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 File::Slurp;
  17 use Getopt::Long::Descriptive;
  18
  19 =head1 NAME
  20
  21 BackupPC_ASA_PostArchive_Update
  22
  23 =head1 DESCRIPTION
  24
  25         # /etc/BackupPC/pc/_search_archive.pl
  26
  27 =cut
  28
  29
  30 my $bpc = BackupPC::Lib->new || die "can't create BackupPC::Lib";
  31 $bpc->ConfigRead('_search_archive'); # read our configuration
  32 my %Conf = $bpc->Conf();
  33
  34 use BackupPC::Search;
  35 %BackupPC::Search::Conf = %Conf;
  36
  37 my ($opt,$usage) = describe_options(
  38 "%c %o",
  39 [ 'host|h=s@',  "import just host(s)" ],
  40 [ 'num|n=s@',   "import just backup number(s)" ],
  41 [ 'ok=n',       "xferOK", { default => 0 } ],
  42 [ 'check|c',    "check archives on disk and sync", { default => 1 } ],
  43 [ 'debug|d',    "debug", { default => 1 } ],
  44 [ 'help',       "show help" ],
  45 );
  46
  47 print($usage->text), exit if $opt->help;
  48
  49 $|=1;
  50
  51 my $start_t = time();
  52
  53 my $t_fmt = '%Y-%m-%d %H:%M:%S';
  54
  55 #warn "## Conf = ",dump( \%Conf );
  56
  57 my $dbh = DBI->connect($Conf{SearchDSN}, $Conf{SearchUser}, "", { RaiseError => 1, AutoCommit => 0 });
  58
  59 #---- subs ----
  60
  61
  62 sub curr_time {
  63         return strftime($t_fmt,localtime());
  64 }
  65
  66 sub fmt_time {
  67         my $t = shift || return;
  68         my $out = "";
  69         my ($ss,$mm,$hh) = gmtime($t);
  70         $out .= "${hh}h" if ($hh);
  71         $out .= sprintf("%02d:%02d", $mm,$ss);
  72         return $out;
  73 }
  74
  75 my $hsn_cache;
  76
  77 sub get_backup_id($$) {
  78         my ($host, $num) = @_;
  79
  80         my $key = "$host $num";
  81         return $hsn_cache->{$key} if ($hsn_cache->{$key});
  82
  83         my $sth = $dbh->prepare(qq{
  84                 SELECT
  85                         backups.id
  86                 FROM backups
  87                 INNER JOIN shares       ON backups.shareID=shares.ID
  88                 INNER JOIN hosts        ON backups.hostID = hosts.ID
  89                 WHERE hosts.name = ? and backups.num = ?
  90         });
  91         $sth->execute($host, $num);
  92         my ($id) = $sth->fetchrow_array;
  93
  94         $hsn_cache->{"$host $num"} = $id;
  95
  96         print STDERR "# $host $num == $id\n" if $opt->debug;
  97
  98         return $id;
  99 }
 100
 101 sub backup_inc_deleted($) {
 102         my $backup_id = shift;
 103         my $sth_inc_deleted = $dbh->prepare(qq{
 104                 update backups set
 105                         inc_deleted = true
 106                 where id = ?
 107         });
 108         $sth_inc_deleted->execute($backup_id);
 109 }
 110
 111 sub system_ok {
 112         warn "## system_ok @_\n";
 113         system(@_) == 0 || die "system @_:$!";
 114 }
 115
 116 my $sth_inc_size = $dbh->prepare(qq{
 117         update backups set
 118                 inc_size = ?,
 119                 parts = ?,
 120                 inc_deleted = false
 121         where id = ?
 122 });
 123
 124 sub read_only {
 125         my $full = shift;
 126         my $perm = (stat $full)[2] & 0444;
 127         warn sprintf("chmod %03o %s\n",$perm,$full);
 128         chmod $perm, $full || die $!;
 129 }
 130
 131 sub check_archive {
 132         my ($host,$num) = @_;
 133         warn "# check_archive $host $num";
 134
 135         my $t = time();
 136
 137         my $glob = "$Conf{ArchiveDest}/$host.$num.*";
 138
 139         my @tar_parts = sort map { s/^\Q$Conf{ArchiveDest}\E\/*//; $_ } glob $glob ;
 140
 141         if ( ! @tar_parts ) {
 142                 warn "ERROR: no files for $glob";
 143                 return;
 144         }
 145
 146         print curr_time, " check $host $num";
 147
 148         my $md5_path = "$Conf{ArchiveDest}/$host.$num.md5";
 149         unlink $md5_path if -s $md5_path == 0; # fix empty
 150
 151         my $read_protect = 0;
 152
 153         if ( ! -e $md5_path ) {
 154                 system_ok "cd $Conf{ArchiveDest} && /usr/bin/md5sum $host.$num.* > $md5_path";
 155                 read_only $md5_path;
 156                 $read_protect = 1;
 157         } else {
 158                 system_ok "cd $Conf{ArchiveDest} && /usr/bin/md5sum -c $md5_path" if $opt->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} = $md5;
 165                 read_only "$Conf{ArchiveDest}/$path" if $read_protect;
 166         }
 167
 168         # depending on expected returned value this is used like:
 169         # my $uncompress_size = get_gzip_size('/full/path/to.gz');
 170         # my ($compress_size, $uncompress_size) = get_gzip_size('/path.gz');
 171         sub get_gzip_size($) {
 172                 my $filename = shift;
 173                 die "file $filename problem: $!" unless (-r $filename);
 174
 175                 if ( $filename !~ m/\.gz$/ ) {
 176                         return -s $filename;
 177                 }
 178
 179                 open(my $gzip, $Conf{GzipPath}." -l $filename |") || die "can't gzip -l $filename: $!";
 180                 local $/ = undef;
 181                 my $line = <$gzip>;
 182                 close($gzip);
 183
 184                 my ($comp, $uncomp) = (0,0);
 185
 186                 if ($line =~ m/\s+(\d+)\s+(\d+)\s+\d+\.\d+/s) {
 187                         if (wantarray) {
 188                                 return [ $1, $2 ];
 189                         } else {
 190                                 return $2;
 191                         }
 192                 } else {
 193                         warn "ERROR can't parse: $line";
 194                         return -s $filename;
 195                 }
 196         }
 197
 198         sub check_part {
 199                 my ($host, $num, $part_nr, $tar_size, $size, $md5, $items, $filename) = @_;
 200                 my $backup_id = get_backup_id($host, $num);
 201                 my $sth_md5 = $dbh->prepare(qq{
 202                         select
 203                                 id, tar_size, size, md5, items, filename
 204                         from backup_parts
 205                         where backup_id = ? and part_nr = ? and filename = ?
 206                 });
 207
 208                 $sth_md5->execute($backup_id, $part_nr, $filename);
 209
 210                 if (my $row = $sth_md5->fetchrow_hashref) {
 211                         return if (
 212                                 $row->{tar_size} >= $tar_size &&
 213                                 $row->{size} == $size &&
 214                                 $row->{md5} eq $md5 &&
 215                                 $row->{items} == $items
 216                         );
 217                         print ", deleting invalid backup_parts $row->{id}";
 218                         $dbh->do(qq{ delete from backup_parts where id = $row->{id} });
 219                 }
 220                 print ", inserting new";
 221                 my $sth_insert = $dbh->prepare(qq{
 222                         insert into backup_parts (
 223                                 backup_id,
 224                                 part_nr,
 225                                 tar_size,
 226                                 size,
 227                                 md5,
 228                                 items,
 229                                 filename
 230                         ) values (?,?,?,?,?,?,?)
 231                 });
 232
 233                 $sth_insert->execute($backup_id, $part_nr, $tar_size, $size, $md5, $items, $filename);
 234         }
 235
 236         print " [parts: ",join(", ", @tar_parts),"]" if $opt->debug;
 237
 238         my @tar_files;
 239
 240         my $backup_part;
 241
 242         print " reading" if $opt->debug;
 243
 244         my $part_nr = 0;
 245         my $inc_size = 0;
 246
 247         foreach my $filename (@tar_parts) {
 248
 249                 next if $filename eq "$host.$num.md5";
 250
 251                 print "\n\t- $filename";
 252
 253                 my $path = "$Conf{ArchiveDest}/$filename";
 254                 $path =~ s{//+}{/}g;
 255
 256                 my $size = (stat( $path ))[7] || die "can't stat $path: $!";
 257
 258                 if ($size > $Conf{ArchiveMediaSize}) {
 259                         print ", part bigger than media $size > $Conf{ArchiveMediaSize}\n";
 260                         return 0;
 261                 }
 262
 263                 print ", $size bytes";
 264
 265 =for later
 266
 267                 open(my $fh, "gzip -cd $path |") or die "can't open $path: $!";
 268                 binmode($fh);
 269                 my $tar = Archive::Tar::Streamed->new($fh);
 270
 271                 my $tar_size_inarc = 0;
 272                 my $items = 0;
 273
 274                 while(my $entry = $tar->next) {
 275                         push @tar_files, $entry->name;
 276                         $items++;
 277                         $tar_size_inarc += $entry->size;
 278
 279                         if ($tar_size_inarc > $Conf{ArchiveMediaSize}) {
 280                                 print ", part $filename is too big $tar_size_inarc > $Conf{ArchiveMediaSize}\n";
 281                                 return 0;
 282                         }
 283
 284                 }
 285
 286                 close($fh);
 287
 288                 print ", $items items";
 289
 290                 if ($tar_size_inarc == 0 && $items == 0) {
 291                         print ", EMPTY tar\n";
 292
 293                         my $backup_id = get_backup_id($host, $share, $num);
 294                         backup_inc_deleted( $backup_id );
 295
 296                         $dbh->commit;
 297
 298                         return 1;
 299                 }
 300
 301 =cut
 302
 303                 # FIXME
 304                 my $tar_size = get_gzip_size( $path );
 305
 306                 #
 307                 # finally, check if backup_parts table in database is valid
 308                 #
 309
 310                 my $md5 = $md5sum->{$filename} || die "no md5sum for $filename in ",dump($md5sum);
 311                 my $items = 1;
 312                 $part_nr++;
 313
 314                 check_part($host, $num, $part_nr, $tar_size, $size, $md5, $items, $filename);
 315
 316                 # round increment size to 2k block size
 317                 $inc_size += int((($size + 2048) / 2048 ) * 2048);
 318         }
 319
 320         $sth_inc_size->execute(
 321                 $inc_size,
 322                 $part_nr,
 323                 get_backup_id($host, $num),
 324         );
 325
 326         @tar_files = sort @tar_files;
 327         print "\n\t",($#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         $dbh->commit;
 369
 370         return $same;
 371 }
 372
 373
 374 #----- main
 375
 376 foreach ( 0 .. $#{ $opt->host } ) {
 377
 378         my $host = lc $opt->host->[$_];
 379         my $num  = $opt->num->[$_];
 380
 381         if ( ! $opt->ok ) {
 382                 warn "ERROR $host $num running cleanup";
 383                 foreach my $path ( glob "$Conf{ArchiveDest}/$host.$num.*" ) {
 384                         warn "# rm $path";
 385                         unlink $path || die $!;
 386                 }
 387         } else {
 388                 check_archive $host => $num;
 389         }
 390
 391 }
 392
 393 exit; # FIXME
 394
 395 my $sth = $dbh->prepare( qq{
 396
 397 select
 398         backups.id as backup_id,
 399         hosts.name as host,
 400         shares.name as share,
 401         backups.num as num,
 402         backups.date,
 403         inc_size,
 404         parts,
 405         count(backup_parts.backup_id) as backup_parts
 406 from backups
 407         join shares on backups.hostid = shares.hostid
 408                 and shares.id = backups.shareid
 409         join hosts on shares.hostid = hosts.id
 410         full outer join backup_parts on backups.id = backup_parts.backup_id
 411 where not inc_deleted and backups.size > 0
 412 group by backups.id, hosts.name, shares.name, backups.num, backups.date, inc_size, parts, backup_parts.backup_id
 413 order by backups.date
 414
 415 } );
 416
 417 $sth->execute();
 418 my $num_backups = $sth->rows;
 419 my $curr_backup = 1;
 420
 421 while (my $row = $sth->fetchrow_hashref) {
 422
 423         $curr_backup++;
 424
 425         my $tar_file = BackupPC::Search::getGzipName($row->{'host'}, $row->{'share'}, $row->{'num'});
 426
 427         # this will return -1 if file doesn't exist
 428         my $size = BackupPC::Search::get_tgz_size_by_name($tar_file);
 429
 430         print "# host: ".$row->{host}.", share: ".$row->{'share'}.", backup_num:".$row->{num}." size: $size backup.size: ", $row->{inc_size},"\n" if $opt->debug;
 431
 432         if ( $row->{'inc_size'} != -1 && $size != -1 && $row->{'inc_size'} >= $size && $row->{parts} == $row->{backup_parts}) {
 433                 if ($opt->check) {
 434                         tar_check($row->{'host'}, $row->{'share'}, $row->{'num'}, $tar_file) && next;
 435                 } else {
 436                         next;
 437                 }
 438         }
 439
 440         print curr_time, " creating $curr_backup/$num_backups ", $row->{host}, ":", $row->{share}, " #", $row->{num},
 441                 " ", strftime('%Y-%m-%d', localtime($row->{date})), " -> $tar_file";
 442
 443         my $t = time();
 444
 445 =for later
 446         # re-create archive?
 447         my $cmd = qq[ $tarIncCreate -h "$row->{host}" -s "$row->{share}" -n $row->{num} -f ];
 448         print STDERR "## $cmd\n" if ($opt->debug);
 449
 450         if (system($cmd) != 0) {
 451                 print STDERR " FAILED, marking this backup deleted";
 452                 backup_inc_deleted( $row->{backup_id} );
 453         }
 454 =cut
 455
 456         print ", dur: ",fmt_time(time() - $t), "\n";
 457
 458 }
 459
 460 undef $sth;
 461 $dbh->disconnect;