fix
[BackupPC.git] / bin / BackupPC_ASA_PostArchive_Update
1 #!/usr/local/bin/perl -w
2
3 use strict;
4 use lib "/usr/local/BackupPC/lib";
5
6 use DBI;
7 use BackupPC::Lib;
8 use BackupPC::View;
9 use BackupPC::Attrib qw/:all/;
10 use Data::Dump qw(dump);
11 use Time::HiRes qw/time/;
12 use POSIX qw/strftime/;
13 use Cwd qw/abs_path/;
14 use Archive::Tar::Streamed;
15 use Algorithm::Diff;
16 use Getopt::Std;
17 use File::Slurp;
18
19 =head1 NAME
20
21 BackupPC_ASA_PostArchive_Update
22
23 =head1 DESCRIPTION
24
25         # /etc/BackupPC/pc/dvd_tar.pl
26
27 =cut
28
29 # FIXME
30 my $debug = $ENV{DEBUG} || 1;
31 my $check = $ENV{CHECK} || 1;
32
33
34 my $bpc = BackupPC::Lib->new || die "can't create BackupPC::Lib";
35 my %Conf = $bpc->Conf();
36 warn "## ARGV=",dump @ARGV;
37
38
39 my $args;
40 my $name;
41 foreach ( @ARGV ) {
42         my $v = $_;
43         if ( m/(\w+)=(.+)/ ) {
44                 $name = $1;
45                 $v = $2;
46         }
47         if ( $name =~ m/List/ ) {
48                 push @{ $args->{$name} }, $v;
49         } else {
50                 $args->{$name} = $v;
51         }
52 }
53
54 warn "args = ",dump($args);
55
56
57 use BackupPC::Search;
58 %BackupPC::Search::Conf = %Conf;
59
60 my $path = abs_path($0);
61 $path =~ s{/[^/]+$}{/}; # FIXME remove?
62
63 $|=1;
64
65 my $start_t = time();
66
67 my $t_fmt = '%Y-%m-%d %H:%M:%S';
68
69 #warn "## Conf = ",dump( \%Conf );
70
71 my $dbh = DBI->connect($Conf{SearchDSN}, $Conf{SearchUser}, "", { RaiseError => 1, AutoCommit => 0 });
72
73 #---- subs ----
74
75
76 sub curr_time {
77         return strftime($t_fmt,localtime());
78 }
79
80 sub fmt_time {
81         my $t = shift || return;
82         my $out = "";
83         my ($ss,$mm,$hh) = gmtime($t);
84         $out .= "${hh}h" if ($hh);
85         $out .= sprintf("%02d:%02d", $mm,$ss);
86         return $out;
87 }
88
89 my $hsn_cache;
90
91 sub get_backup_id($$) {
92         my ($host, $num) = @_;
93
94         my $key = "$host $num";
95         return $hsn_cache->{$key} if ($hsn_cache->{$key});
96
97         my $sth = $dbh->prepare(qq{
98                 SELECT 
99                         backups.id
100                 FROM backups 
101                 INNER JOIN shares       ON backups.shareID=shares.ID
102                 INNER JOIN hosts        ON backups.hostID = hosts.ID
103                 WHERE hosts.name = ? and backups.num = ?
104         });
105         $sth->execute($host, $num);
106         my ($id) = $sth->fetchrow_array;
107
108         $hsn_cache->{"$host $num"} = $id;
109
110         print STDERR "# $host $num == $id\n" if $debug;
111
112         return $id;
113 }
114
115 sub backup_inc_deleted($) {
116         my $backup_id = shift;
117         my $sth_inc_deleted = $dbh->prepare(qq{
118                 update backups set
119                         inc_deleted = true
120                 where id = ?
121         });
122         $sth_inc_deleted->execute($backup_id);
123 }
124
125 sub system_ok {
126         warn "## system_ok @_\n";
127         system(@_) == 0 || die "system @_:$!";
128 }
129
130 my $sth_inc_size = $dbh->prepare(qq{
131         update backups set
132                 inc_size = ?,
133                 parts = ?,
134                 inc_deleted = false
135         where id = ?
136 });
137
138 sub check_archive {
139         my ($host,$num) = @_;
140
141         my $t = time();
142
143         my @tar_parts =
144                 sort map { s/^\Q$Conf{ArchiveDest}\E\/*//; $_ }
145                 glob "$Conf{ArchiveDest}/$host.$num.*"
146                 ;
147
148         return unless @tar_parts;
149
150         print curr_time, " check $host $num";
151
152         my $md5_path = "$Conf{ArchiveDest}/$host.$num.md5";
153         unlink $md5_path if -s $md5_path == 0; # fix empty
154
155         if ( ! -e $md5_path ) {
156                 system_ok "cd $Conf{ArchiveDest} && /usr/bin/md5sum $host.$num.* > $md5_path";
157         } else {
158                 system_ok "cd $Conf{ArchiveDest} && /usr/bin/md5sum -c $md5_path" if $check;
159         }
160
161         my $md5sum;
162         foreach ( split(/\n/, read_file "$Conf{ArchiveDest}/$host.$num.md5" ) ) {
163                 my ( $md5, $path ) = split(/\s+/,$_);
164                 $md5sum->{$path} = $md5;
165         }
166
167         # depending on expected returned value this is used like:
168         # my $uncompress_size = get_gzip_size('/full/path/to.gz');
169         # my ($compress_size, $uncompress_size) = get_gzip_size('/path.gz');
170         sub get_gzip_size($) {
171                 my $filename = shift;
172                 die "file $filename problem: $!" unless (-r $filename);
173
174                 if ( $filename !~ m/\.gz$/ ) {
175                         return -s $filename;
176                 }
177
178                 open(my $gzip, $Conf{GzipPath}." -l $filename |") || die "can't gzip -l $filename: $!";
179                 local $/ = undef;
180                 my $line = <$gzip>;
181                 close($gzip);
182
183                 my ($comp, $uncomp) = (0,0);
184
185                 if ($line =~ m/\s+(\d+)\s+(\d+)\s+\d+\.\d+/s) {
186                         if (wantarray) {
187                                 return [ $1, $2 ];
188                         } else {
189                                 return $2;
190                         }
191                 } else {
192                         warn "ERROR can't parse: $line";
193                         return -s $filename;
194                 }
195         }
196
197         sub check_part {
198                 my ($host, $num, $part_nr, $tar_size, $size, $md5, $items) = @_;
199                 my $backup_id = get_backup_id($host, $num);
200                 my $sth_md5 = $dbh->prepare(qq{
201                         select
202                                 id, tar_size, size, md5, items
203                         from backup_parts
204                         where backup_id = ? and part_nr = ?
205                 });
206
207                 $sth_md5->execute($backup_id, $part_nr);
208
209                 if (my $row = $sth_md5->fetchrow_hashref) {
210                         return if (
211                                 $row->{tar_size} >= $tar_size &&
212                                 $row->{size} == $size &&
213                                 $row->{md5} eq $md5 &&
214                                 $row->{items} == $items
215                         );
216                         print ", deleting invalid backup_parts $row->{id}";
217                         $dbh->do(qq{ delete from backup_parts where id = $row->{id} });
218                 }
219                 print ", inserting new";
220                 my $sth_insert = $dbh->prepare(qq{
221                         insert into backup_parts (
222                                 backup_id,
223                                 part_nr,
224                                 tar_size,
225                                 size,
226                                 md5,
227                                 items
228                         ) values (?,?,?,?,?,?)
229                 });
230
231                 $sth_insert->execute($backup_id, $part_nr, $tar_size, $size, $md5, $items);
232                 $dbh->commit;
233         }
234
235         print " [parts: ",join(", ", @tar_parts),"]" if $debug;
236
237         my @tar_files;
238
239         my $backup_part;
240
241         print " reading" if $debug;
242
243         my $part_nr = 0;
244         my $inc_size = 0;
245
246         foreach my $filename (@tar_parts) {
247
248                 next if $filename eq "$host.$num.md5";
249
250                 print "\n\t- $filename";
251
252                 my $path = "$Conf{ArchiveDest}/$filename";
253                 $path =~ s{//+}{/}g;
254
255                 my $size = (stat( $path ))[7] || die "can't stat $path: $!";
256
257                 if ($size > $Conf{MaxArchiveSize}) {
258                         print ", part bigger than media $size > $Conf{MaxArchiveSize}\n";
259                         return 0;
260                 }
261
262                 print ", $size bytes";
263
264 =for later
265
266                 open(my $fh, "gzip -cd $path |") or die "can't open $path: $!";
267                 binmode($fh);
268                 my $tar = Archive::Tar::Streamed->new($fh);
269
270                 my $tar_size_inarc = 0;
271                 my $items = 0;
272
273                 while(my $entry = $tar->next) {
274                         push @tar_files, $entry->name;
275                         $items++;
276                         $tar_size_inarc += $entry->size;
277
278                         if ($tar_size_inarc > $Conf{MaxArchiveFileSize}) {
279                                 print ", part $filename is too big $tar_size_inarc > $Conf{MaxArchiveFileSize}\n";
280                                 return 0;
281                         }
282
283                 }
284
285                 close($fh);
286
287                 print ", $items items";
288
289                 if ($tar_size_inarc == 0 && $items == 0) {
290                         print ", EMPTY tar\n";
291
292                         my $backup_id = get_backup_id($host, $share, $num);
293                         backup_inc_deleted( $backup_id );
294
295                         $dbh->commit;
296
297                         return 1;
298                 }
299
300 =cut
301
302                 # FIXME
303                 my $tar_size = get_gzip_size( $path );
304
305                 #
306                 # finally, check if backup_parts table in database is valid
307                 #
308
309                 my $md5 = $md5sum->{$filename} || die "no md5sum for $filename in ",dump($md5sum);
310                 my $items = 1;
311                 $part_nr++;
312
313                 check_part($host, $num, $part_nr, $tar_size, $size, $md5, $items);
314
315                 # round increment size to 2k block size
316                 $inc_size += int(($size + 2048) / 2048);
317         }
318
319         $sth_inc_size->execute(
320                 $inc_size,
321                 $part_nr,
322                 get_backup_id($host, $num),
323         );
324         $dbh->commit;
325
326         @tar_files = sort @tar_files;
327         print "\n\t",($#tar_files + 1), " tar files";
328
329         my $sth = $dbh->prepare(qq{
330                 SELECT path,type
331                 FROM files
332                 JOIN shares on shares.id = shareid
333                 JOIN hosts on hosts.id = shares.hostid
334                 WHERE hosts.name = ? and backupnum = ?
335         });
336         $sth->execute($host, $num);
337         my @db_files;
338         while( my $row = $sth->fetchrow_hashref ) {
339
340                 my $path = $row->{'path'} || die "no path?";
341                 $path =~ s#^/#./#;
342                 $path .= '/' if ($row->{'type'} == BPC_FTYPE_DIR);
343                 push @db_files, $path;
344         }
345
346         print " ",($#db_files + 1), " database files, diff";
347
348         @db_files = sort @db_files;
349
350         my $same = 1;
351
352         if ($#tar_files != $#db_files) {
353                 $same = 0;
354                 print " NUMBER";
355         } else {
356                 my $diff = Algorithm::Diff->new(\@tar_files, \@db_files);
357                 while ( $diff->Next() ) {
358                         next if $diff->Same();
359                         $same = 0;
360                         print "< $_\n" for $diff->Items(1);
361                         print "> $_\n" for $diff->Items(2);
362                 }
363         }
364
365         print " ",($same ? 'ok' : 'DIFFERENT'),
366                 ", dur: ",fmt_time(time() - $t), "\n";
367
368         return $same;
369 }
370
371
372 #----- main
373
374 foreach ( 0 .. $#{ $args->{HostList} } ) {
375
376         my $host = $args->{'HostList'}->[$_];
377         my $num  = $args->{'BackupList'}->[$_];
378
379         check_archive $host => $num;
380
381 }
382
383 exit;
384
385 my $sth = $dbh->prepare( qq{
386         
387 select
388         backups.id as backup_id,
389         hosts.name as host,
390         shares.name as share,
391         backups.num as num,
392         backups.date,
393         inc_size,
394         parts,
395         count(backup_parts.backup_id) as backup_parts
396 from backups
397         join shares on backups.hostid = shares.hostid
398                 and shares.id = backups.shareid
399         join hosts on shares.hostid = hosts.id
400         full outer join backup_parts on backups.id = backup_parts.backup_id
401 where not inc_deleted and backups.size > 0
402 group by backups.id, hosts.name, shares.name, backups.num, backups.date, inc_size, parts, backup_parts.backup_id
403 order by backups.date
404
405 } );
406
407 $sth->execute();
408 my $num_backups = $sth->rows;
409 my $curr_backup = 1;
410
411 while (my $row = $sth->fetchrow_hashref) {
412
413         $curr_backup++;
414
415         my $tar_file = BackupPC::Search::getGzipName($row->{'host'}, $row->{'share'}, $row->{'num'});
416
417         # this will return -1 if file doesn't exist
418         my $size = BackupPC::Search::get_tgz_size_by_name($tar_file);
419
420         print "# host: ".$row->{host}.", share: ".$row->{'share'}.", backup_num:".$row->{num}." size: $size backup.size: ", $row->{inc_size},"\n" if $debug;
421
422         if ( $row->{'inc_size'} != -1 && $size != -1 && $row->{'inc_size'} >= $size && $row->{parts} == $row->{backup_parts}) {
423                 if ($check) {
424                         tar_check($row->{'host'}, $row->{'share'}, $row->{'num'}, $tar_file) && next;
425                 } else {
426                         next;
427                 }
428         }
429
430         print curr_time, " creating $curr_backup/$num_backups ", $row->{host}, ":", $row->{share}, " #", $row->{num},
431                 " ", strftime('%Y-%m-%d', localtime($row->{date})), " -> $tar_file";
432
433         my $t = time();
434
435 =for later
436         # re-create archive?
437         my $cmd = qq[ $tarIncCreate -h "$row->{host}" -s "$row->{share}" -n $row->{num} -f ];
438         print STDERR "## $cmd\n" if ($debug);
439
440         if (system($cmd) != 0) {
441                 print STDERR " FAILED, marking this backup deleted";
442                 backup_inc_deleted( $row->{backup_id} );
443         }
444 =cut
445
446         print ", dur: ",fmt_time(time() - $t), "\n";
447
448         $dbh->commit;
449
450 }
451
452 undef $sth;
453 $dbh->disconnect;