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

# 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"

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

# cleanup our lists
rm -f $condorOutput/$book/$dataset/fileList*.txt*

var=`echo $dataDir | grep /castor/cern.ch`
if [ "$var" != "" ]
then
  storageEle="srm-cms.cern.ch"
  storagePath='/srm/managerv2?SFN='
  storageUrl="srm://${storageEle}:8443${storagePath}$dataDir/$book/$dataset"
  #srmls $storageUrl | grep root | tr -s ' ' | cut -d' ' -f 2-3 2> /dev/null 1> \
  #  $condorOutput/$book/$dataset/fileList-all.txt-bak
  list $dataDir/$book/$dataset > $condorOutput/$book/$dataset/fileList-all.txt-bak
else
  storageEle="se01.cmsaf.mit.edu"
  storagePath='/srm/managerv2?SFN='
  storageUrl="srm://${storageEle}:8443${storagePath}$dataDir/$book/$dataset"
  #echo "list $dataDir/$book/$dataset > $condorOutput/$book/$dataset/fileList-all.txt-bak"
  list $dataDir/$book/$dataset > $condorOutput/$book/$dataset/fileList-all.txt-bak
fi

# Make sure there is a kerberos ticket available
cp /tmp/x509up_u5410 ~/.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
opt=''
if [ "`echo $HOSTNAME | grep mit.edu`" != "" ] && [ "`echo $dataDir | grep /castor/cern.ch`" != "" ]
then
  opt='--simple'
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 $line | tr -s ' ' | cut -d ' ' -f 2`
  exists=`grep "$size $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"
  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

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
Queue
EOF

  # 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:	1.1
Committed:	Sun Dec 5 01:01:21 2010 UTC (14 years, 5 months ago) by paus
Content type:	application/x-sh
Branch:	MAIN
CVS Tags:	Mit_017
Log Message:	Next iteration with improved downloading tool.
#	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
17			# Prepare environment
18			echo " "
19			echo " Process dataset: $dataset of book: $book"
20			echo " in directory : $dataDir"
21			echo " to target : $target"
22			echo " n copy procs : $nCopyProcs"
23			echo " condor output: $condorOutput"
24
25			mkdir -p $condorOutput/$book/$dataset
26			mkdir -p $target/$book/$dataset
27			script=`which downloadFiles.sh`
28
29			# cleanup our lists
30			rm -f $condorOutput/$book/$dataset/fileList.txt
31
32			var=`echo $dataDir \| grep /castor/cern.ch`
33			if [ "$var" != "" ]
34			then
35			storageEle="srm-cms.cern.ch"
36			storagePath='/srm/managerv2?SFN='
37			storageUrl="srm://${storageEle}:8443${storagePath}$dataDir/$book/$dataset"
38			#srmls $storageUrl \| grep root \| tr -s ' ' \| cut -d' ' -f 2-3 2> /dev/null 1> \
39			# $condorOutput/$book/$dataset/fileList-all.txt-bak
40			list $dataDir/$book/$dataset > $condorOutput/$book/$dataset/fileList-all.txt-bak
41			else
42			storageEle="se01.cmsaf.mit.edu"
43			storagePath='/srm/managerv2?SFN='
44			storageUrl="srm://${storageEle}:8443${storagePath}$dataDir/$book/$dataset"
45			#echo "list $dataDir/$book/$dataset > $condorOutput/$book/$dataset/fileList-all.txt-bak"
46			list $dataDir/$book/$dataset > $condorOutput/$book/$dataset/fileList-all.txt-bak
47			fi
48
49			# Make sure there is a kerberos ticket available
50			cp /tmp/x509up_u5410 ~/.krb5/
51			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			opt=''
82			if [ "`echo $HOSTNAME \| grep mit.edu`" != "" ] && [ "`echo $dataDir \| grep /castor/cern.ch`" != "" ]
83			then
84			opt='--simple'
85			fi
86			#echo "list $opt $target/$book/$dataset \| grep root > $condorOutput/$book/$dataset/fileList-done.txt"
87			list $opt $target/$book/$dataset \| grep root > $condorOutput/$book/$dataset/fileList-done.txt
88
89			# make list of missing files
90			rm -f $condorOutput/$book/$dataset/fileList.txt
91			touch $condorOutput/$book/$dataset/fileList.txt
92			cat $condorOutput/$book/$dataset/fileList-all.txt \| grep root \| \
93			while read line
94			do
95			size=`echo $line \| tr -s ' ' \| cut -d ' ' -f 1`
96			file=`echo $line \| tr -s ' ' \| cut -d ' ' -f 2`
97			exists=`grep "$size $file" $condorOutput/$book/$dataset/fileList-done.txt`
98			if [ "$exists" == "" ]
99			then
100			echo " -missing- $file with $size bytes"
101			echo "$size $file" >> $condorOutput/$book/$dataset/fileList.txt
102			# else
103			# echo " -exists-- $file with $size bytes - exists"
104			fi
105			done
106
107			# construct our job
108			nFiles=`wc -l $condorOutput/$book/$dataset/fileList.txt \| cut -d ' ' -f1`
109			if [ "$nFiles" == "" ] \|\| [ "$nFiles" == "0" ]
110			then
111			echo " "
112			echo " No more files to download. EXIT."
113			exit 0
114			elif [ $nFiles -lt $nCopyProcs ]
115			then
116			nCopyProcs=$nFiles
117			fi
118			nFilesPerJob=$(( $nFiles/$nCopyProcs ))
119			echo " n files all : $nFiles"
120			echo " n files/proc : $nFilesPerJob"
121
122			i=1
123			next=1
124			last=$nFilesPerJob
125
126			while [ $i -le $nCopyProcs ]
127			do
128			if [ $i == $nCopyProcs ]
129			then
130			last=$nFiles
131			fi
132
133			# say what we are going to submit
134			echo " downloadFiles.sh $dataDir $book $dataset $target $condorOutput $next $last"
135
136			logFile=`echo download:$book/$dataset/${next}-${last}.txt \| tr '/' '+'`
137			logFile=/tmp/$logFile
138			rm -f $logFile
139
140			# prepare the condor_submit files
141			cat > submit_$$.cmd <<EOF
142			Universe = vanilla
143			Requirements = ( (Arch == "X86_64" \|\| Arch == "INTEL") && (OpSys == "LINUX") && (Disk >= DiskUsage) && ((Memory * 1024) >= ImageSize) && (HasFileTransfer) )
144			Notify_user = paus@mit.edu
145			Notification = Error
146			Executable = $script
147			Arguments = $dataDir $book $dataset $target $condorOutput $next $last
148			Rank = Mips
149			GetEnv = True
150			Input = /dev/null
151			Output = $condorOutput/$book/$dataset/${next}-${last}.out
152			Error = $condorOutput/$book/$dataset/${next}-${last}.err
153			Log = $logFile
154			should_transfer_files = YES
155			when_to_transfer_output = ON_EXIT
156			Queue
157			EOF
158
159			# submit the jobs
160			condor_submit submit_$$.cmd >& /dev/null #>& lastSub
161			#cat submit_$$.cmd
162			rm submit_$$.cmd
163
164			# update counters
165			next=$(( $next + $nFilesPerJob ))
166			last=$(( $last + $nFilesPerJob ))
167			i=$(( $i + 1 ))
168			done
169
170			exit 0