1 |
svn |
2581 |
#!/bin/bash |
2 |
|
|
|
3 |
|
|
if [ ! "x$0" = "x${0#-}" ] |
4 |
|
|
then |
5 |
|
|
echo "Don't source me, just run!" 1>&2 |
6 |
|
|
return |
7 |
|
|
fi |
8 |
|
|
. ${0%/*}/libdans.sh |
9 |
|
|
|
10 |
|
|
# Load global defaults (vars: vo, archive, LFC_*, SRM_*) |
11 |
|
|
loadDefaults $0 |
12 |
|
|
|
13 |
|
|
|
14 |
|
|
DEBUG=0 |
15 |
|
|
QUIET=0 |
16 |
|
|
KEEPGOING=0 |
17 |
|
|
START_INDEX=1 |
18 |
|
|
END_INDEX=0 |
19 |
|
|
USAGE="\ |
20 |
|
|
$0 - verify the contents of one or multiple tar.gz files |
21 |
|
|
Usage: ./check-file [-d|--debug] [-k|--keepgoing] [file|lfn|surl] [[file2|lfn2|surl2] ...] |
22 |
|
|
Where: |
23 |
|
|
--keepgoing tells $0 to keep going after an error |
24 |
|
|
file specifies the full path to a local file OR |
25 |
|
|
lfn specifies a Logical File Name on the LFC OR |
26 |
|
|
surl specifies a Storage URL (srm:// or gsiftp://) |
27 |
|
|
" |
28 |
|
|
|
29 |
|
|
# Parse commandline parameters |
30 |
|
|
while [ $# -gt 0 ] |
31 |
|
|
do |
32 |
|
|
case "$1" in |
33 |
|
|
(-h|--help) echo "${USAGE}" |
34 |
|
|
exit 0 |
35 |
|
|
;; |
36 |
|
|
(-d|--debug) let DEBUG+=1 |
37 |
|
|
;; |
38 |
|
|
(-k|--keepgoing) KEEPGOING=1 |
39 |
|
|
;; |
40 |
|
|
(-*) echo "Invalid option: $1" |
41 |
|
|
exit 1 |
42 |
|
|
;; |
43 |
|
|
(*) break |
44 |
|
|
;; |
45 |
|
|
esac |
46 |
|
|
shift |
47 |
|
|
done |
48 |
|
|
|
49 |
|
|
if [ $# -lt 1 ] |
50 |
|
|
then |
51 |
|
|
abort 1 "${USAGE}" |
52 |
|
|
fi |
53 |
|
|
|
54 |
|
|
# Loop over all files|lfns|surls |
55 |
|
|
for file |
56 |
|
|
do |
57 |
|
|
do_get=0 |
58 |
|
|
if [ "${file}" != "${file#lfn://}" ] |
59 |
|
|
then |
60 |
|
|
do_get=1 |
61 |
|
|
fi |
62 |
|
|
if [ "${file}" != "${file#srm://}" ] |
63 |
|
|
then |
64 |
|
|
do_get=1 |
65 |
|
|
fi |
66 |
|
|
if [ "${file}" != "${file#gsiftp://}" ] |
67 |
|
|
then |
68 |
|
|
do_get=1 |
69 |
|
|
fi |
70 |
|
|
|
71 |
|
|
if [ ${do_get} -ne 0 ] |
72 |
|
|
then |
73 |
|
|
info "Retrieving file ${file}" |
74 |
|
|
lcg-cp $SRM_TIMEOUTS "${file}" file://$PWD/"${file##*/}" |
75 |
|
|
if [ $? -ne 0 ] |
76 |
|
|
then |
77 |
|
|
warn "First attempt failed, sleeping 30 seconds and retrying..." |
78 |
|
|
sleep 30 |
79 |
|
|
lcg-cp $SRM_TIMEOUTS "${file}" file://$PWD/"${file##*/}" |
80 |
|
|
fi |
81 |
|
|
if [ $? -ne 0 ] |
82 |
|
|
then |
83 |
|
|
warn "Second attempt failed, sleeping 90 seconds and retrying..." |
84 |
|
|
sleep 90 |
85 |
|
|
date +"%Y/%m/%d-%H:%M:%S start lcg-cp -v -v $SRM_TIMEOUTS" 1>&2 |
86 |
|
|
lcg-cp -v -v $SRM_TIMEOUTS "${file}" file://$PWD/"${file##*/}" |
87 |
|
|
fi |
88 |
|
|
if [ $? -ne 0 ] |
89 |
|
|
then |
90 |
|
|
abort 3 "Error retrieving ${file}" |
91 |
|
|
fi |
92 |
|
|
|
93 |
|
|
file="${file##*/}" |
94 |
|
|
fi |
95 |
|
|
|
96 |
|
|
# the file should be a .tar.gz file, extract the original archive name from it |
97 |
|
|
base=`file "${file}" | sed -n 's/.*compressed data, was \"\(.*\)\",.*/\1/p'` |
98 |
|
|
archive="${base%%-[0-9]*}" |
99 |
|
|
|
100 |
|
|
debug "file=${file} base=${base} archive=${archive}" |
101 |
|
|
|
102 |
|
|
CHECKSUM_DIR="${HOME}/dans/${archive}/checksums" |
103 |
|
|
|
104 |
|
|
# fix numbering issue in RACM tar balls |
105 |
|
|
if [ "x${archive}" = "xRACM" ] |
106 |
|
|
then |
107 |
|
|
index=${base#${archive}-} |
108 |
|
|
index=${index%.tar} |
109 |
|
|
if [ ${#index} -lt 4 ] |
110 |
|
|
then |
111 |
|
|
base="${archive}-0${index}.tar" |
112 |
|
|
fi |
113 |
|
|
fi |
114 |
|
|
|
115 |
|
|
# extract the tarball in a separate directory, for easy cleanup later |
116 |
|
|
mkdir -p workspace |
117 |
|
|
info "Extracting file ${file} (real name: ${base}.gz) ..." |
118 |
|
|
tar -C workspace -xzf "${file}" |
119 |
|
|
|
120 |
|
|
md5file="${base}.md5sum" |
121 |
|
|
if [ -r "${md5file}" ] |
122 |
|
|
then |
123 |
|
|
warn "\"${md5file}\" exists, append \".$$\"" |
124 |
|
|
md5file="${md5file}.$$" |
125 |
|
|
fi |
126 |
|
|
|
127 |
|
|
info "Calculating checksums for all files in archive (output=${md5file})" |
128 |
|
|
echo "# ${base} START" > "${md5file}" |
129 |
|
|
(cd workspace; md5deep -l -r ${archive}) | sort -k 2 >> "${md5file}" |
130 |
|
|
echo "# ${base} END" >> "${md5file}" |
131 |
|
|
|
132 |
|
|
if [ -r "${CHECKSUM_DIR}/${md5file%.$$}" ] |
133 |
|
|
then |
134 |
|
|
debug "Comparing ${md5file} to ${CHECKSUM_DIR}/${md5file%.$$}" |
135 |
|
|
|
136 |
|
|
diff "${CHECKSUM_DIR}/${md5file%.$$}" "${md5file}" && info "OK" || warn "Mismatch" |
137 |
|
|
else |
138 |
|
|
debug "${CHECKSUM_DIR}/${md5file%.$$} not found, skipping" |
139 |
|
|
fi |
140 |
|
|
|
141 |
|
|
if [ "${file##*/}" != "${base}.gz" ] |
142 |
|
|
then |
143 |
|
|
grep "${file##*/}" "${md5file}" |
144 |
|
|
fi |
145 |
|
|
|
146 |
|
|
# delete the tar ball (if downloaded!) |
147 |
|
|
if [ ${do_get} -ne 0 ] |
148 |
|
|
then |
149 |
|
|
rm -f "${file}" |
150 |
|
|
fi |
151 |
|
|
# clear out the workspace |
152 |
|
|
rm -rf workspace |
153 |
|
|
done |
154 |
|
|
|