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                 my $line = <$gzip>;
 180                 chomp($line);
 181                 $line = <$gzip> if ($line =~ /^\s+compressed/);
 182
 183                 my ($comp, $uncomp) = (0,0);
 184
 185                 if ($line =~ m/^\s+(\d+)\s+(\d+)\s+\d+\.\d+/) {
 186                         if (wantarray) {
 187                                 return [ $1, $2 ];
 188                         } else {
 189                                 return $2;
 190                         }
 191                 } else {
 192                         die "can't find size in line: $line";
 193                 }
 194         }
 195
 196         sub check_part {
 197                 my ($host, $num, $part_nr, $tar_size, $size, $md5, $items) = @_;
 198                 my $backup_id = get_backup_id($host, $num);
 199                 my $sth_md5 = $dbh->prepare(qq{
 200                         select
 201                                 id, tar_size, size, md5, items
 202                         from backup_parts
 203                         where backup_id = ? and part_nr = ?
 204                 });
 205
 206                 $sth_md5->execute($backup_id, $part_nr);
 207
 208                 if (my $row = $sth_md5->fetchrow_hashref) {
 209                         return if (
 210                                 $row->{tar_size} >= $tar_size &&
 211                                 $row->{size} == $size &&
 212                                 $row->{md5} eq $md5 &&
 213                                 $row->{items} == $items
 214                         );
 215                         print ", deleting invalid backup_parts $row->{id}";
 216                         $dbh->do(qq{ delete from backup_parts where id = $row->{id} });
 217                 }
 218                 print ", inserting new";
 219                 my $sth_insert = $dbh->prepare(qq{
 220                         insert into backup_parts (
 221                                 backup_id,
 222                                 part_nr,
 223                                 tar_size,
 224                                 size,
 225                                 md5,
 226                                 items
 227                         ) values (?,?,?,?,?,?)
 228                 });
 229
 230                 $sth_insert->execute($backup_id, $part_nr, $tar_size, $size, $md5, $items);
 231                 $dbh->commit;
 232         }
 233
 234         print " [parts: ",join(", ", @tar_parts),"]" if $debug;
 235
 236         my @tar_files;
 237
 238         my $backup_part;
 239
 240         print " reading" if $debug;
 241
 242         my $part_nr = 0;
 243         my $inc_size = 0;
 244
 245         foreach my $filename (@tar_parts) {
 246
 247                 next if $filename eq "$host.$num.md5";
 248
 249                 print "\n\t- $filename";
 250
 251                 my $path = "$Conf{ArchiveDest}/$filename";
 252                 $path =~ s{//+}{/}g;
 253
 254                 my $size = (stat( $path ))[7] || die "can't stat $path: $!";
 255
 256                 if ($size > $Conf{MaxArchiveSize}) {
 257                         print ", part bigger than media $size > $Conf{MaxArchiveSize}\n";
 258                         return 0;
 259                 }
 260
 261                 print ", $size bytes";
 262
 263 =for later
 264
 265                 open(my $fh, "gzip -cd $path |") or die "can't open $path: $!";
 266                 binmode($fh);
 267                 my $tar = Archive::Tar::Streamed->new($fh);
 268
 269                 my $tar_size_inarc = 0;
 270                 my $items = 0;
 271
 272                 while(my $entry = $tar->next) {
 273                         push @tar_files, $entry->name;
 274                         $items++;
 275                         $tar_size_inarc += $entry->size;
 276
 277                         if ($tar_size_inarc > $Conf{MaxArchiveFileSize}) {
 278                                 print ", part $filename is too big $tar_size_inarc > $Conf{MaxArchiveFileSize}\n";
 279                                 return 0;
 280                         }
 281
 282                 }
 283
 284                 close($fh);
 285
 286                 print ", $items items";
 287
 288                 if ($tar_size_inarc == 0 && $items == 0) {
 289                         print ", EMPTY tar\n";
 290
 291                         my $backup_id = get_backup_id($host, $share, $num);
 292                         backup_inc_deleted( $backup_id );
 293
 294                         $dbh->commit;
 295
 296                         return 1;
 297                 }
 298
 299 =cut
 300
 301                 # FIXME
 302                 my $tar_size = get_gzip_size( $path );
 303
 304                 #
 305                 # finally, check if backup_parts table in database is valid
 306                 #
 307
 308                 my $md5 = $md5sum->{$filename} || die "no md5sum for $filename in ",dump($md5sum);
 309                 my $items = 1;
 310                 $part_nr++;
 311
 312                 check_part($host, $num, $part_nr, $tar_size, $size, $md5, $items);
 313
 314                 # round increment size to 2k block size
 315                 $inc_size += int(($size + 2048) / 2048);
 316         }
 317
 318         $sth_inc_size->execute(
 319                 $inc_size,
 320                 $part_nr,
 321                 get_backup_id($host, $num),
 322         );
 323         $dbh->commit;
 324
 325         @tar_files = sort @tar_files;
 326         print "\n\t",($#tar_files + 1), " tar files";
 327
 328         my $sth = $dbh->prepare(qq{
 329                 SELECT path,type
 330                 FROM files
 331                 JOIN shares on shares.id = shareid
 332                 JOIN hosts on hosts.id = shares.hostid
 333                 WHERE hosts.name = ? and backupnum = ?
 334         });
 335         $sth->execute($host, $num);
 336         my @db_files;
 337         while( my $row = $sth->fetchrow_hashref ) {
 338
 339                 my $path = $row->{'path'} || die "no path?";
 340                 $path =~ s#^/#./#;
 341                 $path .= '/' if ($row->{'type'} == BPC_FTYPE_DIR);
 342                 push @db_files, $path;
 343         }
 344
 345         print " ",($#db_files + 1), " database files, diff";
 346
 347         @db_files = sort @db_files;
 348
 349         my $same = 1;
 350
 351         if ($#tar_files != $#db_files) {
 352                 $same = 0;
 353                 print " NUMBER";
 354         } else {
 355                 my $diff = Algorithm::Diff->new(\@tar_files, \@db_files);
 356                 while ( $diff->Next() ) {
 357                         next if $diff->Same();
 358                         $same = 0;
 359                         print "< $_\n" for $diff->Items(1);
 360                         print "> $_\n" for $diff->Items(2);
 361                 }
 362         }
 363
 364         print " ",($same ? 'ok' : 'DIFFERENT'),
 365                 ", dur: ",fmt_time(time() - $t), "\n";
 366
 367         return $same;
 368 }
 369
 370
 371 #----- main
 372
 373 foreach ( 0 .. $#{ $args->{HostList} } ) {
 374
 375         my $host = $args->{'HostList'}->[$_];
 376         my $num  = $args->{'BackupList'}->[$_];
 377
 378         check_archive $host => $num;
 379
 380 }
 381
 382 exit;
 383
 384 my $sth = $dbh->prepare( qq{
 385
 386 select
 387         backups.id as backup_id,
 388         hosts.name as host,
 389         shares.name as share,
 390         backups.num as num,
 391         backups.date,
 392         inc_size,
 393         parts,
 394         count(backup_parts.backup_id) as backup_parts
 395 from backups
 396         join shares on backups.hostid = shares.hostid
 397                 and shares.id = backups.shareid
 398         join hosts on shares.hostid = hosts.id
 399         full outer join backup_parts on backups.id = backup_parts.backup_id
 400 where not inc_deleted and backups.size > 0
 401 group by backups.id, hosts.name, shares.name, backups.num, backups.date, inc_size, parts, backup_parts.backup_id
 402 order by backups.date
 403
 404 } );
 405
 406 $sth->execute();
 407 my $num_backups = $sth->rows;
 408 my $curr_backup = 1;
 409
 410 while (my $row = $sth->fetchrow_hashref) {
 411
 412         $curr_backup++;
 413
 414         my $tar_file = BackupPC::Search::getGzipName($row->{'host'}, $row->{'share'}, $row->{'num'});
 415
 416         # this will return -1 if file doesn't exist
 417         my $size = BackupPC::Search::get_tgz_size_by_name($tar_file);
 418
 419         print "# host: ".$row->{host}.", share: ".$row->{'share'}.", backup_num:".$row->{num}." size: $size backup.size: ", $row->{inc_size},"\n" if $debug;
 420
 421         if ( $row->{'inc_size'} != -1 && $size != -1 && $row->{'inc_size'} >= $size && $row->{parts} == $row->{backup_parts}) {
 422                 if ($check) {
 423                         tar_check($row->{'host'}, $row->{'share'}, $row->{'num'}, $tar_file) && next;
 424                 } else {
 425                         next;
 426                 }
 427         }
 428
 429         print curr_time, " creating $curr_backup/$num_backups ", $row->{host}, ":", $row->{share}, " #", $row->{num},
 430                 " ", strftime('%Y-%m-%d', localtime($row->{date})), " -> $tar_file";
 431
 432         my $t = time();
 433
 434 =for later
 435         # re-create archive?
 436         my $cmd = qq[ $tarIncCreate -h "$row->{host}" -s "$row->{share}" -n $row->{num} -f ];
 437         print STDERR "## $cmd\n" if ($debug);
 438
 439         if (system($cmd) != 0) {
 440                 print STDERR " FAILED, marking this backup deleted";
 441                 backup_inc_deleted( $row->{backup_id} );
 442         }
 443 =cut
 444
 445         print ", dur: ",fmt_time(time() - $t), "\n";
 446
 447         $dbh->commit;
 448
 449 }
 450
 451 undef $sth;
 452 $dbh->disconnect;