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
mkdir -p       $target/$book/$dataset
chmod a+w      $target/$book/$dataset
script=`which downloadFiles.sh`

# cleanup our lists and remake a clean one
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

# 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
nFilesPerJob=$(( $nFiles/$nCopyProcs ))
echo "   n files all  : $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 ]
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 == "X86_64" || Arch == "INTEL") && (OpSys == "LINUX") && (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:	1.3
Committed:	Thu Jan 6 22:30:07 2011 UTC (14 years, 3 months ago) by paus
Content type:	application/x-sh
Branch:	MAIN
CVS Tags:	Mit_020pre1, Mit_018
Changes since 1.2:	+48 -23 lines
Log Message:	Adding more powerful copy.
#	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			mkdir -p $condorOutput/$book/$dataset
30	paus	1.3	mkdir -p $target/$book/$dataset
31			chmod a+w $target/$book/$dataset
32	paus	1.1	script=`which downloadFiles.sh`
33
34	paus	1.3	# cleanup our lists and remake a clean one
35	paus	1.1	rm -f $condorOutput/$book/$dataset/fileList.txt
36
37	paus	1.3	# make list of all local files
38			if [ "`echo $HOSTNAME \| grep mit.edu`" != "" ] && ( [ "`echo $dataDir \| grep /castor/cern.ch`" != "" ] \|\| [ "`echo $target \| grep /castor/cern.ch`" != "" ] )
39	paus	1.1	then
40	paus	1.3	opt="--simple"
41	paus	1.1	else
42	paus	1.3	opt=""
43	paus	1.1	fi
44
45	paus	1.3	#echo "list $opt $dataDir/$book/$dataset > $condorOutput/$book/$dataset/fileList-all.txt-bak"
46			list $opt $dataDir/$book/$dataset > $condorOutput/$book/$dataset/fileList-all.txt-bak
47
48	paus	1.1	# Make sure there is a kerberos ticket available
49	paus	1.3	id=`id -u`
50			cp /tmp/x509up_u${id} ~/.krb5/
51	paus	1.1	KRB5CCNAME=`klist -5 \| grep 'Ticket cache:' \| cut -d' ' -f 3`
52			if ! [ -z $KRB5CCNAME ]
53			then
54			mkdir -p ~/.krb5/
55			chmod 0 ~/.krb5
56			chmod u=rwx ~/.krb5
57			file=`echo $KRB5CCNAME \| cut -d: -f2`
58			if [ -f "$file" ]
59			then
60			cp $file ~/.krb5/ticket
61			else
62			echo " ERROR -- missing kerberos ticket ($KRB5CCNAME)."
63			exit 1
64			fi
65			else
66			echo " ERROR -- missing kerberos ticket ($KRB5CCNAME)."
67			fi
68
69			# make list of all remote files
70			#echo " converting all entries"
71			cat $condorOutput/$book/$dataset/fileList-all.txt-bak \| grep root \| \
72			while read line
73			do
74			size=`echo $line \| tr -s ' ' \| cut -d ' ' -f 1`
75			file=`echo $line \| tr -s ' ' \| cut -d ' ' -f 2`
76			file=`basename $file`
77			echo "$size $file" >> $condorOutput/$book/$dataset/fileList-all.txt
78			done
79
80			# make list of all local files
81	paus	1.3	if [ "`echo $HOSTNAME \| grep mit.edu`" != "" ] && ( [ "`echo $dataDir \| grep /castor/cern.ch`" != "" ] \|\| [ "`echo $target \| grep /castor/cern.ch`" != "" ] )
82	paus	1.1	then
83	paus	1.3	opt="--simple"
84			else
85			opt=""
86	paus	1.1	fi
87	paus	1.3
88	paus	1.1	#echo "list $opt $target/$book/$dataset \| grep root > $condorOutput/$book/$dataset/fileList-done.txt"
89			list $opt $target/$book/$dataset \| grep root > $condorOutput/$book/$dataset/fileList-done.txt
90
91			# make list of missing files
92			rm -f $condorOutput/$book/$dataset/fileList.txt
93			touch $condorOutput/$book/$dataset/fileList.txt
94			cat $condorOutput/$book/$dataset/fileList-all.txt \| grep root \| \
95			while read line
96			do
97			size=`echo $line \| tr -s ' ' \| cut -d ' ' -f 1`
98			file=`echo $line \| tr -s ' ' \| cut -d ' ' -f 2`
99	paus	1.3	exists=`grep "$file" $condorOutput/$book/$dataset/fileList-done.txt`
100	paus	1.1	if [ "$exists" == "" ]
101			then
102	paus	1.3	echo " -missing-- $file with $size bytes"
103	paus	1.1	echo "$size $file" >> $condorOutput/$book/$dataset/fileList.txt
104			# else
105	paus	1.3	# echo " -exists--- $file with $size bytes - exists"
106			else
107			# now check that size matches
108			test=`grep "$size $file" $condorOutput/$book/$dataset/fileList-done.txt`
109			if [ "$test" == "" ]
110			then
111			if [ "$onlyMissing" == "" ]
112			then
113			echo " -fileSize- $exists (remote: $size)"
114			echo "$size $file" >> $condorOutput/$book/$dataset/fileList.txt
115			fi
116			fi
117	paus	1.1	fi
118			done
119
120			# construct our job
121			nFiles=`wc -l $condorOutput/$book/$dataset/fileList.txt \| cut -d ' ' -f1`
122			if [ "$nFiles" == "" ] \|\| [ "$nFiles" == "0" ]
123			then
124			echo " "
125			echo " No more files to download. EXIT."
126			exit 0
127			elif [ $nFiles -lt $nCopyProcs ]
128			then
129			nCopyProcs=$nFiles
130			fi
131			nFilesPerJob=$(( $nFiles/$nCopyProcs ))
132			echo " n files all : $nFiles"
133			echo " n files/proc : $nFilesPerJob"
134
135			i=1
136			next=1
137			last=$nFilesPerJob
138
139	paus	1.3	# make sure condor is properly setup for us
140			if ! [ -z $CONDOR_LOCATION ]
141			then
142			unset CONDOR_LOCATION
143			export CONDOR_CONFIG=/usr/local/condor/etc/condor_config
144			fi
145
146			# loop over the condor jobs and submit them
147	paus	1.1	while [ $i -le $nCopyProcs ]
148			do
149			if [ $i == $nCopyProcs ]
150			then
151			last=$nFiles
152			fi
153
154			# say what we are going to submit
155			echo " downloadFiles.sh $dataDir $book $dataset $target $condorOutput $next $last"
156
157			logFile=`echo download:$book/$dataset/${next}-${last}.txt \| tr '/' '+'`
158			logFile=/tmp/$logFile
159			rm -f $logFile
160
161			# prepare the condor_submit files
162			cat > submit_$$.cmd <<EOF
163			Universe = vanilla
164			Requirements = ( (Arch == "X86_64" \|\| Arch == "INTEL") && (OpSys == "LINUX") && (Disk >= DiskUsage) && ((Memory * 1024) >= ImageSize) && (HasFileTransfer) )
165			Notify_user = paus@mit.edu
166			Notification = Error
167			Executable = $script
168			Arguments = $dataDir $book $dataset $target $condorOutput $next $last
169			Rank = Mips
170			GetEnv = True
171			Input = /dev/null
172			Output = $condorOutput/$book/$dataset/${next}-${last}.out
173			Error = $condorOutput/$book/$dataset/${next}-${last}.err
174			Log = $logFile
175			should_transfer_files = YES
176			when_to_transfer_output = ON_EXIT
177	paus	1.3
178			+AccountingGroup = "group_cmsuser.cmsu0284"
179
180	paus	1.1	Queue
181			EOF
182
183	paus	1.3	#+x509userproxysubject = $DN
184
185	paus	1.1	# submit the jobs
186			condor_submit submit_$$.cmd >& /dev/null #>& lastSub
187			#cat submit_$$.cmd
188			rm submit_$$.cmd
189
190			# update counters
191			next=$(( $next + $nFilesPerJob ))
192			last=$(( $last + $nFilesPerJob ))
193			i=$(( $i + 1 ))
194			done
195
196			exit 0