCATION);
138	ahart	1.11	if (!$isLocation && @files - 2 != $nFiles)
139	ahart	1.1	{
140			print "Number of files does not match database entry!\n";
141	ahart	1.9	print "Continue anyway? (y/N): ";
142			my $response = <STDIN>;
143			$response =~ s/\n//g;
144			exit if !$response \|\| lc ($response) ne "y";
145	ahart	1.1	}
146	ahart	1.11	elsif ($isLocation)
147			{
148			$nFiles = @files - 2;
149			}
150	ahart	1.1	open (RUNLIST, ">$workingDir/runList.py");
151			print RUNLIST "runList = [\n";
152			for (my $i = 0; $i < @files; $i++)
153			{
154			next if $files[$i] eq ".";
155			next if $files[$i] eq "..";
156			print RUNLIST "'file:$location/$files[$i]'";
157			print RUNLIST "," if $i + 1 != @files;
158			print RUNLIST "\n";
159			}
160			print RUNLIST "]";
161			close (RUNLIST);
162	ahart	1.6	if ($crossSection && $crossSection >= 0.0)
163			{
164			open (CROSS_SECTION, ">$workingDir/crossSectionInPicobarn.txt");
165			print CROSS_SECTION "$crossSection\n";
166			close (CROSS_SECTION);
167			}
168	ahart	1.5
169			return $nFiles;
170	ahart	1.1	}
171
172			sub
173			getLocation
174			{
175			my $dataset = shift;
176
177			my $results;
178	ahart	1.2	my $queryDataset = $dataset;
179			$queryDataset =~ s/\*/%/g;
180			$queryDataset =~ s/(.*)/%$1%/g;
181	ahart	1.15	my $query = "select dataset,user,creationTime,location,nFiles,status,crossSectionInPicobarn,higherOrderCrossSectionInPicobarn from ntuple where dataset like '$queryDataset' order by creationTime";
182	ahart	1.1	$db->selectdb ("ntuple");
183			$results = $db->query ($query);
184			if ($results->numrows () == 1)
185			{
186			my @row = $results->fetchrow ();
187	ahart	1.15	return ($row[3], $row[4], $row[5], $row[7]) if $row[7];
188			return ($row[3], $row[4], $row[5], $row[6]) if !$row[7];
189	ahart	1.1	}
190			if ($results->numrows () == 0)
191			{
192			print "Dataset does not exist on the Tier 3!\n";
193			exit;
194			}
195			my %map;
196			print "Found multiple datasets matching\n";
197			print "\"$dataset\":\n";
198			for (my $i = 1; $i <= $results->numrows (); $i++)
199			{
200			my @row = $results->fetchrow ();
201	ahart	1.18	$row[2] =~ s/([^ ]) [^ ]/$1/g;
202	ahart	1.15	$map{"$i"} = [$row[3], $row[4], $row[5], $row[7]] if $row[7];
203			$map{"$i"} = [$row[3], $row[4], $row[5], $row[6]] if !$row[7];
204	ahart	1.2	printf "(%2d) $row[0]\n", $i;
205	ahart	1.18	print " (";
206			print color "green" if $row[5] eq "present";
207			print color "bold yellow" if $row[5] eq "submitted";
208			print color "bold red" if $row[5] eq "created" or $row[5] eq "cancelled" or $row[5] eq "deprecated";
209			print $row[5];
210			print color "reset";
211			print ") created by $row[1] on $row[2]\n";
212	ahart	1.1	}
213			print "\nWhich dataset would you like to use?: ";
214			my $response = <STDIN>;
215			$response =~ s/[ \t\n]//g;
216			if (!(exists $map{$response}))
217			{
218			print "Your selection \"$response\" was not a valid option! Quitting.\n";
219			exit;
220			}
221
222	ahart	1.6	return ($map{$response}[0], $map{$response}[1], $map{$response}[2], $map{$response}[3]);
223	ahart	1.1	}
224
225			sub
226			outputCondor
227			{
228	ahart	1.14	my $condorFileName = shift;
229	ahart	1.1	my $workingDir = shift;
230			my $nJobs = shift;
231	ahart	1.8	my $dataset = shift;
232	ahart	1.16	my $label = shift;
233	ahart	1.1
234			my $cmsRun = `which cmsRun`;
235	ahart	1.14	my $condorFile = "";
236	ahart	1.1
237	ahart	1.14	if (!(-e $condorFileName))
238			{
239	ahart	1.16	my $arguments = "Arguments = config_cfg.py True $nJobs \$(Process)";
240	ahart	1.17	$arguments .= " $dataset" if $dataset;
241			$arguments .= " NULL" if !$dataset;
242			$arguments .= " $label" if $label;
243			$arguments .= " NULL" if !$label;
244			$arguments .= "\n";
245	ahart	1.16
246	ahart	1.14	$condorFile .= "Executable = $cmsRun\n";
247			$condorFile .= "Universe = vanilla\n";
248			$condorFile .= "Getenv = True\n";
249	ahart	1.16	$condorFile .= $arguments;
250	ahart	1.14	$condorFile .= "\n";
251			$condorFile .= "Output = condor_\$(Process).out\n";
252			$condorFile .= "Error = condor_\$(Process).err\n";
253			$condorFile .= "Log = condor_\$(Process).log\n";
254			$condorFile .= "\n";
255			$condorFile .= "+IsLocalJob = true\n";
256			$condorFile .= "Rank = TARGET.IsLocalSlot\n";
257			$condorFile .= "\n";
258			$condorFile .= "Queue $nJobs\n";
259			}
260			else
261			{
262			open (SUB, "<$condorFileName");
263			my @condorFile = <SUB>;
264			close (SUB);
265			$condorFile = join ("", @condorFile);
266			$condorFile =~ s/\$cmsRun/$cmsRun/g;
267			$condorFile =~ s/\$nJobs/$nJobs/g;
268			$condorFile =~ s/\$dataset/$dataset/g if $dataset;
269			$condorFile =~ s/\$dataset/NULL/g if !$dataset;
270	ahart	1.16	$condorFile =~ s/\$label/$label/g if $label;
271			$condorFile =~ s/\$label/NULL/g if !$label;
272	ahart	1.14	}
273	ahart	1.1
274	ahart	1.14	open (SUB, ">$workingDir/condor.sub");
275			print SUB $condorFile;
276	ahart	1.1	close (SUB);
277			}
278
279			sub
280			printHelp
281			{
282			my $exeName = $0;
283			$exeName =~ s/^.\/([^\/])$/$1/;
284
285	ahart	1.12	print "Usage: $exeName [OPTION]... [DATASET \| LOCATION] CONFIG DIRECTORY NJOBS\n";
286	ahart	1.1	print "Submits CMSSW jobs to the OSU Tier 3 compute nodes using Condor.\n";
287			print "\n";
288			printf "%-29s%s\n", " -h, --help", "print this help message";
289	ahart	1.16	printf "%-29s%s\n", " -l, --label LABEL", "give the dataset a short label";
290	ahart	1.13	printf "%-29s%s\n", " -m, --maxEvents N", "only run over N events in the dataset; default is";
291			printf "%-29s%s\n", " ", "to run over all events";
292	ahart	1.1	print "\n";
293	ahart	1.12	print "The optional first argument must be either a DATASET registered in the Tier 3\n";
294			print "ntuple database or a LOCATION which exists on disk. CONFIG must be a valid\n";
295			print "CMSSW python configuration which can be used with cmsRun. DIRECTORY is a\n";
296			print "working directory that is created and in which all output, both from the CMSSW\n";
297			print "jobs and from Condor, is placed. Finally, NJOBS is the number of Condor jobs\n";
298			print "that will be created.\n";
299	ahart	1.1
300			exit;
301			}