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;
17 use Getopt::Long::Descriptive;
21 BackupPC_ASA_PostArchive_Update
25 # /etc/BackupPC/pc/_search_archive.pl
30 my $bpc = BackupPC::Lib->new || die "can't create BackupPC::Lib";
31 $bpc->ConfigRead('_search_archive'); # read our configuration
32 my %Conf = $bpc->Conf();
35 %BackupPC::Search::Conf = %Conf;
37 my ($opt,$usage) = describe_options(
39 [ 'host|h=s@', "import just host(s)" ],
40 [ 'num|n=s@', "import just backup number(s)" ],
41 [ 'ok=n', "xferOK", { default => 0 } ],
42 [ 'check|c', "check archives on disk and sync", { default => 1 } ],
43 [ 'debug|d', "debug", { default => 1 } ],
44 [ 'help', "show help" ],
47 print($usage->text), exit if $opt->help;
53 my $t_fmt = '%Y-%m-%d %H:%M:%S';
55 #warn "## Conf = ",dump( \%Conf );
57 my $dbh = DBI->connect($Conf{SearchDSN}, $Conf{SearchUser}, "", { RaiseError => 1, AutoCommit => 0 });
63 return strftime($t_fmt,localtime());
67 my $t = shift || return;
69 my ($ss,$mm,$hh) = gmtime($t);
70 $out .= "${hh}h" if ($hh);
71 $out .= sprintf("%02d:%02d", $mm,$ss);
77 sub get_backup_id($$) {
78 my ($host, $num) = @_;
80 my $key = "$host $num";
81 return $hsn_cache->{$key} if ($hsn_cache->{$key});
83 my $sth = $dbh->prepare(qq{
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 = ?
91 $sth->execute($host, $num);
92 my ($id) = $sth->fetchrow_array;
94 $hsn_cache->{"$host $num"} = $id;
96 print STDERR "# $host $num == $id\n" if $opt->debug;
101 sub backup_inc_deleted($) {
102 my $backup_id = shift;
103 my $sth_inc_deleted = $dbh->prepare(qq{
108 $sth_inc_deleted->execute($backup_id);
112 warn "## system_ok @_\n";
113 system(@_) == 0 || die "system @_:$!";
116 my $sth_inc_size = $dbh->prepare(qq{
126 my $perm = (stat $full)[2] & 0444;
127 warn sprintf("chmod %03o %s\n",$perm,$full);
128 chmod $perm, $full || die $!;
132 my ($host,$num) = @_;
133 warn "# check_archive $host $num";
137 my $glob = "$Conf{ArchiveDest}/$host.$num.*";
139 my @tar_parts = sort map { s/^\Q$Conf{ArchiveDest}\E\/*//; $_ } glob $glob ;
141 if ( ! @tar_parts ) {
142 warn "ERROR: no files for $glob";
146 print curr_time, " check $host $num";
148 my $md5_path = "$Conf{ArchiveDest}/$host.$num.md5";
149 unlink $md5_path if -s $md5_path == 0; # fix empty
151 my $read_protect = 0;
153 if ( ! -e $md5_path ) {
154 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 $opt->check;
162 foreach ( split(/\n/, read_file "$Conf{ArchiveDest}/$host.$num.md5" ) ) {
163 my ( $md5, $path ) = split(/\s+/,$_);
164 $md5sum->{$path} = $md5;
165 read_only "$Conf{ArchiveDest}/$path" if $read_protect;
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);
175 if ( $filename !~ m/\.gz$/ ) {
179 open(my $gzip, $Conf{GzipPath}." -l $filename |") || die "can't gzip -l $filename: $!";
184 my ($comp, $uncomp) = (0,0);
186 if ($line =~ m/\s+(\d+)\s+(\d+)\s+\d+\.\d+/s) {
193 warn "ERROR can't parse: $line";
199 my ($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{
203 id, tar_size, size, md5, items, filename
205 where backup_id = ? and part_nr = ? and filename = ?
208 $sth_md5->execute($backup_id, $part_nr, $filename);
210 if (my $row = $sth_md5->fetchrow_hashref) {
212 $row->{tar_size} >= $tar_size &&
213 $row->{size} == $size &&
214 $row->{md5} eq $md5 &&
215 $row->{items} == $items
217 print ", deleting invalid backup_parts $row->{id}";
218 $dbh->do(qq{ delete from backup_parts where id = $row->{id} });
220 print ", inserting new";
221 my $sth_insert = $dbh->prepare(qq{
222 insert into backup_parts (
230 ) values (?,?,?,?,?,?,?)
233 $sth_insert->execute($backup_id, $part_nr, $tar_size, $size, $md5, $items, $filename);
236 print " [parts: ",join(", ", @tar_parts),"]" if $opt->debug;
242 print " reading" if $opt->debug;
247 foreach my $filename (@tar_parts) {
249 next if $filename eq "$host.$num.md5";
251 print "\n\t- $filename";
253 my $path = "$Conf{ArchiveDest}/$filename";
256 my $size = (stat( $path ))[7] || die "can't stat $path: $!";
258 if ($size > $Conf{ArchiveMediaSize}) {
259 print ", part bigger than media $size > $Conf{ArchiveMediaSize}\n";
263 print ", $size bytes";
267 open(my $fh, "gzip -cd $path |") or die "can't open $path: $!";
269 my $tar = Archive::Tar::Streamed->new($fh);
271 my $tar_size_inarc = 0;
274 while(my $entry = $tar->next) {
275 push @tar_files, $entry->name;
277 $tar_size_inarc += $entry->size;
279 if ($tar_size_inarc > $Conf{ArchiveMediaSize}) {
280 print ", part $filename is too big $tar_size_inarc > $Conf{ArchiveMediaSize}\n";
288 print ", $items items";
290 if ($tar_size_inarc == 0 && $items == 0) {
291 print ", EMPTY tar\n";
293 my $backup_id = get_backup_id($host, $share, $num);
294 backup_inc_deleted( $backup_id );
304 my $tar_size = get_gzip_size( $path );
307 # finally, check if backup_parts table in database is valid
310 my $md5 = $md5sum->{$filename} || die "no md5sum for $filename in ",dump($md5sum);
314 check_part($host, $num, $part_nr, $tar_size, $size, $md5, $items, $filename);
316 # round increment size to 2k block size
317 $inc_size += int((($size + 2048) / 2048 ) * 2048);
320 $sth_inc_size->execute(
323 get_backup_id($host, $num),
326 @tar_files = sort @tar_files;
327 print "\n\t",($#tar_files + 1), " tar files";
329 my $sth = $dbh->prepare(qq{
332 JOIN shares on shares.id = shareid
333 JOIN hosts on hosts.id = shares.hostid
334 WHERE hosts.name = ? and backupnum = ?
336 $sth->execute($host, $num);
338 while( my $row = $sth->fetchrow_hashref ) {
340 my $path = $row->{'path'} || die "no path?";
342 $path .= '/' if ($row->{'type'} == BPC_FTYPE_DIR);
343 push @db_files, $path;
346 print " ",($#db_files + 1), " database files, diff";
348 @db_files = sort @db_files;
352 if ($#tar_files != $#db_files) {
356 my $diff = Algorithm::Diff->new(\@tar_files, \@db_files);
357 while ( $diff->Next() ) {
358 next if $diff->Same();
360 print "< $_\n" for $diff->Items(1);
361 print "> $_\n" for $diff->Items(2);
365 print " ",($same ? 'ok' : 'DIFFERENT'),
366 ", dur: ",fmt_time(time() - $t), "\n";
376 foreach ( 0 .. $#{ $opt->host } ) {
378 my $host = lc $opt->host->[$_];
379 my $num = $opt->num->[$_];
382 warn "ERROR $host $num running cleanup";
383 foreach my $path ( glob "$Conf{ArchiveDest}/$host.$num.*" ) {
385 unlink $path || die $!;
388 check_archive $host => $num;
395 my $sth = $dbh->prepare( qq{
398 backups.id as backup_id,
400 shares.name as share,
405 count(backup_parts.backup_id) as backup_parts
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
418 my $num_backups = $sth->rows;
421 while (my $row = $sth->fetchrow_hashref) {
425 my $tar_file = BackupPC::Search::getGzipName($row->{'host'}, $row->{'share'}, $row->{'num'});
427 # this will return -1 if file doesn't exist
428 my $size = BackupPC::Search::get_tgz_size_by_name($tar_file);
430 print "# host: ".$row->{host}.", share: ".$row->{'share'}.", backup_num:".$row->{num}." size: $size backup.size: ", $row->{inc_size},"\n" if $opt->debug;
432 if ( $row->{'inc_size'} != -1 && $size != -1 && $row->{'inc_size'} >= $size && $row->{parts} == $row->{backup_parts}) {
434 tar_check($row->{'host'}, $row->{'share'}, $row->{'num'}, $tar_file) && next;
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";
447 my $cmd = qq[ $tarIncCreate -h "$row->{host}" -s "$row->{share}" -n $row->{num} -f ];
448 print STDERR "## $cmd\n" if ($opt->debug);
450 if (system($cmd) != 0) {
451 print STDERR " FAILED, marking this backup deleted";
452 backup_inc_deleted( $row->{backup_id} );
456 print ", dur: ",fmt_time(time() - $t), "\n";