annotate unarchiver.pl @ 55:7958ac21f9bf

fixed unarchiving of multiple versions by adding -ifnewer option to dsmc ret
author casties
date Tue, 11 Aug 2009 16:34:59 +0200
parents b762b5af6e42
children 2208ed7370cb
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
1 #!/usr/local/bin/perl -w
30497c6a3eca Initial revision
casties
parents:
diff changeset
2
30497c6a3eca Initial revision
casties
parents:
diff changeset
3 use strict;
30497c6a3eca Initial revision
casties
parents:
diff changeset
4
30497c6a3eca Initial revision
casties
parents:
diff changeset
5 use XML::LibXML;
7
c4e6fc065b6d fixed problem with dsmc because of new filespace.
casties
parents: 0
diff changeset
6 use FileHandle;
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
7
30497c6a3eca Initial revision
casties
parents:
diff changeset
8 # MPIWG libraries
30497c6a3eca Initial revision
casties
parents:
diff changeset
9 use lib '/usr/local/mpiwg/archive';
30497c6a3eca Initial revision
casties
parents:
diff changeset
10 use MPIWGStor;
30497c6a3eca Initial revision
casties
parents:
diff changeset
11
30497c6a3eca Initial revision
casties
parents:
diff changeset
12 # make output unbuffered
30497c6a3eca Initial revision
casties
parents:
diff changeset
13 $|=1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
14
30497c6a3eca Initial revision
casties
parents:
diff changeset
15
30497c6a3eca Initial revision
casties
parents:
diff changeset
16 #######################################################
30497c6a3eca Initial revision
casties
parents:
diff changeset
17 # internal parameters
30497c6a3eca Initial revision
casties
parents:
diff changeset
18 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
19
30497c6a3eca Initial revision
casties
parents:
diff changeset
20 # program version
55
7958ac21f9bf fixed unarchiving of multiple versions by adding -ifnewer option to dsmc ret
casties
parents: 52
diff changeset
21 my $version = "0.3.4 (11.8.2009)";
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
22
30497c6a3eca Initial revision
casties
parents:
diff changeset
23 # read command line parameters
30497c6a3eca Initial revision
casties
parents:
diff changeset
24 my $args = parseargs;
30497c6a3eca Initial revision
casties
parents:
diff changeset
25
30497c6a3eca Initial revision
casties
parents:
diff changeset
26 # debug level
30497c6a3eca Initial revision
casties
parents:
diff changeset
27 $debug = (exists $$args{'debug'}) ? ($$args{'debug'}) : 0;
30497c6a3eca Initial revision
casties
parents:
diff changeset
28
30497c6a3eca Initial revision
casties
parents:
diff changeset
29 # rewrite XML file (necessary for archive date!)
30497c6a3eca Initial revision
casties
parents:
diff changeset
30 my $fix_xml = 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
31 my $xml_changed = 0;
30497c6a3eca Initial revision
casties
parents:
diff changeset
32 # XML namespace (not really implemented!)
30497c6a3eca Initial revision
casties
parents:
diff changeset
33 my $namespace = "";
30497c6a3eca Initial revision
casties
parents:
diff changeset
34
30497c6a3eca Initial revision
casties
parents:
diff changeset
35 # archive name (archive-path element, usually == $docdir)
30497c6a3eca Initial revision
casties
parents:
diff changeset
36 my $archname;
30497c6a3eca Initial revision
casties
parents:
diff changeset
37 # archive storage date
30497c6a3eca Initial revision
casties
parents:
diff changeset
38 my $archdate;
30497c6a3eca Initial revision
casties
parents:
diff changeset
39
30497c6a3eca Initial revision
casties
parents:
diff changeset
40 #######################################################
30497c6a3eca Initial revision
casties
parents:
diff changeset
41 # external programs
30497c6a3eca Initial revision
casties
parents:
diff changeset
42 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
43 my $archprog = "/opt/tivoli/tsm/client/ba/bin/dsmc";
30497c6a3eca Initial revision
casties
parents:
diff changeset
44 if (! -x $archprog) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
45 logger("ABORT", "TSM client program '$archprog' missing!!");
30497c6a3eca Initial revision
casties
parents:
diff changeset
46 exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
47 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
48 my $checkprog = "/usr/local/mpiwg/archive/archivecheck";
30497c6a3eca Initial revision
casties
parents:
diff changeset
49 if (! -x $checkprog) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
50 logger("ABORT", "archive checking program '$checkprog' missing!!");
30497c6a3eca Initial revision
casties
parents:
diff changeset
51 exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
52 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
53 # log file for archiver
30497c6a3eca Initial revision
casties
parents:
diff changeset
54 my $log_file = "/var/tmp/unarchiver.log";
30497c6a3eca Initial revision
casties
parents:
diff changeset
55 if (! open LOG, ">>$log_file") {
30497c6a3eca Initial revision
casties
parents:
diff changeset
56 logger("ABORT", "unable to write log file '$log_file'!!");
30497c6a3eca Initial revision
casties
parents:
diff changeset
57 exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
58 }
7
c4e6fc065b6d fixed problem with dsmc because of new filespace.
casties
parents: 0
diff changeset
59 LOG->autoflush(1);
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
60
30497c6a3eca Initial revision
casties
parents:
diff changeset
61 #######################################################
30497c6a3eca Initial revision
casties
parents:
diff changeset
62 # check parameters that were passed to the program
30497c6a3eca Initial revision
casties
parents:
diff changeset
63 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
64 my $docdir = $$args{'path'};
30497c6a3eca Initial revision
casties
parents:
diff changeset
65 if (! $docdir) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
66 print "ABORT: no document directory given!\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
67 exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
68 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
69 # strip trailing slashes
30497c6a3eca Initial revision
casties
parents:
diff changeset
70 $docdir =~ s/\/$//;
30497c6a3eca Initial revision
casties
parents:
diff changeset
71 if (! -d $docdir) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
72 print "ABORT: document directory \'$docdir\' doesn't exist!\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
73 exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
74 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
75
30497c6a3eca Initial revision
casties
parents:
diff changeset
76 my $metafile = "$docdir/index.meta";
30497c6a3eca Initial revision
casties
parents:
diff changeset
77 if (! -f $metafile) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
78 print "ABORT: metadata index file \'$metafile\' doesn't exist!\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
79 exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
80 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
81
30497c6a3eca Initial revision
casties
parents:
diff changeset
82 #######################################################
30497c6a3eca Initial revision
casties
parents:
diff changeset
83 # internal variables
30497c6a3eca Initial revision
casties
parents:
diff changeset
84 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
85
30497c6a3eca Initial revision
casties
parents:
diff changeset
86 # number of errors
30497c6a3eca Initial revision
casties
parents:
diff changeset
87 my $errcnt = 0;
30497c6a3eca Initial revision
casties
parents:
diff changeset
88 # number of warnings
30497c6a3eca Initial revision
casties
parents:
diff changeset
89 my $warncnt = 0;
30497c6a3eca Initial revision
casties
parents:
diff changeset
90
30497c6a3eca Initial revision
casties
parents:
diff changeset
91 #######################################################
30497c6a3eca Initial revision
casties
parents:
diff changeset
92 # subroutines
30497c6a3eca Initial revision
casties
parents:
diff changeset
93 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
94
10
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
95 # construct document's parent dir
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
96 sub get_parent {
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
97 my ($dirname) = @_;
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
98 my $dirparent = $dirname;
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
99 $dirparent =~ s!/[^/]+$!!;
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
100 return $dirparent;
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
101 }
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
102
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
103
30497c6a3eca Initial revision
casties
parents:
diff changeset
104 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
105 # $files = read_resource_meta($rootnode)
30497c6a3eca Initial revision
casties
parents:
diff changeset
106 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
107 # checks general resource meta information and reads the list of files
30497c6a3eca Initial revision
casties
parents:
diff changeset
108 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
109 sub read_resource_meta {
30497c6a3eca Initial revision
casties
parents:
diff changeset
110 my ($rootnode) = @_;
30497c6a3eca Initial revision
casties
parents:
diff changeset
111 my %files;
30497c6a3eca Initial revision
casties
parents:
diff changeset
112 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
113 # archive path
30497c6a3eca Initial revision
casties
parents:
diff changeset
114 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
115 # get archive-path
30497c6a3eca Initial revision
casties
parents:
diff changeset
116 $archname = sstrip($rootnode->findvalue('child::archive-path'));
30497c6a3eca Initial revision
casties
parents:
diff changeset
117 if (! $archname) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
118 logger("ABORT", "archive-name element missing!!");
30497c6a3eca Initial revision
casties
parents:
diff changeset
119 exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
120 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
121
30497c6a3eca Initial revision
casties
parents:
diff changeset
122 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
123 # files
30497c6a3eca Initial revision
casties
parents:
diff changeset
124 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
125 my @filenodes = $rootnode->findnodes('child::file');
30497c6a3eca Initial revision
casties
parents:
diff changeset
126 foreach my $fn (@filenodes) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
127 my $name = sstrip($fn->findvalue('child::name'));
30497c6a3eca Initial revision
casties
parents:
diff changeset
128 my $path = sstrip($fn->findvalue('child::path'));
30497c6a3eca Initial revision
casties
parents:
diff changeset
129 logger("DEBUG", "FILE: ($path)$name");
30497c6a3eca Initial revision
casties
parents:
diff changeset
130 my $f = ($path) ? "$path/$name" : "$name";
30497c6a3eca Initial revision
casties
parents:
diff changeset
131 $files{$f} = [$name];
30497c6a3eca Initial revision
casties
parents:
diff changeset
132 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
133
30497c6a3eca Initial revision
casties
parents:
diff changeset
134 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
135 # dirs
30497c6a3eca Initial revision
casties
parents:
diff changeset
136 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
137 my @dirnodes = $rootnode->findnodes('child::dir');
30497c6a3eca Initial revision
casties
parents:
diff changeset
138 foreach my $fn (@dirnodes) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
139 my $name = sstrip($fn->findvalue('child::name'));
30497c6a3eca Initial revision
casties
parents:
diff changeset
140 my $path = sstrip($fn->findvalue('child::path'));
30497c6a3eca Initial revision
casties
parents:
diff changeset
141 logger("DEBUG", "DIR: ($path)$name");
30497c6a3eca Initial revision
casties
parents:
diff changeset
142 my $f = ($path) ? "$path/$name" : "$name";
30497c6a3eca Initial revision
casties
parents:
diff changeset
143 $files{$f} = [$name];
30497c6a3eca Initial revision
casties
parents:
diff changeset
144 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
145
30497c6a3eca Initial revision
casties
parents:
diff changeset
146 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
147 # archive-storage-date
30497c6a3eca Initial revision
casties
parents:
diff changeset
148 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
149 my $archdate = $rootnode->find('child::archive-storage-date');
30497c6a3eca Initial revision
casties
parents:
diff changeset
150 if ($archdate) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
151 logger("INFO", "archive storage date: $archdate");
30497c6a3eca Initial revision
casties
parents:
diff changeset
152 } else {
30497c6a3eca Initial revision
casties
parents:
diff changeset
153 logger("ERROR", "archive storage date missing!");
30497c6a3eca Initial revision
casties
parents:
diff changeset
154 $errcnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
155 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
156
30497c6a3eca Initial revision
casties
parents:
diff changeset
157 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
158 # archive-recall-date
30497c6a3eca Initial revision
casties
parents:
diff changeset
159 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
160 my $recalldatenode = ($rootnode->find('child::archive-recall-date'))->get_node(1);
30497c6a3eca Initial revision
casties
parents:
diff changeset
161 if ($recalldatenode) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
162 print "INFO: archive recall date exists!\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
163 # delete old date
30497c6a3eca Initial revision
casties
parents:
diff changeset
164 $recalldatenode->removeChildNodes;
30497c6a3eca Initial revision
casties
parents:
diff changeset
165 } else {
30497c6a3eca Initial revision
casties
parents:
diff changeset
166 # create new storage date node
30497c6a3eca Initial revision
casties
parents:
diff changeset
167 $recalldatenode = $rootnode->addNewChild($namespace, "archive-recall-date");
30497c6a3eca Initial revision
casties
parents:
diff changeset
168 # move after archive-path
30497c6a3eca Initial revision
casties
parents:
diff changeset
169 $rootnode->insertAfter($recalldatenode, ($rootnode->find('child::archive-storage-date'))->get_node(1));
30497c6a3eca Initial revision
casties
parents:
diff changeset
170 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
171 $recalldatenode->appendTextNode(scalar localtime);
30497c6a3eca Initial revision
casties
parents:
diff changeset
172 $xml_changed++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
173
30497c6a3eca Initial revision
casties
parents:
diff changeset
174 return \%files;
30497c6a3eca Initial revision
casties
parents:
diff changeset
175 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
176
30497c6a3eca Initial revision
casties
parents:
diff changeset
177
30497c6a3eca Initial revision
casties
parents:
diff changeset
178 #
10
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
179 # $num_files = run_retrieve($docdir, $docmount, \%files)
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
180 #
10
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
181 # Runs the retriever program on $docdir and returns the number of unarchived files.
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
182 # All filenames are put in %files.
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
183 # $docmount is the mount point of the doc partition in cases when the new mount point
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
184 # is different.
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
185 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
186 # Sample output:
7
c4e6fc065b6d fixed problem with dsmc because of new filespace.
casties
parents: 0
diff changeset
187 # (old!) Retrieving 17,234 /mpiwg/archive/data/test/auto_titit_123/pageimg/essen-wind1.jpg [Done]
c4e6fc065b6d fixed problem with dsmc because of new filespace.
casties
parents: 0
diff changeset
188 # Retrieving 42,406,326 /mpiwg/archive/data/library/B980G582/raw/00015.tif --> /mpiwg/archive/data/library/B980G582/raw/00015.tif [Done]
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
189 sub run_retrieve {
10
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
190 my ($archdir, $archmount, $files) = @_;
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
191 my $archparent;
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
192 if ($archmount eq $archdir) {
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
193 # no explicit mount point
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
194 $archparent = get_parent($archdir);
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
195 } else {
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
196 # destination dir is mount point
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
197 $archparent = $archmount;
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
198 }
20
79c6618e8dfa small fixes
casties
parents: 11
diff changeset
199 logger("INFO", "looking for archives in $archmount...");
10
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
200
7
c4e6fc065b6d fixed problem with dsmc because of new filespace.
casties
parents: 0
diff changeset
201 print LOG "START unarchive $version on ", scalar localtime, "\n";
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
202 my $archcmd = $archprog;
55
7958ac21f9bf fixed unarchiving of multiple versions by adding -ifnewer option to dsmc ret
casties
parents: 52
diff changeset
203 $archcmd .= " retrieve -subdir=yes -replace=all -ifnewer";
10
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
204 $archcmd .= " -description='$archname'"; # archive name
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
205 $archcmd .= " '$archmount/'"; # archive mount point
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
206 $archcmd .= " '$archparent/'"; # destination dir name
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
207
50
724c615b5982 added please-wait-message
casties
parents: 20
diff changeset
208 logger('INFO', "querying TSM server for $archmount, please wait...");
724c615b5982 added please-wait-message
casties
parents: 20
diff changeset
209
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
210 my $archcnt = 0;
10
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
211 my $numfiles = 0;
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
212 print LOG "CMD: $archcmd\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
213 if (open ARCH, "$archcmd 2>&1 |") {
30497c6a3eca Initial revision
casties
parents:
diff changeset
214 while (<ARCH>) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
215 chomp;
30497c6a3eca Initial revision
casties
parents:
diff changeset
216 print LOG "ARCH: $_\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
217 if (/
30497c6a3eca Initial revision
casties
parents:
diff changeset
218 Retrieving
30497c6a3eca Initial revision
casties
parents:
diff changeset
219 \s+([\d,]+) # size
30497c6a3eca Initial revision
casties
parents:
diff changeset
220 \s+(\S+) # file name
7
c4e6fc065b6d fixed problem with dsmc because of new filespace.
casties
parents: 0
diff changeset
221 \s+-->
c4e6fc065b6d fixed problem with dsmc because of new filespace.
casties
parents: 0
diff changeset
222 \s+(\S+) # destination file name
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
223 \s+\[Done\]
30497c6a3eca Initial revision
casties
parents:
diff changeset
224 /x) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
225 my $size = $1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
226 my $file = $2;
30497c6a3eca Initial revision
casties
parents:
diff changeset
227 $size =~ s/,//g;
30497c6a3eca Initial revision
casties
parents:
diff changeset
228 logger("DEBUG", " RETRIEVE: file '$file'");
30497c6a3eca Initial revision
casties
parents:
diff changeset
229 $archcnt++;
10
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
230 if ($$files{$file}) {
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
231 logger("WARNING", "file $file seems to be archived multiple times.");
30497c6a3eca Initial revision
casties
parents:
diff changeset
232 $warncnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
233 }
10
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
234 $$files{$file} = [$size];
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
235 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
236 }
10
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
237 $numfiles = (scalar keys %$files);
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
238 logger("INFO", "$archcnt archives of $numfiles files (in $archmount).");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
239 } else {
30497c6a3eca Initial revision
casties
parents:
diff changeset
240 logger("ABORT", "unable to start archive command '$archcmd'!!");
30497c6a3eca Initial revision
casties
parents:
diff changeset
241 exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
242 }
10
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
243 return $numfiles;
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
244 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
245
30497c6a3eca Initial revision
casties
parents:
diff changeset
246
30497c6a3eca Initial revision
casties
parents:
diff changeset
247 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
248 # check_files(\%files_to_retrieve, \%retrieved_files)
30497c6a3eca Initial revision
casties
parents:
diff changeset
249 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
250 # compares the list of archived and retrieved files
30497c6a3eca Initial revision
casties
parents:
diff changeset
251 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
252 sub check_files {
30497c6a3eca Initial revision
casties
parents:
diff changeset
253 my ($to_retrieve, $retrieved) = @_;
30497c6a3eca Initial revision
casties
parents:
diff changeset
254
30497c6a3eca Initial revision
casties
parents:
diff changeset
255 my $nt = scalar keys %$to_retrieve;
30497c6a3eca Initial revision
casties
parents:
diff changeset
256 my $na = scalar keys %$retrieved;
30497c6a3eca Initial revision
casties
parents:
diff changeset
257
30497c6a3eca Initial revision
casties
parents:
diff changeset
258 foreach my $ft (sort keys %$to_retrieve) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
259 my $fp = "$docdir/$ft";
30497c6a3eca Initial revision
casties
parents:
diff changeset
260 #logger("DEBUG", " fp: $fp");
30497c6a3eca Initial revision
casties
parents:
diff changeset
261 if ($$retrieved{$fp}) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
262 logger("DEBUG", "$ft retrieved OK");
30497c6a3eca Initial revision
casties
parents:
diff changeset
263 $$retrieved{$fp}->[1] = "OK";
30497c6a3eca Initial revision
casties
parents:
diff changeset
264 } else {
30497c6a3eca Initial revision
casties
parents:
diff changeset
265 logger("ERROR", "file entry '$ft' missing from archive!");
30497c6a3eca Initial revision
casties
parents:
diff changeset
266 $errcnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
267 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
268 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
269
30497c6a3eca Initial revision
casties
parents:
diff changeset
270 foreach my $fa (sort keys %$retrieved) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
271 if (! $$retrieved{$fa}->[1]) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
272 my ($fn, $fp) = split_file_path($fa);
30497c6a3eca Initial revision
casties
parents:
diff changeset
273 if ($index_files{$fn}) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
274 logger("DEBUG", "$fa ignored");
30497c6a3eca Initial revision
casties
parents:
diff changeset
275 $na--;
30497c6a3eca Initial revision
casties
parents:
diff changeset
276 } else {
30497c6a3eca Initial revision
casties
parents:
diff changeset
277 logger("WARNING", "$fa retrieved but not in list!");
30497c6a3eca Initial revision
casties
parents:
diff changeset
278 $warncnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
279 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
280 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
281 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
282
30497c6a3eca Initial revision
casties
parents:
diff changeset
283 if ($nt > $na) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
284 logger("WARNING", "less files were retrieved ($na vs. $nt)!");
30497c6a3eca Initial revision
casties
parents:
diff changeset
285 $warncnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
286 } elsif ($na > $nt) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
287 logger("WARNING", "more files were retrieved ($na vs. $nt)!");
30497c6a3eca Initial revision
casties
parents:
diff changeset
288 $warncnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
289 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
290
30497c6a3eca Initial revision
casties
parents:
diff changeset
291 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
292
30497c6a3eca Initial revision
casties
parents:
diff changeset
293
30497c6a3eca Initial revision
casties
parents:
diff changeset
294
30497c6a3eca Initial revision
casties
parents:
diff changeset
295 #######################################################
30497c6a3eca Initial revision
casties
parents:
diff changeset
296 # main
30497c6a3eca Initial revision
casties
parents:
diff changeset
297 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
298
30497c6a3eca Initial revision
casties
parents:
diff changeset
299 logger("INFO", "unarchiver $version");
30497c6a3eca Initial revision
casties
parents:
diff changeset
300
30497c6a3eca Initial revision
casties
parents:
diff changeset
301 # make shure the right user is running this program
52
b762b5af6e42 fixed problem with user recogintion by getlogin
casties
parents: 50
diff changeset
302 my $user = getlogin || getpwuid($<);
7
c4e6fc065b6d fixed problem with dsmc because of new filespace.
casties
parents: 0
diff changeset
303 if (($user)&&($user ne "archive")&&($user ne "root")) {
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
304 logger("ABORT", "you must be archive or root user to run this program!");
30497c6a3eca Initial revision
casties
parents:
diff changeset
305 exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
306 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
307
30497c6a3eca Initial revision
casties
parents:
diff changeset
308 # use checkarchive first
30497c6a3eca Initial revision
casties
parents:
diff changeset
309 if (system("$checkprog $docdir >/dev/null") == 0) {
10
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
310 logger("INFO", "archive \"$docdir\" check OK");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
311 } else {
10
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
312 logger("ABORT", "archive \"$docdir\" check failed!!");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
313 exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
314 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
315
30497c6a3eca Initial revision
casties
parents:
diff changeset
316 # read index.meta file
30497c6a3eca Initial revision
casties
parents:
diff changeset
317 my ($document, $rootnode) = read_xml($metafile);
30497c6a3eca Initial revision
casties
parents:
diff changeset
318
30497c6a3eca Initial revision
casties
parents:
diff changeset
319 # check index file
30497c6a3eca Initial revision
casties
parents:
diff changeset
320 my $archived_files = read_resource_meta($rootnode);
30497c6a3eca Initial revision
casties
parents:
diff changeset
321 my $num_archived_files = scalar keys %$archived_files;
30497c6a3eca Initial revision
casties
parents:
diff changeset
322
30497c6a3eca Initial revision
casties
parents:
diff changeset
323 # check for .archived file
30497c6a3eca Initial revision
casties
parents:
diff changeset
324 if (-f "$docdir/.archived") {
30497c6a3eca Initial revision
casties
parents:
diff changeset
325 logger("INFO", ".archived file exists.");
30497c6a3eca Initial revision
casties
parents:
diff changeset
326 } else {
30497c6a3eca Initial revision
casties
parents:
diff changeset
327 logger("WARNING", "no .archived file!");
30497c6a3eca Initial revision
casties
parents:
diff changeset
328 $warncnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
329 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
330
30497c6a3eca Initial revision
casties
parents:
diff changeset
331 logger("INFO", "$num_archived_files files to retrieve.");
30497c6a3eca Initial revision
casties
parents:
diff changeset
332
10
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
333 # save current index.meta
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
334 park_file($metafile);
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
335
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
336 # retrieve
10
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
337 my %retrieved_files = ();
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
338 my $archcnt = 0;
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
339
10
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
340 if ($docdir =~ /\/mpiwg\/archive\/data\/(.*)/) {
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
341 # TSM needs two different paths because of historical mount points :-(
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
342 # try the new one first
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
343 $archcnt = run_retrieve($docdir, "/mpiwg/archive", \%retrieved_files);
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
344 if ($archcnt == 0) {
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
345 # and then the old one
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
346 $archcnt = run_retrieve($docdir, "/mpiwg/archive/data", \%retrieved_files);
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
347 }
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
348 } else {
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
349 # otherwise we assume that it works
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
350 $archcnt += run_retrieve($docdir, $docdir, \%retrieved_files);
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
351 }
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
352
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
353 # restore current index.meta
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
354 unpark_file($metafile);
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
355
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
356 if ($archcnt == 0) {
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
357 logger("ABORT", "no files retrieved!!");
30497c6a3eca Initial revision
casties
parents:
diff changeset
358 exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
359 }
10
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
360 logger("INFO", "$archcnt files retrieved");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
361
30497c6a3eca Initial revision
casties
parents:
diff changeset
362 # check list of archived files
10
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
363 check_files($archived_files, \%retrieved_files);
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
364
30497c6a3eca Initial revision
casties
parents:
diff changeset
365 # rewrite index.meta file
30497c6a3eca Initial revision
casties
parents:
diff changeset
366 write_xml($document, $metafile);
30497c6a3eca Initial revision
casties
parents:
diff changeset
367
30497c6a3eca Initial revision
casties
parents:
diff changeset
368 logger("INFO", "$warncnt warnings");
30497c6a3eca Initial revision
casties
parents:
diff changeset
369 logger("INFO", "$errcnt errors");
30497c6a3eca Initial revision
casties
parents:
diff changeset
370 if ($errcnt == 0) {
10
4417be0e2f07 adapted to dsmc's problem with mount points
casties
parents: 7
diff changeset
371 logger("DONE", "$archcnt archived files retrieved");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
372 } else {
30497c6a3eca Initial revision
casties
parents:
diff changeset
373 logger("ABORT", "there were $errcnt errors!!");
30497c6a3eca Initial revision
casties
parents:
diff changeset
374 exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
375 }