1 #!/usr/local/bin/perl -w
4 use lib "__INSTALLDIR__/lib";
9 use BackupPC::Attrib qw/:all/;
11 use Time::HiRes qw/time/;
12 use POSIX qw/strftime/;
15 use Archive::Tar::Streamed;
20 my $bpc = BackupPC::Lib->new || die "can't create BackupPC::Lib";
21 my %Conf = $bpc->Conf();
23 use BackupPC::SearchLib;
24 %BackupPC::SearchLib::Conf = %Conf;
26 my $path = abs_path($0);
27 $path =~ s#/[^/]+$#/#;
28 my $tarIncCreate = $path .= 'BackupPC_tarIncCreate';
30 die "can't find $tarIncCreate: $!\n" unless (-x $tarIncCreate);
33 foreach my $c (qw/gzip md5sum/) {
34 $bin->{$c} = which($c) || die "$0 needs $c, install it\n";
38 getopts("cd", \%opt );
41 my $check = $opt{c} && print STDERR "NOTICE: tar archive check forced\n";
47 my $t_fmt = '%Y-%m-%d %H:%M:%S';
49 my $dsn = $Conf{SearchDSN} || die "Need SearchDSN in config.pl\n";
50 my $user = $Conf{SearchUser} || '';
52 my $dbh = DBI->connect($dsn, $user, "", { RaiseError => 1, AutoCommit => 0 });
54 my $tar_dir = $Conf{InstallDir}.'/'.$Conf{GzipTempDir};
56 die "problem with $tar_dir, check GzipTempDir in configuration\n" unless (-d $tar_dir && -w $tar_dir);
61 my $t = shift || return;
63 my ($ss,$mm,$hh) = gmtime($t);
64 $out .= "${hh}h" if ($hh);
65 $out .= sprintf("%02d:%02d", $mm,$ss);
70 return strftime($t_fmt,localtime());
75 sub get_backup_id($$$) {
76 my ($host, $share, $num) = @_;
78 my $key = "$host $share $num";
79 return $hsn_cache->{$key} if ($hsn_cache->{$key});
81 my $sth = $dbh->prepare(qq{
85 INNER JOIN shares ON backups.shareID=shares.ID
86 INNER JOIN hosts ON backups.hostID = hosts.ID
87 where hosts.name = ? and shares.name = ? and backups.num = ?
89 $sth->execute($host, $share, $num);
90 my ($id) = $sth->fetchrow_array;
92 $hsn_cache->{"$host $share $num"} = $id;
94 print STDERR "# $host $share $num == $id\n" if ($opt{d});
100 sub tar_check($$$$) {
101 my ($host,$share,$num,$filename) = @_;
104 my ($host, $share, $num, $part_nr, $tar_size, $size, $md5, $items) = @_;
105 my $backup_id = get_backup_id($host, $share, $num);
106 my $sth_md5 = $dbh->prepare(qq{
108 id, tar_size, size, md5, items
110 where backup_id = ? and part_nr = ?
113 $sth_md5->execute($backup_id, $part_nr);
115 if (my $row = $sth_md5->fetchrow_hashref) {
117 $row->{tar_size} == $tar_size &&
118 $row->{size} == $size &&
119 $row->{md5} eq $md5 &&
120 $row->{items} == $items
122 print STDERR "# deleting invalid row $row->{id}\n" if ($opt{d});
123 $dbh->do(qq{ delete from backup_parts where id = $row->{id} });
125 print STDERR "# inserting new backup_part row\n";
126 my $sth_insert = $dbh->prepare(qq{
127 insert into backup_parts (
134 ) values (?,?,?,?,?,?)
137 $sth_insert->execute($backup_id, $part_nr, $tar_size, $size, $md5, $items);
142 print STDERR " {{ CHECK: ${host}:${share}#${num} and $filename";
149 if (-d "$tar_dir/$filename") {
150 print STDERR " multi-part" if ($opt{d});
151 opendir(my $dir, "$tar_dir/$filename") || die "can't readdir $tar_dir/$filename: $!";
152 @tar_parts = map { my $p = $_; $p =~ s#^#${filename}/#; $p } grep { !/^\./ && !/md5/ && -f "$tar_dir/$filename/$_" } readdir($dir);
155 push @tar_parts, "${filename}.tar.gz";
158 print " [parts: ",join(", ", @tar_parts),"]" if ($opt{d});
167 foreach my $tarfilename (@tar_parts) {
169 print STDERR " $tarfilename" if ($debug);
171 my $path = "$tar_dir/$tarfilename";
172 my $md5_path = $path;
173 $md5_path =~ s/\.tar\.gz$/.md5/ || die "can't create md5 filename from $md5_path";
174 if (! -e $md5_path) {
175 print ", creating md5";
176 system( $bin->{md5sum} . " $path > $md5_path") == 0 or die "can't create md5 $path: $!";
179 my $md5 = read_file( $md5_path ) || die "can't read md5sum file $md5_path: $!";
182 $part_nr = $1 if ($tarfilename =~ m#/(\d+)\.tar\.gz#);
184 my $size = (stat( "$tar_dir/$tarfilename" ))[7] || die "can't stat $tar_dir/$tarfilename";
186 open(my $fh, "gzip -cd $tar_dir/$tarfilename |") or die "can't open $tar_dir/$tarfilename: $!";
188 my $tar = Archive::Tar::Streamed->new($fh);
193 while(my $entry = $tar->next) {
194 push @tar_files, $entry->name;
196 $tar_size += $entry->size;
199 if ($tar_size > $Conf{MaxArchiveFileSize}) {
200 print STDERR " part too big $tar_size > $Conf{MaxArchiveFileSize} }}" if ($debug);
203 } elsif ($size > $Conf{MaxArchiveSize}) {
204 print STDERR " part bigger than media $size > $Conf{MaxArchiveSize} }}" if ($debug);
209 check_part($host, $share, $num, $part_nr, $tar_size, $size, $md5, $items);
212 # short-cut and exit;
213 return $same unless($same);
215 @tar_files = sort @tar_files;
216 print STDERR " ",($#tar_files + 1), " files" if ($debug);
218 print STDERR ", database" if ($debug);
220 my $sth = $dbh->prepare(qq{
223 JOIN shares on shares.id = shareid
224 JOIN hosts on hosts.id = shares.hostid
225 WHERE hosts.name = ? and shares.name = ? and backupnum = ?
227 $sth->execute($host, $share, $num);
229 while( my $row = $sth->fetchrow_hashref ) {
231 my $path = $row->{'path'} || die "no path?";
233 $path .= '/' if ($row->{'type'} == BPC_FTYPE_DIR);
234 push @db_files, $path;
237 print STDERR " ",($#db_files + 1), " files, diff" if ($debug);
239 @db_files = sort @db_files;
241 if ($#tar_files != $#db_files) {
243 print STDERR " NUMBER" if ($debug);
245 my $diff = Algorithm::Diff->new(\@tar_files, \@db_files);
246 while ( $diff->Next() ) {
247 next if $diff->Same();
249 print "< $_\n" for $diff->Items(1);
250 print "> $_\n" for $diff->Items(2);
254 print " ",($same ? 'ok' : 'DIFFERENT');
255 print STDERR " }} " if ($debug);
263 my $sth = $dbh->prepare( qq{
266 backups.id as backup_id,
268 shares.name as share,
273 join shares on backups.hostid = shares.hostid
274 and shares.id = backups.shareid
275 join hosts on shares.hostid = hosts.id
276 where not inc_deleted
277 order by backups.date
282 my $num_backups = $sth->rows;
285 while (my $row = $sth->fetchrow_hashref) {
286 my $tar_file = BackupPC::SearchLib::getGzipName($row->{'host'}, $row->{'share'}, $row->{'num'});
288 # this will return -1 if file doesn't exist
289 my $size = BackupPC::SearchLib::get_tgz_size_by_name($tar_file);
291 print "# size: $size backup.size: ", $row->{inc_size},"\n" if ($opt{d});
293 if ( $row->{'inc_size'} != -1 && $size != -1 && $row->{'inc_size'} == $size) {
295 tar_check($row->{'host'}, $row->{'share'}, $row->{'num'}, $tar_file) && next;
301 print curr_time, " $curr_backup/$num_backups ", $row->{'host'}, ":", $row->{'share'}, " #", $row->{'num'}, " -> $tar_file";
307 my $cmd = qq{ $tarIncCreate -h "$row->{'host'}" -s "$row->{'share'}" -n $row->{'num'} -f };
308 print STDERR "## $cmd\n" if ($debug);
310 if (system($cmd) != 0) {
311 print STDERR " FAILED";
314 print ", dur: ",fmt_time(time() - $t), "\n";