/[BackupPC]/trunk/bin/BackupPC_incPartsUpdate
This is repository of my old source code which isn't updated any more. Go to git.rot13.org for current projects!
ViewVC logotype

Annotation of /trunk/bin/BackupPC_incPartsUpdate

Parent Directory Parent Directory | Revision Log Revision Log


Revision 280 - (hide annotations)
Wed Dec 14 10:40:10 2005 UTC (18 years, 5 months ago) by dpavlin
File size: 8223 byte(s)
 r11683@llin:  dpavlin | 2005-12-14 13:40:02 +0100
 fix tars without files

1 dpavlin 157 #!/usr/local/bin/perl -w
2    
3     use strict;
4     use lib "__INSTALLDIR__/lib";
5    
6     use DBI;
7     use BackupPC::Lib;
8     use BackupPC::View;
9 dpavlin 214 use BackupPC::Attrib qw/:all/;
10 dpavlin 157 use Data::Dumper;
11     use Time::HiRes qw/time/;
12     use POSIX qw/strftime/;
13     use Cwd qw/abs_path/;
14 dpavlin 201 use File::Which;
15 dpavlin 215 use Archive::Tar::Streamed;
16 dpavlin 214 use Algorithm::Diff;
17     use Getopt::Std;
18 dpavlin 264 use File::Slurp;
19 dpavlin 157
20 dpavlin 230 my $bpc = BackupPC::Lib->new || die "can't create BackupPC::Lib";
21     my %Conf = $bpc->Conf();
22    
23     use BackupPC::SearchLib;
24     %BackupPC::SearchLib::Conf = %Conf;
25    
26 dpavlin 157 my $path = abs_path($0);
27     $path =~ s#/[^/]+$#/#;
28     my $tarIncCreate = $path .= 'BackupPC_tarIncCreate';
29    
30     die "can't find $tarIncCreate: $!\n" unless (-x $tarIncCreate);
31    
32 dpavlin 201 my $bin;
33 dpavlin 254 foreach my $c (qw/gzip md5sum/) {
34 dpavlin 201 $bin->{$c} = which($c) || die "$0 needs $c, install it\n";
35     }
36    
37 dpavlin 214 my %opt;
38     getopts("cd", \%opt );
39 dpavlin 201
40 dpavlin 214 my $debug = $opt{d};
41     my $check = $opt{c} && print STDERR "NOTICE: tar archive check forced\n";
42    
43 dpavlin 157 $|=1;
44    
45     my $start_t = time();
46    
47     my $t_fmt = '%Y-%m-%d %H:%M:%S';
48    
49     my $dsn = $Conf{SearchDSN} || die "Need SearchDSN in config.pl\n";
50     my $user = $Conf{SearchUser} || '';
51    
52     my $dbh = DBI->connect($dsn, $user, "", { RaiseError => 1, AutoCommit => 0 });
53    
54     my $tar_dir = $Conf{InstallDir}.'/'.$Conf{GzipTempDir};
55    
56     die "problem with $tar_dir, check GzipTempDir in configuration\n" unless (-d $tar_dir && -w $tar_dir);
57    
58     #---- subs ----
59    
60     sub fmt_time {
61     my $t = shift || return;
62     my $out = "";
63     my ($ss,$mm,$hh) = gmtime($t);
64     $out .= "${hh}h" if ($hh);
65     $out .= sprintf("%02d:%02d", $mm,$ss);
66     return $out;
67     }
68    
69     sub curr_time {
70     return strftime($t_fmt,localtime());
71     }
72    
73 dpavlin 264 my $hsn_cache;
74    
75     sub get_backup_id($$$) {
76     my ($host, $share, $num) = @_;
77    
78     my $key = "$host $share $num";
79     return $hsn_cache->{$key} if ($hsn_cache->{$key});
80    
81     my $sth = $dbh->prepare(qq{
82     SELECT
83     backups.id
84     FROM backups
85     INNER JOIN shares ON backups.shareID=shares.ID
86     INNER JOIN hosts ON backups.hostID = hosts.ID
87     where hosts.name = ? and shares.name = ? and backups.num = ?
88     });
89     $sth->execute($host, $share, $num);
90     my ($id) = $sth->fetchrow_array;
91    
92     $hsn_cache->{"$host $share $num"} = $id;
93    
94     print STDERR "# $host $share $num == $id\n" if ($opt{d});
95    
96     return $id;
97     }
98    
99    
100 dpavlin 214 sub tar_check($$$$) {
101     my ($host,$share,$num,$filename) = @_;
102    
103 dpavlin 270 my $t = time();
104     print curr_time, " check $host:$share#$num -> $filename";
105    
106 dpavlin 264 sub check_part {
107     my ($host, $share, $num, $part_nr, $tar_size, $size, $md5, $items) = @_;
108     my $backup_id = get_backup_id($host, $share, $num);
109     my $sth_md5 = $dbh->prepare(qq{
110     select
111     id, tar_size, size, md5, items
112     from backup_parts
113     where backup_id = ? and part_nr = ?
114     });
115    
116     $sth_md5->execute($backup_id, $part_nr);
117    
118     if (my $row = $sth_md5->fetchrow_hashref) {
119     return if (
120 dpavlin 265 $row->{tar_size} >= $tar_size &&
121 dpavlin 264 $row->{size} == $size &&
122     $row->{md5} eq $md5 &&
123     $row->{items} == $items
124     );
125 dpavlin 270 print ", deleting invalid backup_parts $row->{id}";
126 dpavlin 264 $dbh->do(qq{ delete from backup_parts where id = $row->{id} });
127     }
128 dpavlin 270 print ", inserting new";
129 dpavlin 264 my $sth_insert = $dbh->prepare(qq{
130     insert into backup_parts (
131     backup_id,
132     part_nr,
133     tar_size,
134     size,
135     md5,
136     items
137     ) values (?,?,?,?,?,?)
138     });
139    
140     $sth_insert->execute($backup_id, $part_nr, $tar_size, $size, $md5, $items);
141     $dbh->commit;
142     }
143    
144 dpavlin 254 my @tar_parts;
145    
146     if (-d "$tar_dir/$filename") {
147 dpavlin 270 print ", multi-part";
148 dpavlin 254 opendir(my $dir, "$tar_dir/$filename") || die "can't readdir $tar_dir/$filename: $!";
149     @tar_parts = map { my $p = $_; $p =~ s#^#${filename}/#; $p } grep { !/^\./ && !/md5/ && -f "$tar_dir/$filename/$_" } readdir($dir);
150     closedir($dir);
151     } else {
152     push @tar_parts, "${filename}.tar.gz";
153 dpavlin 214 }
154    
155 dpavlin 256 print " [parts: ",join(", ", @tar_parts),"]" if ($opt{d});
156    
157     my $same = 1;
158 dpavlin 254 my @tar_files;
159 dpavlin 214
160 dpavlin 264 my $backup_part;
161    
162 dpavlin 268 print " reading" if ($opt{d});
163 dpavlin 254
164     foreach my $tarfilename (@tar_parts) {
165    
166 dpavlin 270 print "\n\t- $tarfilename";
167 dpavlin 254
168 dpavlin 269 my $size = (stat( "$tar_dir/$tarfilename" ))[7] || die "can't stat $tar_dir/$tarfilename";
169    
170     if ($size > $Conf{MaxArchiveSize}) {
171 dpavlin 274 print ", part bigger than media $size > $Conf{MaxArchiveSize}\n";
172     return 0;
173 dpavlin 254 }
174    
175 dpavlin 270 print ", $size bytes";
176    
177 dpavlin 269 my $path = "$tar_dir/$tarfilename";
178 dpavlin 264
179 dpavlin 254 open(my $fh, "gzip -cd $tar_dir/$tarfilename |") or die "can't open $tar_dir/$tarfilename: $!";
180     binmode($fh);
181     my $tar = Archive::Tar::Streamed->new($fh);
182    
183 dpavlin 264 my $tar_size = 0;
184     my $items = 0;
185 dpavlin 256
186 dpavlin 254 while(my $entry = $tar->next) {
187     push @tar_files, $entry->name;
188 dpavlin 264 $items++;
189     $tar_size += $entry->size;
190 dpavlin 269
191     if ($tar_size > $Conf{MaxArchiveFileSize}) {
192 dpavlin 270 print ", part $tarfilename is too big $tar_size > $Conf{MaxArchiveFileSize}\n";
193 dpavlin 274 return 0;
194 dpavlin 269 }
195    
196 dpavlin 254 }
197 dpavlin 256
198 dpavlin 270 print ", $items items";
199    
200 dpavlin 280 if ($tar_size == 0 && $items == 0) {
201 dpavlin 279 print ", EMPTY tar\n";
202    
203     my $backup_id = get_backup_id($host, $share, $num);
204    
205     my $sth_inc_deleted = $dbh->prepare(qq{
206     update backups set
207     inc_deleted = true
208     where id = ?
209     });
210     $sth_inc_deleted->execute($backup_id);
211 dpavlin 280
212     $dbh->commit;
213    
214     return 1;
215 dpavlin 279 }
216    
217 dpavlin 280 # fix tar_size for tars without any files
218     $tar_size ||= 512 * $items;
219 dpavlin 279
220 dpavlin 280
221 dpavlin 269 #
222     # check if md5 exists, and if not, create one
223     #
224    
225     my $md5_path = $path;
226     $md5_path =~ s/\.tar\.gz$/.md5/ || die "can't create md5 filename from $md5_path";
227     if (! -e $md5_path || -z $md5_path) {
228     print ", creating md5";
229     system( $bin->{md5sum} . " $path > $md5_path") == 0 or die "can't create md5 $path: $!";
230 dpavlin 270 } else {
231     ## FIXME check if existing md5 is valid
232 dpavlin 256 }
233 dpavlin 264
234 dpavlin 269 my $md5 = read_file( $md5_path ) || die "can't read md5sum file $md5_path: $!";
235     $md5 =~ s#\s.*$##;
236    
237     # extract part number from filename
238     my $part_nr = 1;
239     $part_nr = $1 if ($tarfilename =~ m#/(\d+)\.tar\.gz#);
240    
241     #
242     # finally, check if backup_parts table in database is valid
243     #
244    
245 dpavlin 264 check_part($host, $share, $num, $part_nr, $tar_size, $size, $md5, $items);
246 dpavlin 215 }
247 dpavlin 254
248 dpavlin 256 # short-cut and exit;
249     return $same unless($same);
250    
251 dpavlin 215 @tar_files = sort @tar_files;
252 dpavlin 270 print "\n\t",($#tar_files + 1), " tar files";
253 dpavlin 214
254     my $sth = $dbh->prepare(qq{
255     SELECT path,type
256     FROM files
257     JOIN shares on shares.id = shareid
258     JOIN hosts on hosts.id = shares.hostid
259     WHERE hosts.name = ? and shares.name = ? and backupnum = ?
260     });
261     $sth->execute($host, $share, $num);
262     my @db_files;
263     while( my $row = $sth->fetchrow_hashref ) {
264    
265     my $path = $row->{'path'} || die "no path?";
266     $path =~ s#^/#./#;
267     $path .= '/' if ($row->{'type'} == BPC_FTYPE_DIR);
268     push @db_files, $path;
269     }
270    
271 dpavlin 270 print " ",($#db_files + 1), " database files, diff";
272 dpavlin 214
273     @db_files = sort @db_files;
274    
275     if ($#tar_files != $#db_files) {
276     $same = 0;
277 dpavlin 270 print " NUMBER";
278 dpavlin 214 } else {
279     my $diff = Algorithm::Diff->new(\@tar_files, \@db_files);
280     while ( $diff->Next() ) {
281     next if $diff->Same();
282     $same = 0;
283     print "< $_\n" for $diff->Items(1);
284     print "> $_\n" for $diff->Items(2);
285     }
286     }
287    
288 dpavlin 270 print " ",($same ? 'ok' : 'DIFFERENT'),
289     ", dur: ",fmt_time(time() - $t), "\n";
290 dpavlin 214
291     return $same;
292     }
293    
294    
295 dpavlin 157 #----- main
296    
297     my $sth = $dbh->prepare( qq{
298    
299     select
300 dpavlin 158 backups.id as backup_id,
301 dpavlin 157 hosts.name as host,
302     shares.name as share,
303 dpavlin 192 backups.num as num,
304 dpavlin 196 inc_size,
305     parts
306 dpavlin 157 from backups
307     join shares on backups.hostid = shares.hostid
308     and shares.id = backups.shareid
309     join hosts on shares.hostid = hosts.id
310 dpavlin 192 where not inc_deleted
311 dpavlin 157 order by backups.date
312    
313     } );
314    
315 dpavlin 213 $sth->execute();
316     my $num_backups = $sth->rows;
317     my $curr_backup = 1;
318    
319 dpavlin 157 while (my $row = $sth->fetchrow_hashref) {
320 dpavlin 271
321     $curr_backup++;
322    
323 dpavlin 157 my $tar_file = BackupPC::SearchLib::getGzipName($row->{'host'}, $row->{'share'}, $row->{'num'});
324 dpavlin 192
325 dpavlin 194 # this will return -1 if file doesn't exist
326     my $size = BackupPC::SearchLib::get_tgz_size_by_name($tar_file);
327 dpavlin 192
328 dpavlin 253 print "# size: $size backup.size: ", $row->{inc_size},"\n" if ($opt{d});
329    
330 dpavlin 266 if ( $row->{'inc_size'} != -1 && $size != -1 && $row->{'inc_size'} >= $size) {
331 dpavlin 256 if ($check) {
332     tar_check($row->{'host'}, $row->{'share'}, $row->{'num'}, $tar_file) && next;
333     } else {
334     next;
335     }
336 dpavlin 253 }
337    
338 dpavlin 271 print curr_time, " creating $curr_backup/$num_backups ", $row->{'host'}, ":", $row->{'share'}, " #", $row->{'num'}, " -> $tar_file";
339 dpavlin 157
340     my $t = time();
341    
342 dpavlin 194 # re-create archive?
343 dpavlin 254 my $cmd = qq{ $tarIncCreate -h "$row->{'host'}" -s "$row->{'share'}" -n $row->{'num'} -f };
344 dpavlin 253 print STDERR "## $cmd\n" if ($debug);
345 dpavlin 157
346 dpavlin 253 if (system($cmd) != 0) {
347     print STDERR " FAILED";
348 dpavlin 194 }
349 dpavlin 157
350 dpavlin 194 print ", dur: ",fmt_time(time() - $t), "\n";
351 dpavlin 158
352     $dbh->commit;
353    
354 dpavlin 157 }
355    
356     undef $sth;
357     $dbh->disconnect;

Properties

Name Value
svn:executable *

  ViewVC Help
Powered by ViewVC 1.1.26