Processing/bin/downloadSample.sh

#!/bin/bash
#---------------------------------------------------------------------------------------------------
# Download a list of files
#---------------------------------------------------------------------------------------------------

# Read the arguments
echo ""
echo "downloadSample.sh  $*"
echo ""
dataDir=$1;      shift
book=$1;         shift
dataset=$1;      shift
target=$1;       shift
nCopyProcs=$1;   shift
condorOutput=$1; shift
onlyMissing=$1;  shift

DN=`grid-proxy-info -subject`

# Prepare environment
echo " "
echo " Process dataset: $dataset  of book: $book"
echo "   in directory : $dataDir"
echo "   to target    : $target"
echo "   n copy procs : $nCopyProcs"
echo "   condor output: $condorOutput"
echo "   only missing : $onlyMissing"

mkdir -p $condorOutput/$book/$dataset
makedir --exe  $target
makedir --exe  $target/$book
makedir --exe  $target/$book/$dataset
script=`which downloadFiles.sh`

# cleanup our lists and remake a clean one
#echo "rm -f $condorOutput/$book/$dataset/fileList*.txt*"
rm -f $condorOutput/$book/$dataset/fileList*.txt*

# make list of all local files
if [ "`echo $HOSTNAME | grep mit.edu`" != "" ] && \
 ( [ "`echo $dataDir | grep /castor/cern.ch`" != "" ] || \
   [ "`echo $target | grep /castor/cern.ch`" != "" ] )
then
  opt="--simple"
else
  opt=""
fi

#echo "list $opt $dataDir/$book/$dataset > $condorOutput/$book/$dataset/fileList-all.txt-bak"
list $opt $dataDir/$book/$dataset > $condorOutput/$book/$dataset/fileList-all.txt-bak

# Make sure there is a kerberos ticket available
id=`id -u`
cp /tmp/x509up_u${id} ~/.krb5/
KRB5CCNAME=`klist -5 | grep 'Ticket cache:' | cut -d' ' -f 3`
if ! [ -z $KRB5CCNAME ]
then
  mkdir    -p  ~/.krb5/
  chmod 0      ~/.krb5
  chmod u=rwx  ~/.krb5
  file=`echo $KRB5CCNAME | cut -d: -f2`
  if [ -f "$file" ]
  then
    cp $file ~/.krb5/ticket
  else
    echo " ERROR -- missing kerberos ticket ($KRB5CCNAME)."
    exit 1
  fi
else
  echo " ERROR -- missing kerberos ticket ($KRB5CCNAME)."
fi

# make list of all remote files
#echo " converting all entries"
cat $condorOutput/$book/$dataset/fileList-all.txt-bak | grep root | \
while read line
do
  size=`echo $line | tr -s ' ' | cut -d ' ' -f 1`
  file=`echo $line | tr -s ' ' | cut -d ' ' -f 2`
  file=`basename $file`
  echo "$size $file" >> $condorOutput/$book/$dataset/fileList-all.txt
done

# make list of all local files
if [ "`echo $HOSTNAME | grep mit.edu`" != "" ] && \
    ( [ "`echo $dataDir | grep /castor/cern.ch`" != "" ] || \
      [ "`echo $target | grep /castor/cern.ch`" != "" ] )
then
  opt="--simple"
else
  opt=""
fi

#echo "list $opt $target/$book/$dataset | grep root > $condorOutput/$book/$dataset/fileList-done.txt"
list $opt $target/$book/$dataset | grep root > $condorOutput/$book/$dataset/fileList-done.txt

# make list of missing files
rm -f $condorOutput/$book/$dataset/fileList.txt
touch $condorOutput/$book/$dataset/fileList.txt
cat   $condorOutput/$book/$dataset/fileList-all.txt | grep root | \
while read line
do
  size=`echo $line | tr -s ' ' | cut -d ' ' -f 1`
  file=`echo $line | tr -s ' ' | cut -d ' ' -f 2`
  exists=`grep "$file" $condorOutput/$book/$dataset/fileList-done.txt`
  if [ "$exists" == "" ]
  then
    echo "   -missing-- $file with $size bytes"
    echo "$size $file" >> $condorOutput/$book/$dataset/fileList.txt
  # else
  #   echo "   -exists--- $file with $size bytes - exists"
  else
    # now check that size matches
    test=`grep "$size $file" $condorOutput/$book/$dataset/fileList-done.txt`
    if [ "$test" == "" ]
    then
      if [ "$onlyMissing" == "" ]
      then
        echo "   -fileSize- $exists (remote: $size)"
        echo "$size $file" >> $condorOutput/$book/$dataset/fileList.txt
      fi
    fi
  fi
done
nAll=`wc -l $condorOutput/$book/$dataset/fileList-all.txt | cut -d ' ' -f1`
nMissing=`wc -l $condorOutput/$book/$dataset/fileList.txt | cut -d ' ' -f1`
nDone=`wc -l $condorOutput/$book/$dataset/fileList-done.txt | cut -d ' ' -f1`
echo ""
echo " Download Summary "
echo "   All       $nAll"
echo "   Done      $nDone"
echo "   Missing   $nMissing"
echo ""

# construct our job
nFiles=`wc -l $condorOutput/$book/$dataset/fileList.txt | cut -d ' ' -f1`
if   [ "$nFiles" == "" ] || [ "$nFiles" == "0" ]
then
  echo " "
  echo " No more files to download. EXIT."
  exit 0
elif [ $nFiles -lt $nCopyProcs ]
then
  nCopyProcs=$nFiles
fi
# how many files per job?
nFilesPerJob=$(( $nFiles/$nCopyProcs ))
nFilesTmp=$(( $nFilesPerJob*$nCopyProcs ))
if [ $nFilesPerJob == 1 ] && [ $nFiles -gt $nCopyProcs ]
then
  nFilesPerJob=2
elif [ $nFilesTmp -lt $nFiles ]
then
  nFilesPerJob=$(( $nFilesPerJob+1 ))
fi

echo "   n files to copy: $nFiles"
echo "   n files/proc   : $nFilesPerJob"

i=1
next=1
last=$nFilesPerJob

# make sure condor is properly setup for us
if ! [ -z $CONDOR_LOCATION ]
then
  unset  CONDOR_LOCATION
  export CONDOR_CONFIG=/usr/local/condor/etc/condor_config
fi

# loop over the condor jobs and submit them
while [ $i -le $nCopyProcs ] && [ $last -le $nFiles ]
do
  if [ $i == $nCopyProcs ]
  then
    last=$nFiles
  fi

  # say what we are going to submit
  echo "   downloadFiles.sh $dataDir $book $dataset $target $condorOutput $next $last"

  logFile=`echo download:$book/$dataset/${next}-${last}.txt | tr '/' '+'`
  logFile=/tmp/$logFile
  rm -f $logFile

  # prepare the condor_submit files
  cat > submit_$$.cmd <<EOF
Universe                = vanilla
Requirements            = ( (Arch == "INTEL") && (Disk >= DiskUsage) && ((Memory * 1024) >= ImageSize) && (HasFileTransfer) )
Notify_user             = paus@mit.edu
Notification            = Error
Executable              = $script
Arguments               = $dataDir $book $dataset $target $condorOutput $next $last
Rank                    = Mips
GetEnv                  = True
Input                   = /dev/null
Output                  = $condorOutput/$book/$dataset/${next}-${last}.out
Error                   = $condorOutput/$book/$dataset/${next}-${last}.err
Log                     = $logFile
should_transfer_files   = YES
when_to_transfer_output = ON_EXIT

+AccountingGroup        = "group_cmsuser.cmsu0284"

Queue
EOF

#+x509userproxysubject   = $DN

  # submit the jobs
  condor_submit submit_$$.cmd >& /dev/null #>& lastSub
  #cat submit_$$.cmd
  rm  submit_$$.cmd

  # update counters
  next=$(( $next + $nFilesPerJob ))
  last=$(( $last + $nFilesPerJob ))
  i=$(( $i + 1 ))
done

exit 0
Revision:	1.10
Committed:	Sat Oct 22 13:07:04 2011 UTC (13 years, 6 months ago) by paus
Content type:	application/x-sh
Branch:	MAIN
CVS Tags:	Mit_025c_branch1, Mit_025c_branch0, Mit_025c, Mit_025b, Mit_025a
Branch point for:	Mit_025c_branch
Changes since 1.9:	+11 -2 lines
Log Message:	Private file productions.
#	User	Rev	Content
1	paus	1.1	#!/bin/bash
2			#---------------------------------------------------------------------------------------------------
3			# Download a list of files
4			#---------------------------------------------------------------------------------------------------
5
6			# Read the arguments
7			echo ""
8			echo "downloadSample.sh $*"
9			echo ""
10			dataDir=$1; shift
11			book=$1; shift
12			dataset=$1; shift
13			target=$1; shift
14			nCopyProcs=$1; shift
15			condorOutput=$1; shift
16	paus	1.3	onlyMissing=$1; shift
17
18			DN=`grid-proxy-info -subject`
19	paus	1.1
20			# Prepare environment
21			echo " "
22			echo " Process dataset: $dataset of book: $book"
23			echo " in directory : $dataDir"
24			echo " to target : $target"
25			echo " n copy procs : $nCopyProcs"
26			echo " condor output: $condorOutput"
27	paus	1.3	echo " only missing : $onlyMissing"
28	paus	1.1
29	paus	1.8	mkdir -p $condorOutput/$book/$dataset
30	paus	1.9	makedir --exe $target
31			makedir --exe $target/$book
32	paus	1.8	makedir --exe $target/$book/$dataset
33	paus	1.1	script=`which downloadFiles.sh`
34
35	paus	1.3	# cleanup our lists and remake a clean one
36	paus	1.4	#echo "rm -f $condorOutput/$book/$dataset/fileList.txt"
37	paus	1.1	rm -f $condorOutput/$book/$dataset/fileList.txt
38
39	paus	1.3	# make list of all local files
40	paus	1.4	if [ "`echo $HOSTNAME \| grep mit.edu`" != "" ] && \
41			( [ "`echo $dataDir \| grep /castor/cern.ch`" != "" ] \|\| \
42			[ "`echo $target \| grep /castor/cern.ch`" != "" ] )
43	paus	1.1	then
44	paus	1.3	opt="--simple"
45	paus	1.1	else
46	paus	1.3	opt=""
47	paus	1.1	fi
48
49	paus	1.3	#echo "list $opt $dataDir/$book/$dataset > $condorOutput/$book/$dataset/fileList-all.txt-bak"
50			list $opt $dataDir/$book/$dataset > $condorOutput/$book/$dataset/fileList-all.txt-bak
51
52	paus	1.1	# Make sure there is a kerberos ticket available
53	paus	1.3	id=`id -u`
54			cp /tmp/x509up_u${id} ~/.krb5/
55	paus	1.1	KRB5CCNAME=`klist -5 \| grep 'Ticket cache:' \| cut -d' ' -f 3`
56			if ! [ -z $KRB5CCNAME ]
57			then
58			mkdir -p ~/.krb5/
59			chmod 0 ~/.krb5
60			chmod u=rwx ~/.krb5
61			file=`echo $KRB5CCNAME \| cut -d: -f2`
62			if [ -f "$file" ]
63			then
64			cp $file ~/.krb5/ticket
65			else
66			echo " ERROR -- missing kerberos ticket ($KRB5CCNAME)."
67			exit 1
68			fi
69			else
70			echo " ERROR -- missing kerberos ticket ($KRB5CCNAME)."
71			fi
72
73			# make list of all remote files
74			#echo " converting all entries"
75			cat $condorOutput/$book/$dataset/fileList-all.txt-bak \| grep root \| \
76			while read line
77			do
78			size=`echo $line \| tr -s ' ' \| cut -d ' ' -f 1`
79			file=`echo $line \| tr -s ' ' \| cut -d ' ' -f 2`
80			file=`basename $file`
81			echo "$size $file" >> $condorOutput/$book/$dataset/fileList-all.txt
82			done
83
84			# make list of all local files
85	paus	1.9	if [ "`echo $HOSTNAME \| grep mit.edu`" != "" ] && \
86			( [ "`echo $dataDir \| grep /castor/cern.ch`" != "" ] \|\| \
87			[ "`echo $target \| grep /castor/cern.ch`" != "" ] )
88	paus	1.1	then
89	paus	1.3	opt="--simple"
90			else
91			opt=""
92	paus	1.1	fi
93	paus	1.3
94	paus	1.1	#echo "list $opt $target/$book/$dataset \| grep root > $condorOutput/$book/$dataset/fileList-done.txt"
95			list $opt $target/$book/$dataset \| grep root > $condorOutput/$book/$dataset/fileList-done.txt
96
97			# make list of missing files
98			rm -f $condorOutput/$book/$dataset/fileList.txt
99			touch $condorOutput/$book/$dataset/fileList.txt
100			cat $condorOutput/$book/$dataset/fileList-all.txt \| grep root \| \
101			while read line
102			do
103			size=`echo $line \| tr -s ' ' \| cut -d ' ' -f 1`
104			file=`echo $line \| tr -s ' ' \| cut -d ' ' -f 2`
105	paus	1.3	exists=`grep "$file" $condorOutput/$book/$dataset/fileList-done.txt`
106	paus	1.1	if [ "$exists" == "" ]
107			then
108	paus	1.3	echo " -missing-- $file with $size bytes"
109	paus	1.1	echo "$size $file" >> $condorOutput/$book/$dataset/fileList.txt
110			# else
111	paus	1.3	# echo " -exists--- $file with $size bytes - exists"
112			else
113			# now check that size matches
114			test=`grep "$size $file" $condorOutput/$book/$dataset/fileList-done.txt`
115			if [ "$test" == "" ]
116			then
117			if [ "$onlyMissing" == "" ]
118			then
119			echo " -fileSize- $exists (remote: $size)"
120			echo "$size $file" >> $condorOutput/$book/$dataset/fileList.txt
121			fi
122			fi
123	paus	1.1	fi
124			done
125	paus	1.10	nAll=`wc -l $condorOutput/$book/$dataset/fileList-all.txt \| cut -d ' ' -f1`
126			nMissing=`wc -l $condorOutput/$book/$dataset/fileList.txt \| cut -d ' ' -f1`
127			nDone=`wc -l $condorOutput/$book/$dataset/fileList-done.txt \| cut -d ' ' -f1`
128			echo ""
129			echo " Download Summary "
130			echo " All $nAll"
131			echo " Done $nDone"
132			echo " Missing $nMissing"
133			echo ""
134	paus	1.1
135			# construct our job
136			nFiles=`wc -l $condorOutput/$book/$dataset/fileList.txt \| cut -d ' ' -f1`
137			if [ "$nFiles" == "" ] \|\| [ "$nFiles" == "0" ]
138			then
139			echo " "
140			echo " No more files to download. EXIT."
141			exit 0
142			elif [ $nFiles -lt $nCopyProcs ]
143			then
144			nCopyProcs=$nFiles
145			fi
146	paus	1.4	# how many files per job?
147	paus	1.1	nFilesPerJob=$(( $nFiles/$nCopyProcs ))
148	paus	1.4	nFilesTmp=$(( $nFilesPerJob*$nCopyProcs ))
149			if [ $nFilesPerJob == 1 ] && [ $nFiles -gt $nCopyProcs ]
150			then
151			nFilesPerJob=2
152			elif [ $nFilesTmp -lt $nFiles ]
153			then
154			nFilesPerJob=$(( $nFilesPerJob+1 ))
155			fi
156
157	paus	1.10	echo " n files to copy: $nFiles"
158			echo " n files/proc : $nFilesPerJob"
159	paus	1.1
160			i=1
161			next=1
162			last=$nFilesPerJob
163
164	paus	1.3	# make sure condor is properly setup for us
165			if ! [ -z $CONDOR_LOCATION ]
166			then
167			unset CONDOR_LOCATION
168			export CONDOR_CONFIG=/usr/local/condor/etc/condor_config
169			fi
170
171			# loop over the condor jobs and submit them
172	paus	1.4	while [ $i -le $nCopyProcs ] && [ $last -le $nFiles ]
173	paus	1.1	do
174			if [ $i == $nCopyProcs ]
175			then
176			last=$nFiles
177			fi
178
179			# say what we are going to submit
180			echo " downloadFiles.sh $dataDir $book $dataset $target $condorOutput $next $last"
181
182			logFile=`echo download:$book/$dataset/${next}-${last}.txt \| tr '/' '+'`
183			logFile=/tmp/$logFile
184			rm -f $logFile
185
186			# prepare the condor_submit files
187			cat > submit_$$.cmd <<EOF
188			Universe = vanilla
189	paus	1.7	Requirements = ( (Arch == "INTEL") && (Disk >= DiskUsage) && ((Memory * 1024) >= ImageSize) && (HasFileTransfer) )
190	paus	1.1	Notify_user = paus@mit.edu
191			Notification = Error
192			Executable = $script
193			Arguments = $dataDir $book $dataset $target $condorOutput $next $last
194			Rank = Mips
195			GetEnv = True
196			Input = /dev/null
197			Output = $condorOutput/$book/$dataset/${next}-${last}.out
198			Error = $condorOutput/$book/$dataset/${next}-${last}.err
199			Log = $logFile
200			should_transfer_files = YES
201			when_to_transfer_output = ON_EXIT
202	paus	1.3
203			+AccountingGroup = "group_cmsuser.cmsu0284"
204
205	paus	1.1	Queue
206			EOF
207
208	paus	1.3	#+x509userproxysubject = $DN
209
210	paus	1.1	# submit the jobs
211			condor_submit submit_$$.cmd >& /dev/null #>& lastSub
212			#cat submit_$$.cmd
213			rm submit_$$.cmd
214
215			# update counters
216			next=$(( $next + $nFilesPerJob ))
217			last=$(( $last + $nFilesPerJob ))
218			i=$(( $i + 1 ))
219			done
220
221			exit 0