[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;

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, "help|h");
my $argc = @ARGV;

printHelp () if $opt{"help"};
printHelp () if $argc != 4;
if (!(-e $ARGV[1]))
  {
    print "\"$ARGV[1]\" does not exist!\n";
    exit;
  }
if (-e $ARGV[2])
  {
    print "Directory \"$ARGV[2]\" already exists!\n";
    print "Please delete it or specify another working directory.\n";
    exit;
  }
mkdir $ARGV[2];
my $nFiles = outputRunList ($ARGV[0], $ARGV[2]);
my $nJobs = $ARGV[3];
my $realNJobs = ceil ($nFiles / ceil ($nFiles / $nJobs));
outputPset ($ARGV[2]);
outputCondor ($ARGV[2], $realNJobs);
copy ($ARGV[1], "$ARGV[2]/userConfig_cfg.py");
chdir $ARGV[2];
print "Submitting $realNJobs jobs to run on $nFiles files.\n";
system ("condor_submit condor.sub");

sub
outputPset
{
  my $workingDir = 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";
  print PSET "pset.process.source.fileNames = cms.untracked.vstring (osusub.runList)\n";
  print PSET "pset.process.maxEvents.input = cms.untracked.int32 (-1)\n";
  print PSET "process = pset.process\n";

  close (PSET);
}

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

  my ($location, $nFiles, $status) = getLocation ($dataset);
  if ($status ne "present")
    {
      print "This dataset is not marked as present on the Tier 3!\n";
      exit;
    }
  if (!(-e $location))
    {
      print "The database does not know where this dataset is!\n";
      exit;
    }
  opendir (LOCATION, $location);
  my @files = readdir (LOCATION);
  closedir (LOCATION);
  if (@files - 2 != $nFiles)
    {
      print "Number of files does not match database entry!\n";
      exit;
    }
  open (RUNLIST, ">$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);

  return $nFiles;
}

sub
getLocation
{
  my $dataset = shift;

  my $results;
  my $queryDataset = $dataset;
  $queryDataset =~ s/\*/%/g;
  $queryDataset =~ s/(.*)/%$1%/g;
  my $query = "select dataset,user,creationTime,location,nFiles,status from ntuple where dataset like '$queryDataset' and status='present' order by creationTime";
  $db->selectdb ("ntuple");
  $results = $db->query ($query);
  if ($results->numrows () == 1)
    {
      my @row = $results->fetchrow ();
      return ($row[3], $row[4], $row[5]);
    }
  if ($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 ();
      $map{"$i"} = [$row[3], $row[4], $row[5]];
      printf "(%2d) $row[0]\n", $i;
      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]);
}

sub
outputCondor
{
  my $workingDir = shift;
  my $nJobs = shift;

  my $cmsRun = `which cmsRun`;
  open (SUB, ">$workingDir/condor.sub");

  print SUB "Executable              = $cmsRun\n";
  print SUB "Universe                = vanilla\n";
  print SUB "Getenv                  = True\n";
  print SUB "Arguments               = config_cfg.py $nJobs \$(Process)\n";
  print SUB "\n";
  print SUB "Output                  = condor_\$(Process).out\n";
  print SUB "Error                   = condor_\$(Process).err\n";
  print SUB "Log                     = condor_\$(Process).log\n";
  print SUB "\n";
  print SUB "Queue $nJobs\n";

  close (SUB);
}

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

  print "Usage: $exeName [OPTION]... DATASET 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";
  print "\n";
  print "The DATASET must exist in the Tier 3 ntuple database, and CONFIG must be a valid\n";
  print "CMSSW python configuration which can be used with cmsRun. DIRECTORY is a working\n";
  print "directory that is created and in which all output, both from the CMSSW jobs and\n";
  print "from Condor, is placed. Finally, NJOBS is the number of Condor jobs that will\n";
  print "be created.\n";

  exit;
}
Revision:	1.5
Committed:	Wed Aug 15 08:43:26 2012 UTC (12 years, 8 months ago) by ahart
Branch:	MAIN
CVS Tags:	V00-00-04
Changes since 1.4:	+8 -2 lines
Log Message:	Do not submit empty jobs.
#	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	1.2
9	ahart	1.1	sub printHelp;
10			sub outputPset;
11			sub outputRunList;
12			sub getLocation;
13			sub outputCondor;
14
15	ahart	1.3	our $db = Mysql->connect ("cmshead", "ntuple", "osuT3User");
16	ahart	1.1
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	ahart	1.5	my $nFiles = outputRunList ($ARGV[0], $ARGV[2]);
37			my $nJobs = $ARGV[3];
38			my $realNJobs = ceil ($nFiles / ceil ($nFiles / $nJobs));
39	ahart	1.1	outputPset ($ARGV[2]);
40	ahart	1.5	outputCondor ($ARGV[2], $realNJobs);
41	ahart	1.1	copy ($ARGV[1], "$ARGV[2]/userConfig_cfg.py");
42			chdir $ARGV[2];
43	ahart	1.5	print "Submitting $realNJobs jobs to run on $nFiles files.\n";
44	ahart	1.1	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, $status) = getLocation ($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	ahart	1.5
109			return $nFiles;
110	ahart	1.1	}
111
112			sub
113			getLocation
114			{
115			my $dataset = shift;
116
117			my $results;
118	ahart	1.2	my $queryDataset = $dataset;
119			$queryDataset =~ s/\*/%/g;
120			$queryDataset =~ s/(.*)/%$1%/g;
121	ahart	1.4	my $query = "select dataset,user,creationTime,location,nFiles,status from ntuple where dataset like '$queryDataset' and status='present' order by creationTime";
122	ahart	1.1	$db->selectdb ("ntuple");
123			$results = $db->query ($query);
124			if ($results->numrows () == 1)
125			{
126			my @row = $results->fetchrow ();
127	ahart	1.2	return ($row[3], $row[4], $row[5]);
128	ahart	1.1	}
129			if ($results->numrows () == 0)
130			{
131			print "Dataset does not exist on the Tier 3!\n";
132			exit;
133			}
134			my %map;
135			print "Found multiple datasets matching\n";
136			print "\"$dataset\":\n";
137			for (my $i = 1; $i <= $results->numrows (); $i++)
138			{
139			my @row = $results->fetchrow ();
140	ahart	1.2	$map{"$i"} = [$row[3], $row[4], $row[5]];
141			printf "(%2d) $row[0]\n", $i;
142			print " created by $row[1] on $row[2]\n";
143	ahart	1.1	}
144			print "\nWhich dataset would you like to use?: ";
145			my $response = <STDIN>;
146			$response =~ s/[ \t\n]//g;
147			if (!(exists $map{$response}))
148			{
149			print "Your selection \"$response\" was not a valid option! Quitting.\n";
150			exit;
151			}
152
153			return ($map{$response}[0], $map{$response}[1], $map{$response}[2]);
154			}
155
156			sub
157			outputCondor
158			{
159			my $workingDir = shift;
160			my $nJobs = shift;
161
162			my $cmsRun = `which cmsRun`;
163			open (SUB, ">$workingDir/condor.sub");
164
165			print SUB "Executable = $cmsRun\n";
166			print SUB "Universe = vanilla\n";
167			print SUB "Getenv = True\n";
168			print SUB "Arguments = config_cfg.py $nJobs \$(Process)\n";
169			print SUB "\n";
170			print SUB "Output = condor_\$(Process).out\n";
171			print SUB "Error = condor_\$(Process).err\n";
172			print SUB "Log = condor_\$(Process).log\n";
173			print SUB "\n";
174			print SUB "Queue $nJobs\n";
175
176			close (SUB);
177			}
178
179			sub
180			printHelp
181			{
182			my $exeName = $0;
183			$exeName =~ s/^.\/([^\/])$/$1/;
184
185	ahart	1.3	print "Usage: $exeName [OPTION]... DATASET CONFIG DIRECTORY NJOBS\n";
186	ahart	1.1	print "Submits CMSSW jobs to the OSU Tier 3 compute nodes using Condor.\n";
187			print "\n";
188			printf "%-29s%s\n", " -h, --help", "print this help message";
189			print "\n";
190			print "The DATASET must exist in the Tier 3 ntuple database, and CONFIG must be a valid\n";
191			print "CMSSW python configuration which can be used with cmsRun. DIRECTORY is a working\n";
192			print "directory that is created and in which all output, both from the CMSSW jobs and\n";
193			print "from Condor, is placed. Finally, NJOBS is the number of Condor jobs that will\n";
194			print "be created.\n";
195
196			exit;
197			}