1 #!/usr/local/bin/perl -w
4 use lib "/usr/local/BackupPC/lib";
9 use BackupPC::Attrib qw/:all/;
10 use Data::Dump qw(dump);
11 use Time::HiRes qw/time/;
12 use POSIX qw/strftime/;
14 use Archive::Tar::Streamed;
21 BackupPC_ASA_PostArchive_Update
25 # /etc/BackupPC/pc/dvd_tar.pl
30 my $debug = $ENV{DEBUG} || 1;
31 my $check = $ENV{CHECK} || 1;
34 my $bpc = BackupPC::Lib->new || die "can't create BackupPC::Lib";
35 my %Conf = $bpc->Conf();
36 warn "## ARGV=",dump @ARGV;
43 if ( m/(\w+)=(.+)/ ) {
47 if ( $name =~ m/List/ ) {
48 push @{ $args->{$name} }, $v;
54 warn "args = ",dump($args);
58 %BackupPC::Search::Conf = %Conf;
60 my $path = abs_path($0);
61 $path =~ s{/[^/]+$}{/}; # FIXME remove?
67 my $t_fmt = '%Y-%m-%d %H:%M:%S';
69 #warn "## Conf = ",dump( \%Conf );
71 my $dbh = DBI->connect($Conf{SearchDSN}, $Conf{SearchUser}, "", { RaiseError => 1, AutoCommit => 0 });
77 return strftime($t_fmt,localtime());
81 my $t = shift || return;
83 my ($ss,$mm,$hh) = gmtime($t);
84 $out .= "${hh}h" if ($hh);
85 $out .= sprintf("%02d:%02d", $mm,$ss);
91 sub get_backup_id($$) {
92 my ($host, $num) = @_;
94 my $key = "$host $num";
95 return $hsn_cache->{$key} if ($hsn_cache->{$key});
97 my $sth = $dbh->prepare(qq{
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 = ?
105 $sth->execute($host, $num);
106 my ($id) = $sth->fetchrow_array;
108 $hsn_cache->{"$host $num"} = $id;
110 print STDERR "# $host $num == $id\n" if $debug;
115 sub backup_inc_deleted($) {
116 my $backup_id = shift;
117 my $sth_inc_deleted = $dbh->prepare(qq{
122 $sth_inc_deleted->execute($backup_id);
126 warn "## system_ok @_\n";
127 system(@_) == 0 || die "system @_:$!";
130 my $sth_inc_size = $dbh->prepare(qq{
139 my ($host,$num) = @_;
144 sort map { s/^\Q$Conf{ArchiveDest}\E\/*//; $_ }
145 glob "$Conf{ArchiveDest}/$host.$num.*"
148 return unless @tar_parts;
150 print curr_time, " check $host $num";
152 my $md5_path = "$Conf{ArchiveDest}/$host.$num.md5";
153 unlink $md5_path if -s $md5_path == 0; # fix empty
155 if ( ! -e $md5_path ) {
156 system_ok "cd $Conf{ArchiveDest} && /usr/bin/md5sum $host.$num.* > $md5_path";
158 system_ok "cd $Conf{ArchiveDest} && /usr/bin/md5sum -c $md5_path" if $check;
162 foreach ( split(/\n/, read_file "$Conf{ArchiveDest}/$host.$num.md5" ) ) {
163 my ( $md5, $path ) = split(/\s+/,$_);
164 $md5sum->{$path} = $md5;
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);
174 if ( $filename !~ m/\.gz$/ ) {
178 open(my $gzip, $Conf{GzipPath}." -l $filename |") || die "can't gzip -l $filename: $!";
181 $line = <$gzip> if ($line =~ /^\s+compressed/);
183 my ($comp, $uncomp) = (0,0);
185 if ($line =~ m/^\s+(\d+)\s+(\d+)\s+\d+\.\d+/) {
192 die "can't find size in line: $line";
197 my ($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{
201 id, tar_size, size, md5, items
203 where backup_id = ? and part_nr = ?
206 $sth_md5->execute($backup_id, $part_nr);
208 if (my $row = $sth_md5->fetchrow_hashref) {
210 $row->{tar_size} >= $tar_size &&
211 $row->{size} == $size &&
212 $row->{md5} eq $md5 &&
213 $row->{items} == $items
215 print ", deleting invalid backup_parts $row->{id}";
216 $dbh->do(qq{ delete from backup_parts where id = $row->{id} });
218 print ", inserting new";
219 my $sth_insert = $dbh->prepare(qq{
220 insert into backup_parts (
227 ) values (?,?,?,?,?,?)
230 $sth_insert->execute($backup_id, $part_nr, $tar_size, $size, $md5, $items);
234 print " [parts: ",join(", ", @tar_parts),"]" if $debug;
240 print " reading" if $debug;
245 foreach my $filename (@tar_parts) {
247 next if $filename eq "$host.$num.md5";
249 print "\n\t- $filename";
251 my $path = "$Conf{ArchiveDest}/$filename";
254 my $size = (stat( $path ))[7] || die "can't stat $path: $!";
256 if ($size > $Conf{MaxArchiveSize}) {
257 print ", part bigger than media $size > $Conf{MaxArchiveSize}\n";
261 print ", $size bytes";
265 open(my $fh, "gzip -cd $path |") or die "can't open $path: $!";
267 my $tar = Archive::Tar::Streamed->new($fh);
269 my $tar_size_inarc = 0;
272 while(my $entry = $tar->next) {
273 push @tar_files, $entry->name;
275 $tar_size_inarc += $entry->size;
277 if ($tar_size_inarc > $Conf{MaxArchiveFileSize}) {
278 print ", part $filename is too big $tar_size_inarc > $Conf{MaxArchiveFileSize}\n";
286 print ", $items items";
288 if ($tar_size_inarc == 0 && $items == 0) {
289 print ", EMPTY tar\n";
291 my $backup_id = get_backup_id($host, $share, $num);
292 backup_inc_deleted( $backup_id );
302 my $tar_size = get_gzip_size( $path );
305 # finally, check if backup_parts table in database is valid
308 my $md5 = $md5sum->{$filename} || die "no md5sum for $filename in ",dump($md5sum);
312 check_part($host, $num, $part_nr, $tar_size, $size, $md5, $items);
314 # round increment size to 2k block size
315 $inc_size += int(($size + 2048) / 2048);
318 $sth_inc_size->execute(
321 get_backup_id($host, $num),
325 @tar_files = sort @tar_files;
326 print "\n\t",($#tar_files + 1), " tar files";
328 my $sth = $dbh->prepare(qq{
331 JOIN shares on shares.id = shareid
332 JOIN hosts on hosts.id = shares.hostid
333 WHERE hosts.name = ? and backupnum = ?
335 $sth->execute($host, $num);
337 while( my $row = $sth->fetchrow_hashref ) {
339 my $path = $row->{'path'} || die "no path?";
341 $path .= '/' if ($row->{'type'} == BPC_FTYPE_DIR);
342 push @db_files, $path;
345 print " ",($#db_files + 1), " database files, diff";
347 @db_files = sort @db_files;
351 if ($#tar_files != $#db_files) {
355 my $diff = Algorithm::Diff->new(\@tar_files, \@db_files);
356 while ( $diff->Next() ) {
357 next if $diff->Same();
359 print "< $_\n" for $diff->Items(1);
360 print "> $_\n" for $diff->Items(2);
364 print " ",($same ? 'ok' : 'DIFFERENT'),
365 ", dur: ",fmt_time(time() - $t), "\n";
373 foreach ( 0 .. $#{ $args->{HostList} } ) {
375 my $host = $args->{'HostList'}->[$_];
376 my $num = $args->{'BackupList'}->[$_];
378 check_archive $host => $num;
384 my $sth = $dbh->prepare( qq{
387 backups.id as backup_id,
389 shares.name as share,
394 count(backup_parts.backup_id) as backup_parts
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
407 my $num_backups = $sth->rows;
410 while (my $row = $sth->fetchrow_hashref) {
414 my $tar_file = BackupPC::Search::getGzipName($row->{'host'}, $row->{'share'}, $row->{'num'});
416 # this will return -1 if file doesn't exist
417 my $size = BackupPC::Search::get_tgz_size_by_name($tar_file);
419 print "# host: ".$row->{host}.", share: ".$row->{'share'}.", backup_num:".$row->{num}." size: $size backup.size: ", $row->{inc_size},"\n" if $debug;
421 if ( $row->{'inc_size'} != -1 && $size != -1 && $row->{'inc_size'} >= $size && $row->{parts} == $row->{backup_parts}) {
423 tar_check($row->{'host'}, $row->{'share'}, $row->{'num'}, $tar_file) && next;
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";
436 my $cmd = qq[ $tarIncCreate -h "$row->{host}" -s "$row->{share}" -n $row->{num} -f ];
437 print STDERR "## $cmd\n" if ($debug);
439 if (system($cmd) != 0) {
440 print STDERR " FAILED, marking this backup deleted";
441 backup_inc_deleted( $row->{backup_id} );
445 print ", dur: ",fmt_time(time() - $t), "\n";