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

list $opt $dataDir/$book/$dataset | sort > $condorOutput/$book/$dataset/fileList-all.$$.txt-bak

# Make sure there are kerberos and globus tickets available
id=`id -u`
mkdir             -p  ~/.krb5/
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
rm -f $condorOutput/$book/$dataset/fileList-all.$$.txt
touch $condorOutput/$book/$dataset/fileList-all.$$.txt

cat   $condorOutput/$book/$dataset/fileList-all.$$.txt-bak | grep root | sort | \
while read line
do
  size=`echo $line | tr -s ' ' | cut -d ' ' -f 1`
  file=`echo $line | tr -s ' ' | cut -d ' ' -f 2`
  file=`basename $file`
  echo "$size $file" >> $condorOutput/$book/$dataset/fileList-all.$$.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

#dcache=/pnfs/cmsaf.mit.edu/t2bat/cms/store/user/paus
#list $opt $dcache/$book/$dataset $target/$book/$dataset | grep root | sort \
list $opt $target/$book/$dataset | grep root | sort \
     > $condorOutput/$book/$dataset/fileList-done.$$.txt

diff -y $condorOutput/$book/$dataset/fileList-all.$$.txt  \
        $condorOutput/$book/$dataset/fileList-done.$$.txt > diff.$$
echo ""
echo " Files different in size: "
grep \| diff.$$
echo ""
echo " Files available in all and not done: "
grep \< diff.$$
echo ""
echo " Files done but not listed in all available: "
grep \> diff.$$
echo ""
rm      diff.$$

# 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

# stage in the missing files if it is at CERN
if [ "`echo $dataDir | grep /castor/cern.ch`" != "" ]
then
  echo "  scp $condorOutput/$book/$dataset/fileList.$$.txt  $TICKET_HOLDER@lxplus.cern.ch:"
  scp $condorOutput/$book/$dataset/fileList.$$.txt  $TICKET_HOLDER@lxplus.cern.ch:
  echo "  ssh $TICKET_HOLDER@lxplus.cern.ch ./stageSample.py --dataDir=$dataDir/$book/$dataset --fileList=fileList.$$.txt"
  ssh $TICKET_HOLDER@lxplus.cern.ch ./stageSample.py --dataDir=$dataDir/$book/$dataset --fileList=fileList.$$.txt
  echo "  ssh $TICKET_HOLDER@lxplus.cern.ch rm fileList.$$.txt"
  ssh $TICKET_HOLDER@lxplus.cern.ch rm fileList.$$.txt
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_$$.cmd <<EOF
Universe                = vanilla
Requirements            = ( (Arch == "INTEL") && (Disk >= DiskUsage) && ((Memory * 1024) >= ImageSize) && (HasFileTransfer) )
Notify_user             = $TICKET_HOLDER@mit.edu
Notification            = Error
Executable              = $script
Arguments               = $dataDir $book $dataset $target $condorOutput $$ $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:	1.12
Committed:	Wed Jul 25 03:05:45 2012 UTC (12 years, 9 months ago) by paus
Content type:	application/x-sh
Branch:	MAIN
CVS Tags:	Mit_029_pre1
Changes since 1.11:	+18 -5 lines
Log Message:	Preparing for new tag (Mit_029).
#	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	paus	1.11	echo " downloadSample.sh $*"
9	paus	1.1	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	1.11	#echo "rm -f $condorOutput/$book/$dataset/fileList.$$.txt"
37			rm -f $condorOutput/$book/$dataset/fileList.$$.txt
38	paus	1.1
39	paus	1.3	# make list of all local files
40	paus	1.4	if [ "`echo $HOSTNAME \| grep mit.edu`" != "" ] && \
41	paus	1.12	( [ "`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	1.11	list $opt $dataDir/$book/$dataset \| sort > $condorOutput/$book/$dataset/fileList-all.$$.txt-bak
50	paus	1.3
51	paus	1.11	# Make sure there are kerberos and globus tickets available
52	paus	1.3	id=`id -u`
53	paus	1.11	mkdir -p ~/.krb5/
54	paus	1.3	cp /tmp/x509up_u${id} ~/.krb5/
55	paus	1.1	KRB5CCNAME=`klist -5 \| grep 'Ticket cache:' \| cut -d' ' -f 3`
56			if ! [ -z $KRB5CCNAME ]
57			then
58	paus	1.12	mkdir -p ~/.krb5/
59	paus	1.1	chmod 0 ~/.krb5
60			chmod u=rwx ~/.krb5
61			file=`echo $KRB5CCNAME \| cut -d: -f2`
62			if [ -f "$file" ]
63			then
64			cp $file ~/.krb5/ticket
65			else
66			echo " ERROR -- missing kerberos ticket ($KRB5CCNAME)."
67			exit 1
68			fi
69			else
70			echo " ERROR -- missing kerberos ticket ($KRB5CCNAME)."
71			fi
72
73			# make list of all remote files
74	paus	1.11	rm -f $condorOutput/$book/$dataset/fileList-all.$$.txt
75			touch $condorOutput/$book/$dataset/fileList-all.$$.txt
76
77			cat $condorOutput/$book/$dataset/fileList-all.$$.txt-bak \| grep root \| sort \| \
78	paus	1.1	while read line
79			do
80			size=`echo $line \| tr -s ' ' \| cut -d ' ' -f 1`
81			file=`echo $line \| tr -s ' ' \| cut -d ' ' -f 2`
82			file=`basename $file`
83	paus	1.11	echo "$size $file" >> $condorOutput/$book/$dataset/fileList-all.$$.txt
84	paus	1.1	done
85
86			# make list of all local files
87	paus	1.9	if [ "`echo $HOSTNAME \| grep mit.edu`" != "" ] && \
88			( [ "`echo $dataDir \| grep /castor/cern.ch`" != "" ] \|\| \
89			[ "`echo $target \| grep /castor/cern.ch`" != "" ] )
90	paus	1.1	then
91	paus	1.3	opt="--simple"
92			else
93			opt=""
94	paus	1.1	fi
95	paus	1.3
96	paus	1.12	#dcache=/pnfs/cmsaf.mit.edu/t2bat/cms/store/user/paus
97			#list $opt $dcache/$book/$dataset $target/$book/$dataset \| grep root \| sort \
98			list $opt $target/$book/$dataset \| grep root \| sort \
99	paus	1.11	> $condorOutput/$book/$dataset/fileList-done.$$.txt
100
101			diff -y $condorOutput/$book/$dataset/fileList-all.$$.txt \
102			$condorOutput/$book/$dataset/fileList-done.$$.txt > diff.$$
103			echo ""
104			echo " Files different in size: "
105			grep \\| diff.$$
106			echo ""
107			echo " Files available in all and not done: "
108			grep \< diff.$$
109			echo ""
110			echo " Files done but not listed in all available: "
111			grep \> diff.$$
112			echo ""
113			rm diff.$$
114	paus	1.1
115			# make list of missing files
116	paus	1.11	rm -f $condorOutput/$book/$dataset/fileList.$$.txt
117			touch $condorOutput/$book/$dataset/fileList.$$.txt
118
119			cat $condorOutput/$book/$dataset/fileList-all.$$.txt \| grep root \| \
120	paus	1.1	while read line
121			do
122			size=`echo $line \| tr -s ' ' \| cut -d ' ' -f 1`
123			file=`echo $line \| tr -s ' ' \| cut -d ' ' -f 2`
124	paus	1.11	exists=`grep "$file" $condorOutput/$book/$dataset/fileList-done.$$.txt`
125	paus	1.1	if [ "$exists" == "" ]
126			then
127	paus	1.3	echo " -missing-- $file with $size bytes"
128	paus	1.11	echo "$size $file" >> $condorOutput/$book/$dataset/fileList.$$.txt
129	paus	1.1	# else
130	paus	1.3	# echo " -exists--- $file with $size bytes - exists"
131			else
132			# now check that size matches
133	paus	1.11	test=`grep "$size $file" $condorOutput/$book/$dataset/fileList-done.$$.txt`
134	paus	1.3	if [ "$test" == "" ]
135			then
136			if [ "$onlyMissing" == "" ]
137			then
138			echo " -fileSize- $exists (remote: $size)"
139	paus	1.11	echo "$size $file" >> $condorOutput/$book/$dataset/fileList.$$.txt
140	paus	1.3	fi
141			fi
142	paus	1.1	fi
143			done
144	paus	1.11
145			nAll=`wc -l $condorOutput/$book/$dataset/fileList-all.$$.txt \| cut -d ' ' -f1`
146			nMissing=`wc -l $condorOutput/$book/$dataset/fileList.$$.txt \| cut -d ' ' -f1`
147			nDone=`wc -l $condorOutput/$book/$dataset/fileList-done.$$.txt \| cut -d ' ' -f1`
148	paus	1.10	echo ""
149			echo " Download Summary "
150			echo " All $nAll"
151			echo " Done $nDone"
152			echo " Missing $nMissing"
153			echo ""
154	paus	1.1
155			# construct our job
156	paus	1.11	nFiles=`wc -l $condorOutput/$book/$dataset/fileList.$$.txt \| cut -d ' ' -f1`
157	paus	1.1	if [ "$nFiles" == "" ] \|\| [ "$nFiles" == "0" ]
158			then
159			echo " "
160			echo " No more files to download. EXIT."
161			exit 0
162			elif [ $nFiles -lt $nCopyProcs ]
163			then
164			nCopyProcs=$nFiles
165			fi
166	paus	1.4	# how many files per job?
167	paus	1.1	nFilesPerJob=$(( $nFiles/$nCopyProcs ))
168	paus	1.4	nFilesTmp=$(( $nFilesPerJob*$nCopyProcs ))
169			if [ $nFilesPerJob == 1 ] && [ $nFiles -gt $nCopyProcs ]
170			then
171			nFilesPerJob=2
172			elif [ $nFilesTmp -lt $nFiles ]
173			then
174			nFilesPerJob=$(( $nFilesPerJob+1 ))
175			fi
176
177	paus	1.10	echo " n files to copy: $nFiles"
178			echo " n files/proc : $nFilesPerJob"
179	paus	1.1
180			i=1
181			next=1
182			last=$nFilesPerJob
183
184	paus	1.3	# make sure condor is properly setup for us
185			if ! [ -z $CONDOR_LOCATION ]
186			then
187			unset CONDOR_LOCATION
188			export CONDOR_CONFIG=/usr/local/condor/etc/condor_config
189			fi
190
191	paus	1.12	# stage in the missing files if it is at CERN
192			if [ "`echo $dataDir \| grep /castor/cern.ch`" != "" ]
193			then
194			echo " scp $condorOutput/$book/$dataset/fileList.$$.txt $TICKET_HOLDER@lxplus.cern.ch:"
195			scp $condorOutput/$book/$dataset/fileList.$$.txt $TICKET_HOLDER@lxplus.cern.ch:
196			echo " ssh $TICKET_HOLDER@lxplus.cern.ch ./stageSample.py --dataDir=$dataDir/$book/$dataset --fileList=fileList.$$.txt"
197			ssh $TICKET_HOLDER@lxplus.cern.ch ./stageSample.py --dataDir=$dataDir/$book/$dataset --fileList=fileList.$$.txt
198			echo " ssh $TICKET_HOLDER@lxplus.cern.ch rm fileList.$$.txt"
199			ssh $TICKET_HOLDER@lxplus.cern.ch rm fileList.$$.txt
200			fi
201
202	paus	1.3	# loop over the condor jobs and submit them
203	paus	1.4	while [ $i -le $nCopyProcs ] && [ $last -le $nFiles ]
204	paus	1.1	do
205			if [ $i == $nCopyProcs ]
206			then
207			last=$nFiles
208			fi
209
210			# say what we are going to submit
211	paus	1.11	echo " downloadFiles.sh $dataDir $book $dataset $target $condorOutput $$ $next $last"
212	paus	1.1
213			logFile=`echo download:$book/$dataset/${next}-${last}.txt \| tr '/' '+'`
214			logFile=/tmp/$logFile
215			rm -f $logFile
216
217			# prepare the condor_submit files
218			cat > submit_$$.cmd <<EOF
219			Universe = vanilla
220	paus	1.7	Requirements = ( (Arch == "INTEL") && (Disk >= DiskUsage) && ((Memory * 1024) >= ImageSize) && (HasFileTransfer) )
221	paus	1.12	Notify_user = $TICKET_HOLDER@mit.edu
222	paus	1.1	Notification = Error
223			Executable = $script
224	paus	1.11	Arguments = $dataDir $book $dataset $target $condorOutput $$ $next $last
225	paus	1.1	Rank = Mips
226			GetEnv = True
227			Input = /dev/null
228			Output = $condorOutput/$book/$dataset/${next}-${last}.out
229			Error = $condorOutput/$book/$dataset/${next}-${last}.err
230			Log = $logFile
231			should_transfer_files = YES
232			when_to_transfer_output = ON_EXIT
233	paus	1.3
234			+AccountingGroup = "group_cmsuser.cmsu0284"
235
236	paus	1.1	Queue
237			EOF
238
239	paus	1.3	#+x509userproxysubject = $DN
240
241	paus	1.1	# submit the jobs
242			condor_submit submit_$$.cmd >& /dev/null #>& lastSub
243			#cat submit_$$.cmd
244			rm submit_$$.cmd
245
246			# update counters
247			next=$(( $next + $nFilesPerJob ))
248			last=$(( $last + $nFilesPerJob ))
249			i=$(( $i + 1 ))
250			done
251
252			exit 0