Annotation of /trunk/bin/BackupPC_incPartsUpdate

#!/usr/local/bin/perl -w

use strict;
use lib "__INSTALLDIR__/lib";

use DBI;
use BackupPC::Lib;
use BackupPC::View;
use BackupPC::Attrib qw/:all/;
use Data::Dumper;
use Time::HiRes qw/time/;
use POSIX qw/strftime/;
use Cwd qw/abs_path/;
use File::Which;
use Archive::Tar::Streamed;
use Algorithm::Diff;
use Getopt::Std;
use File::Slurp;
use File::Pid;

=head1 NAME

BackupPC_incPartsUpdate

=head1 DESCRIPTION

Create C<.tar.gz> increments on disk calling C<BackupPC_tarIncCreate>.

Following options are supported (but all are optional):

=over 4

=item -h hostname

Update parts for just single C<hostname>

=item -c

Force check for tar archives which exist on disk

=item -d

Turn debugging output

=back

=cut

my %opt;
getopts("cdh:", \%opt );

my $debug = $opt{d};
my $check = $opt{c} && print STDERR "NOTICE: tar archive check forced\n";

my $pid_path = abs_path($0);
$pid_path =~ s/\W+/_/g;

my $pidfile = new File::Pid({
        file => "/tmp/$pid_path",
});

if (my $pid = $pidfile->running ) {
        die "$0 already running: $pid\n";
} elsif ($pidfile->pid ne $$) {
        $pidfile->remove;
        $pidfile = new File::Pid;
}

print STDERR "$0 using pid ",$pidfile->pid," file ",$pidfile->file,"\n";
$pidfile->write;

my $bpc = BackupPC::Lib->new || die "can't create BackupPC::Lib";
my %Conf = $bpc->Conf();

use BackupPC::SearchLib;
%BackupPC::SearchLib::Conf = %Conf;

my $path = abs_path($0);
$path =~ s#/[^/]+$#/#;
my $tarIncCreate = $path .= 'BackupPC_tarIncCreate';

die "can't find $tarIncCreate: $!\n" unless (-x $tarIncCreate);

my $bin;
foreach my $c (qw/gzip md5sum/) {
        $bin->{$c} = which($c) || die "$0 needs $c, install it\n";
}

$|=1;

my $start_t = time();

my $t_fmt = '%Y-%m-%d %H:%M:%S';

my $dsn = $Conf{SearchDSN} || die "Need SearchDSN in config.pl\n";
my $user = $Conf{SearchUser} || '';

my $dbh = DBI->connect($dsn, $user, "", { RaiseError => 1, AutoCommit => 0 });

my $tar_dir = $Conf{InstallDir}.'/'.$Conf{GzipTempDir};

die "problem with $tar_dir, check GzipTempDir in configuration\n" unless (-d $tar_dir && -w $tar_dir);

#---- subs ----

sub fmt_time {
        my $t = shift || return;
        my $out = "";
        my ($ss,$mm,$hh) = gmtime($t);
        $out .= "${hh}h" if ($hh);
        $out .= sprintf("%02d:%02d", $mm,$ss);
        return $out;
}

sub curr_time {
        return strftime($t_fmt,localtime());
}

my $hsn_cache;

sub get_backup_id($$$) {
        my ($host, $share, $num) = @_;

        my $key = "$host $share $num";
        return $hsn_cache->{$key} if ($hsn_cache->{$key});

        my $sth = $dbh->prepare(qq{
                SELECT 
                        backups.id
                FROM backups 
                INNER JOIN shares       ON backups.shareID=shares.ID
                INNER JOIN hosts        ON backups.hostID = hosts.ID
                where hosts.name = ? and shares.name = ? and backups.num = ?
        });
        $sth->execute($host, $share, $num);
        my ($id) = $sth->fetchrow_array;

        $hsn_cache->{"$host $share $num"} = $id;

        print STDERR "# $host $share $num == $id\n" if ($opt{d});

        return $id;
}


sub tar_check($$$$) {
        my ($host,$share,$num,$filename) = @_;

        my $t = time();
        print curr_time, " check $host:$share#$num -> $filename";

        # depending on expected returned value this is used like:
        # my $uncompress_size = get_gzip_size('/full/path/to.gz');
        # my ($compress_size, $uncompress_size) = get_gzip_size('/path.gz');
        sub get_gzip_size($) {
                my $filename = shift;
                die "file $filename problem: $!" unless (-r $filename);
                open(my $gzip, $bin->{gzip}." -l $filename |") || die "can't gzip -l $filename: $!";
                my $line = <$gzip>;
                chomp($line);
                $line = <$gzip> if ($line =~ /^\s+compressed/);

                my ($comp, $uncomp) = (0,0);

                if ($line =~ m/^\s+(\d+)\s+(\d+)\s+\d+\.\d+/) {
                        if (wantarray) {
                                return [ $1, $2 ];
                        } else {
                                return $2;
                        }
                } else {
                        die "can't find size in line: $line";
                }
        }

        sub check_part {
                my ($host, $share, $num, $part_nr, $tar_size, $size, $md5, $items) = @_;
                my $backup_id = get_backup_id($host, $share, $num);
                my $sth_md5 = $dbh->prepare(qq{
                        select
                                id, tar_size, size, md5, items
                        from backup_parts
                        where backup_id = ? and part_nr = ?
                });

                $sth_md5->execute($backup_id, $part_nr);

                if (my $row = $sth_md5->fetchrow_hashref) {
                        return if (
                                $row->{tar_size} >= $tar_size &&
                                $row->{size} == $size &&
                                $row->{md5} eq $md5 &&
                                $row->{items} == $items
                        );
                        print ", deleting invalid backup_parts $row->{id}";
                        $dbh->do(qq{ delete from backup_parts where id = $row->{id} });
                }
                print ", inserting new";
                my $sth_insert = $dbh->prepare(qq{
                        insert into backup_parts (
                                backup_id,
                                part_nr,
                                tar_size,
                                size,
                                md5,
                                items
                        ) values (?,?,?,?,?,?)
                });

                $sth_insert->execute($backup_id, $part_nr, $tar_size, $size, $md5, $items);
                $dbh->commit;
        }

        my @tar_parts;

        if (-d "$tar_dir/$filename") {
                print ", multi-part";
                opendir(my $dir, "$tar_dir/$filename") || die "can't readdir $tar_dir/$filename: $!";
                @tar_parts = map { my $p = $_; $p =~ s#^#${filename}/#; $p } grep { !/^\./ && !/md5/ && -f "$tar_dir/$filename/$_" } readdir($dir);
                closedir($dir);
        } else {
                push @tar_parts, "${filename}.tar.gz";
        }

        print " [parts: ",join(", ", @tar_parts),"]" if ($opt{d});

        my $same = 1;
        my @tar_files;

        my $backup_part;

        print " reading" if ($opt{d});

        foreach my $tarfilename (@tar_parts) {

                print "\n\t- $tarfilename";

                my $path = "$tar_dir/$tarfilename";

                my $size = (stat( $path ))[7] || die "can't stat $path: $!";

                if ($size > $Conf{MaxArchiveSize}) {
                        print ", part bigger than media $size > $Conf{MaxArchiveSize}\n";
                        return 0;
                }

                print ", $size bytes";


                open(my $fh, "gzip -cd $path |") or die "can't open $path: $!";
                binmode($fh);
                my $tar = Archive::Tar::Streamed->new($fh);

                my $tar_size_inarc = 0;
                my $items = 0;

                while(my $entry = $tar->next) {
                        push @tar_files, $entry->name;
                        $items++;
                        $tar_size_inarc += $entry->size;

                        if ($tar_size_inarc > $Conf{MaxArchiveFileSize}) {
                                print ", part $tarfilename is too big $tar_size_inarc > $Conf{MaxArchiveFileSize}\n";
                                return 0;
                        }

                }

                close($fh);

                print ", $items items";

                if ($tar_size_inarc == 0 && $items == 0) {
                        print ", EMPTY tar\n";

                        my $backup_id = get_backup_id($host, $share, $num);

                        my $sth_inc_deleted = $dbh->prepare(qq{
                                update backups set
                                        inc_deleted = true
                                where id = ?
                        });
                        $sth_inc_deleted->execute($backup_id);

                        $dbh->commit;

                        return 1;
                }

                my $tar_size = get_gzip_size( $path );

                # real tar size is bigger because of padding    
                if ($tar_size_inarc > $tar_size) {
                        print ", size of files in tar ($tar_size_inarc) bigger than whole tar ($tar_size)!\n";
                        return 0;
                }

                #
                # check if md5 exists, and if not, create one
                #

                my $md5_path = $path;
                $md5_path =~ s/\.tar\.gz$/.md5/ || die "can't create md5 filename from $md5_path";
                if (! -e $md5_path || -z $md5_path) {
                        print ", creating md5";
                        system( $bin->{md5sum} . " $path > $md5_path") == 0 or die "can't create md5 $path: $!";
                } else {
                        ## FIXME check if existing md5 is valid
                }

                my $md5 = read_file( $md5_path ) || die "can't read md5sum file $md5_path: $!";
                $md5 =~ s#\s.*$##;

                # extract part number from filename
                my $part_nr = 1;
                $part_nr = $1 if ($tarfilename =~ m#/(\d+)\.tar\.gz#);

                #
                # finally, check if backup_parts table in database is valid
                #

                check_part($host, $share, $num, $part_nr, $tar_size, $size, $md5, $items);
        }

        # short-cut and exit;
        return $same unless($same);

        @tar_files = sort @tar_files;
        print "\n\t",($#tar_files + 1), " tar files";

        my $sth = $dbh->prepare(qq{
                SELECT path,type
                FROM files
                JOIN shares on shares.id = shareid
                JOIN hosts on hosts.id = shares.hostid
                WHERE hosts.name = ? and shares.name = ? and backupnum = ?
        });
        $sth->execute($host, $share, $num);
        my @db_files;
        while( my $row = $sth->fetchrow_hashref ) {

                my $path = $row->{'path'} || die "no path?";
                $path =~ s#^/#./#;
                $path .= '/' if ($row->{'type'} == BPC_FTYPE_DIR);
                push @db_files, $path;
        }

        print " ",($#db_files + 1), " database files, diff";

        @db_files = sort @db_files;

        if ($#tar_files != $#db_files) {
                $same = 0;
                print " NUMBER";
        } else {
                my $diff = Algorithm::Diff->new(\@tar_files, \@db_files);
                while ( $diff->Next() ) {
                        next if $diff->Same();
                        $same = 0;
                        print "< $_\n" for $diff->Items(1);
                        print "> $_\n" for $diff->Items(2);
                }
        }

        print " ",($same ? 'ok' : 'DIFFERENT'),
                ", dur: ",fmt_time(time() - $t), "\n";

        return $same;
}


#----- main

my $sth = $dbh->prepare( qq{
        
select
        backups.id as backup_id,
        hosts.name as host,
        shares.name as share,
        backups.num as num,
        inc_size,
        parts
from backups
        join shares on backups.hostid = shares.hostid
                and shares.id = backups.shareid
        join hosts on shares.hostid = hosts.id
where not inc_deleted
order by backups.date

} );

$sth->execute();
my $num_backups = $sth->rows;
my $curr_backup = 1;

if ($opt{h}) {
        warn "making increments just for host $opt{h}\n";
}

while (my $row = $sth->fetchrow_hashref) {

        if ($opt{h} && $row->{host} ne $opt{h}) {
                warn "skipped $row->{host}\n" if ($debug);
                next;
        }

        $curr_backup++;

        my $tar_file = BackupPC::SearchLib::getGzipName($row->{'host'}, $row->{'share'}, $row->{'num'});

        # this will return -1 if file doesn't exist
        my $size = BackupPC::SearchLib::get_tgz_size_by_name($tar_file);

        print "# size: $size backup.size: ", $row->{inc_size},"\n" if ($opt{d});

        if ( $row->{'inc_size'} != -1 && $size != -1 && $row->{'inc_size'} >= $size) {
                if ($check) {
                        tar_check($row->{'host'}, $row->{'share'}, $row->{'num'}, $tar_file) && next;
                } else {
                        next;
                }
        }

        print curr_time, " creating $curr_backup/$num_backups ", $row->{'host'}, ":", $row->{'share'}, " #", $row->{'num'}, " -> $tar_file";

        my $t = time();

        # re-create archive?
        my $cmd = qq{ $tarIncCreate -h "$row->{'host'}" -s "$row->{'share'}" -n $row->{'num'} -f };
        print STDERR "## $cmd\n" if ($debug);

        if (system($cmd) != 0) {
                print STDERR " FAILED";
        }

        print ", dur: ",fmt_time(time() - $t), "\n";

        $dbh->commit;

}

undef $sth;
$dbh->disconnect;
1	dpavlin	157	#!/usr/local/bin/perl -w
2
3			use strict;
4			use lib "__INSTALLDIR__/lib";
5
6			use DBI;
7			use BackupPC::Lib;
8			use BackupPC::View;
9	dpavlin	214	use BackupPC::Attrib qw/:all/;
10	dpavlin	157	use Data::Dumper;
11			use Time::HiRes qw/time/;
12			use POSIX qw/strftime/;
13			use Cwd qw/abs_path/;
14	dpavlin	201	use File::Which;
15	dpavlin	215	use Archive::Tar::Streamed;
16	dpavlin	214	use Algorithm::Diff;
17			use Getopt::Std;
18	dpavlin	264	use File::Slurp;
19	dpavlin	289	use File::Pid;
20	dpavlin	157
21	dpavlin	293	=head1 NAME
22
23			BackupPC_incPartsUpdate
24
25			=head1 DESCRIPTION
26
27			Create C<.tar.gz> increments on disk calling C<BackupPC_tarIncCreate>.
28
29			Following options are supported (but all are optional):
30
31			=over 4
32
33			=item -h hostname
34
35			Update parts for just single C<hostname>
36
37			=item -c
38
39			Force check for tar archives which exist on disk
40
41			=item -d
42
43			Turn debugging output
44
45			=back
46
47			=cut
48
49			my %opt;
50			getopts("cdh:", \%opt );
51
52			my $debug = $opt{d};
53			my $check = $opt{c} && print STDERR "NOTICE: tar archive check forced\n";
54
55	dpavlin	289	my $pid_path = abs_path($0);
56			$pid_path =~ s/\W+/_/g;
57
58			my $pidfile = new File::Pid({
59			file => "/tmp/$pid_path",
60			});
61
62			if (my $pid = $pidfile->running ) {
63			die "$0 already running: $pid\n";
64			} elsif ($pidfile->pid ne $$) {
65			$pidfile->remove;
66			$pidfile = new File::Pid;
67			}
68
69			print STDERR "$0 using pid ",$pidfile->pid," file ",$pidfile->file,"\n";
70			$pidfile->write;
71
72	dpavlin	230	my $bpc = BackupPC::Lib->new \|\| die "can't create BackupPC::Lib";
73			my %Conf = $bpc->Conf();
74
75			use BackupPC::SearchLib;
76			%BackupPC::SearchLib::Conf = %Conf;
77
78	dpavlin	157	my $path = abs_path($0);
79			$path =~ s#/[^/]+$#/#;
80			my $tarIncCreate = $path .= 'BackupPC_tarIncCreate';
81
82			die "can't find $tarIncCreate: $!\n" unless (-x $tarIncCreate);
83
84	dpavlin	201	my $bin;
85	dpavlin	254	foreach my $c (qw/gzip md5sum/) {
86	dpavlin	201	$bin->{$c} = which($c) \|\| die "$0 needs $c, install it\n";
87			}
88
89	dpavlin	157	$\|=1;
90
91			my $start_t = time();
92
93			my $t_fmt = '%Y-%m-%d %H:%M:%S';
94
95			my $dsn = $Conf{SearchDSN} \|\| die "Need SearchDSN in config.pl\n";
96			my $user = $Conf{SearchUser} \|\| '';
97
98			my $dbh = DBI->connect($dsn, $user, "", { RaiseError => 1, AutoCommit => 0 });
99
100			my $tar_dir = $Conf{InstallDir}.'/'.$Conf{GzipTempDir};
101
102			die "problem with $tar_dir, check GzipTempDir in configuration\n" unless (-d $tar_dir && -w $tar_dir);
103
104			#---- subs ----
105
106			sub fmt_time {
107			my $t = shift \|\| return;
108			my $out = "";
109			my ($ss,$mm,$hh) = gmtime($t);
110			$out .= "${hh}h" if ($hh);
111			$out .= sprintf("%02d:%02d", $mm,$ss);
112			return $out;
113			}
114
115			sub curr_time {
116			return strftime($t_fmt,localtime());
117			}
118
119	dpavlin	264	my $hsn_cache;
120
121			sub get_backup_id($$$) {
122			my ($host, $share, $num) = @_;
123
124			my $key = "$host $share $num";
125			return $hsn_cache->{$key} if ($hsn_cache->{$key});
126
127			my $sth = $dbh->prepare(qq{
128			SELECT
129			backups.id
130			FROM backups
131			INNER JOIN shares ON backups.shareID=shares.ID
132			INNER JOIN hosts ON backups.hostID = hosts.ID
133			where hosts.name = ? and shares.name = ? and backups.num = ?
134			});
135			$sth->execute($host, $share, $num);
136			my ($id) = $sth->fetchrow_array;
137
138			$hsn_cache->{"$host $share $num"} = $id;
139
140			print STDERR "# $host $share $num == $id\n" if ($opt{d});
141
142			return $id;
143			}
144
145
146	dpavlin	214	sub tar_check($$$$) {
147			my ($host,$share,$num,$filename) = @_;
148
149	dpavlin	270	my $t = time();
150			print curr_time, " check $host:$share#$num -> $filename";
151
152	dpavlin	281	# depending on expected returned value this is used like:
153			# my $uncompress_size = get_gzip_size('/full/path/to.gz');
154			# my ($compress_size, $uncompress_size) = get_gzip_size('/path.gz');
155			sub get_gzip_size($) {
156			my $filename = shift;
157			die "file $filename problem: $!" unless (-r $filename);
158			open(my $gzip, $bin->{gzip}." -l $filename \|") \|\| die "can't gzip -l $filename: $!";
159			my $line = <$gzip>;
160			chomp($line);
161			$line = <$gzip> if ($line =~ /^\s+compressed/);
162
163			my ($comp, $uncomp) = (0,0);
164
165			if ($line =~ m/^\s+(\d+)\s+(\d+)\s+\d+\.\d+/) {
166			if (wantarray) {
167			return [ $1, $2 ];
168			} else {
169			return $2;
170			}
171			} else {
172			die "can't find size in line: $line";
173			}
174			}
175
176	dpavlin	264	sub check_part {
177			my ($host, $share, $num, $part_nr, $tar_size, $size, $md5, $items) = @_;
178			my $backup_id = get_backup_id($host, $share, $num);
179			my $sth_md5 = $dbh->prepare(qq{
180			select
181			id, tar_size, size, md5, items
182			from backup_parts
183			where backup_id = ? and part_nr = ?
184			});
185
186			$sth_md5->execute($backup_id, $part_nr);
187
188			if (my $row = $sth_md5->fetchrow_hashref) {
189			return if (
190	dpavlin	265	$row->{tar_size} >= $tar_size &&
191	dpavlin	264	$row->{size} == $size &&
192			$row->{md5} eq $md5 &&
193			$row->{items} == $items
194			);
195	dpavlin	270	print ", deleting invalid backup_parts $row->{id}";
196	dpavlin	264	$dbh->do(qq{ delete from backup_parts where id = $row->{id} });
197			}
198	dpavlin	270	print ", inserting new";
199	dpavlin	264	my $sth_insert = $dbh->prepare(qq{
200			insert into backup_parts (
201			backup_id,
202			part_nr,
203			tar_size,
204			size,
205			md5,
206			items
207			) values (?,?,?,?,?,?)
208			});
209
210			$sth_insert->execute($backup_id, $part_nr, $tar_size, $size, $md5, $items);
211			$dbh->commit;
212			}
213
214	dpavlin	254	my @tar_parts;
215
216			if (-d "$tar_dir/$filename") {
217	dpavlin	270	print ", multi-part";
218	dpavlin	254	opendir(my $dir, "$tar_dir/$filename") \|\| die "can't readdir $tar_dir/$filename: $!";
219			@tar_parts = map { my $p = $_; $p =~ s#^#${filename}/#; $p } grep { !/^\./ && !/md5/ && -f "$tar_dir/$filename/$_" } readdir($dir);
220			closedir($dir);
221			} else {
222			push @tar_parts, "${filename}.tar.gz";
223	dpavlin	214	}
224
225	dpavlin	256	print " [parts: ",join(", ", @tar_parts),"]" if ($opt{d});
226
227			my $same = 1;
228	dpavlin	254	my @tar_files;
229	dpavlin	214
230	dpavlin	264	my $backup_part;
231
232	dpavlin	268	print " reading" if ($opt{d});
233	dpavlin	254
234			foreach my $tarfilename (@tar_parts) {
235
236	dpavlin	270	print "\n\t- $tarfilename";
237	dpavlin	254
238	dpavlin	281	my $path = "$tar_dir/$tarfilename";
239	dpavlin	269
240	dpavlin	281	my $size = (stat( $path ))[7] \|\| die "can't stat $path: $!";
241
242	dpavlin	269	if ($size > $Conf{MaxArchiveSize}) {
243	dpavlin	274	print ", part bigger than media $size > $Conf{MaxArchiveSize}\n";
244			return 0;
245	dpavlin	254	}
246
247	dpavlin	270	print ", $size bytes";
248
249	dpavlin	264
250	dpavlin	281	open(my $fh, "gzip -cd $path \|") or die "can't open $path: $!";
251	dpavlin	254	binmode($fh);
252			my $tar = Archive::Tar::Streamed->new($fh);
253
254	dpavlin	281	my $tar_size_inarc = 0;
255	dpavlin	264	my $items = 0;
256	dpavlin	256
257	dpavlin	254	while(my $entry = $tar->next) {
258			push @tar_files, $entry->name;
259	dpavlin	264	$items++;
260	dpavlin	281	$tar_size_inarc += $entry->size;
261	dpavlin	269
262	dpavlin	281	if ($tar_size_inarc > $Conf{MaxArchiveFileSize}) {
263			print ", part $tarfilename is too big $tar_size_inarc > $Conf{MaxArchiveFileSize}\n";
264	dpavlin	274	return 0;
265	dpavlin	269	}
266
267	dpavlin	254	}
268	dpavlin	256
269	dpavlin	281	close($fh);
270
271	dpavlin	270	print ", $items items";
272
273	dpavlin	281	if ($tar_size_inarc == 0 && $items == 0) {
274	dpavlin	279	print ", EMPTY tar\n";
275
276			my $backup_id = get_backup_id($host, $share, $num);
277
278			my $sth_inc_deleted = $dbh->prepare(qq{
279			update backups set
280			inc_deleted = true
281			where id = ?
282			});
283			$sth_inc_deleted->execute($backup_id);
284	dpavlin	280
285			$dbh->commit;
286
287			return 1;
288	dpavlin	279	}
289
290	dpavlin	281	my $tar_size = get_gzip_size( $path );
291	dpavlin	279
292	dpavlin	281	# real tar size is bigger because of padding
293			if ($tar_size_inarc > $tar_size) {
294	dpavlin	282	print ", size of files in tar ($tar_size_inarc) bigger than whole tar ($tar_size)!\n";
295	dpavlin	281	return 0;
296			}
297	dpavlin	280
298	dpavlin	269	#
299			# check if md5 exists, and if not, create one
300			#
301
302			my $md5_path = $path;
303			$md5_path =~ s/\.tar\.gz$/.md5/ \|\| die "can't create md5 filename from $md5_path";
304			if (! -e $md5_path \|\| -z $md5_path) {
305			print ", creating md5";
306			system( $bin->{md5sum} . " $path > $md5_path") == 0 or die "can't create md5 $path: $!";
307	dpavlin	270	} else {
308			## FIXME check if existing md5 is valid
309	dpavlin	256	}
310	dpavlin	264
311	dpavlin	269	my $md5 = read_file( $md5_path ) \|\| die "can't read md5sum file $md5_path: $!";
312			$md5 =~ s#\s.*$##;
313
314			# extract part number from filename
315			my $part_nr = 1;
316			$part_nr = $1 if ($tarfilename =~ m#/(\d+)\.tar\.gz#);
317
318			#
319			# finally, check if backup_parts table in database is valid
320			#
321
322	dpavlin	264	check_part($host, $share, $num, $part_nr, $tar_size, $size, $md5, $items);
323	dpavlin	215	}
324	dpavlin	254
325	dpavlin	256	# short-cut and exit;
326			return $same unless($same);
327
328	dpavlin	215	@tar_files = sort @tar_files;
329	dpavlin	270	print "\n\t",($#tar_files + 1), " tar files";
330	dpavlin	214
331			my $sth = $dbh->prepare(qq{
332			SELECT path,type
333			FROM files
334			JOIN shares on shares.id = shareid
335			JOIN hosts on hosts.id = shares.hostid
336			WHERE hosts.name = ? and shares.name = ? and backupnum = ?
337			});
338			$sth->execute($host, $share, $num);
339			my @db_files;
340			while( my $row = $sth->fetchrow_hashref ) {
341
342			my $path = $row->{'path'} \|\| die "no path?";
343			$path =~ s#^/#./#;
344			$path .= '/' if ($row->{'type'} == BPC_FTYPE_DIR);
345			push @db_files, $path;
346			}
347
348	dpavlin	270	print " ",($#db_files + 1), " database files, diff";
349	dpavlin	214
350			@db_files = sort @db_files;
351
352			if ($#tar_files != $#db_files) {
353			$same = 0;
354	dpavlin	270	print " NUMBER";
355	dpavlin	214	} 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 $diff->Items(2);
362			}
363			}
364
365	dpavlin	270	print " ",($same ? 'ok' : 'DIFFERENT'),
366			", dur: ",fmt_time(time() - $t), "\n";
367	dpavlin	214
368			return $same;
369			}
370
371
372	dpavlin	157	#----- main
373
374			my $sth = $dbh->prepare( qq{
375
376			select
377	dpavlin	158	backups.id as backup_id,
378	dpavlin	157	hosts.name as host,
379			shares.name as share,
380	dpavlin	192	backups.num as num,
381	dpavlin	196	inc_size,
382			parts
383	dpavlin	157	from backups
384			join shares on backups.hostid = shares.hostid
385			and shares.id = backups.shareid
386			join hosts on shares.hostid = hosts.id
387	dpavlin	192	where not inc_deleted
388	dpavlin	157	order by backups.date
389
390			} );
391
392	dpavlin	213	$sth->execute();
393			my $num_backups = $sth->rows;
394			my $curr_backup = 1;
395
396	dpavlin	293	if ($opt{h}) {
397			warn "making increments just for host $opt{h}\n";
398			}
399
400	dpavlin	157	while (my $row = $sth->fetchrow_hashref) {
401	dpavlin	271
402	dpavlin	293	if ($opt{h} && $row->{host} ne $opt{h}) {
403			warn "skipped $row->{host}\n" if ($debug);
404			next;
405			}
406
407	dpavlin	271	$curr_backup++;
408
409	dpavlin	157	my $tar_file = BackupPC::SearchLib::getGzipName($row->{'host'}, $row->{'share'}, $row->{'num'});
410	dpavlin	192
411	dpavlin	194	# this will return -1 if file doesn't exist
412			my $size = BackupPC::SearchLib::get_tgz_size_by_name($tar_file);
413	dpavlin	192
414	dpavlin	253	print "# size: $size backup.size: ", $row->{inc_size},"\n" if ($opt{d});
415
416	dpavlin	266	if ( $row->{'inc_size'} != -1 && $size != -1 && $row->{'inc_size'} >= $size) {
417	dpavlin	256	if ($check) {
418			tar_check($row->{'host'}, $row->{'share'}, $row->{'num'}, $tar_file) && next;
419			} else {
420			next;
421			}
422	dpavlin	253	}
423
424	dpavlin	271	print curr_time, " creating $curr_backup/$num_backups ", $row->{'host'}, ":", $row->{'share'}, " #", $row->{'num'}, " -> $tar_file";
425	dpavlin	157
426			my $t = time();
427
428	dpavlin	194	# re-create archive?
429	dpavlin	254	my $cmd = qq{ $tarIncCreate -h "$row->{'host'}" -s "$row->{'share'}" -n $row->{'num'} -f };
430	dpavlin	253	print STDERR "## $cmd\n" if ($debug);
431	dpavlin	157
432	dpavlin	253	if (system($cmd) != 0) {
433			print STDERR " FAILED";
434	dpavlin	194	}
435	dpavlin	157
436	dpavlin	194	print ", dur: ",fmt_time(time() - $t), "\n";
437	dpavlin	158
438			$dbh->commit;
439
440	dpavlin	157	}
441
442			undef $sth;
443			$dbh->disconnect;