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;

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";
}

my %opt;
getopts("cd", \%opt );

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

        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 STDERR "# deleting invalid row $row->{id}\n" if ($opt{d});
                        $dbh->do(qq{ delete from backup_parts where id = $row->{id} });
                }
                print STDERR "# inserting new backup_part row\n";
                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;
        }

        if ($debug) {
                print STDERR " {{ CHECK: ${host}:${share}#${num} and $filename";
        } else {
                print " check";
        }

        my @tar_parts;

        if (-d "$tar_dir/$filename") {
                print STDERR " multi-part" if ($opt{d});
                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";

        foreach my $tarfilename (@tar_parts) {

                print STDERR " $tarfilename" if ($debug);

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

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

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

                my $size = (stat( "$tar_dir/$tarfilename" ))[7] || die "can't stat $tar_dir/$tarfilename";

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

                my $tar_size = 0;
                my $items = 0;

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

                if ($tar_size > $Conf{MaxArchiveFileSize}) {
                        print STDERR " part too big $tar_size > $Conf{MaxArchiveFileSize} }}" if ($debug);
                        $same = 0;
                        last;
                } elsif ($size > $Conf{MaxArchiveSize}) {
                        print STDERR " part bigger than media $size > $Conf{MaxArchiveSize} }}" if ($debug);
                        $same = 0;
                        last;
                }

                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 STDERR " ",($#tar_files + 1), " files" if ($debug);

        print STDERR ", database" if ($debug);

        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 STDERR " ",($#db_files + 1), " files, diff" if ($debug);

        @db_files = sort @db_files;

        if ($#tar_files != $#db_files) {
                $same = 0;
                print STDERR " NUMBER" if ($debug);
        } 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 " ",($same ? 'ok' : 'DIFFERENT');
        print STDERR " }} " if ($debug);

        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;

while (my $row = $sth->fetchrow_hashref) {
        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, " $curr_backup/$num_backups ", $row->{'host'}, ":", $row->{'share'}, " #", $row->{'num'}, " -> $tar_file";
        $curr_backup++;

        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	157
20	dpavlin	230	my $bpc = BackupPC::Lib->new \|\| die "can't create BackupPC::Lib";
21			my %Conf = $bpc->Conf();
22
23			use BackupPC::SearchLib;
24			%BackupPC::SearchLib::Conf = %Conf;
25
26	dpavlin	157	my $path = abs_path($0);
27			$path =~ s#/[^/]+$#/#;
28			my $tarIncCreate = $path .= 'BackupPC_tarIncCreate';
29
30			die "can't find $tarIncCreate: $!\n" unless (-x $tarIncCreate);
31
32	dpavlin	201	my $bin;
33	dpavlin	254	foreach my $c (qw/gzip md5sum/) {
34	dpavlin	201	$bin->{$c} = which($c) \|\| die "$0 needs $c, install it\n";
35			}
36
37	dpavlin	214	my %opt;
38			getopts("cd", \%opt );
39	dpavlin	201
40	dpavlin	214	my $debug = $opt{d};
41			my $check = $opt{c} && print STDERR "NOTICE: tar archive check forced\n";
42
43	dpavlin	157	$\|=1;
44
45			my $start_t = time();
46
47			my $t_fmt = '%Y-%m-%d %H:%M:%S';
48
49			my $dsn = $Conf{SearchDSN} \|\| die "Need SearchDSN in config.pl\n";
50			my $user = $Conf{SearchUser} \|\| '';
51
52			my $dbh = DBI->connect($dsn, $user, "", { RaiseError => 1, AutoCommit => 0 });
53
54			my $tar_dir = $Conf{InstallDir}.'/'.$Conf{GzipTempDir};
55
56			die "problem with $tar_dir, check GzipTempDir in configuration\n" unless (-d $tar_dir && -w $tar_dir);
57
58			#---- subs ----
59
60			sub fmt_time {
61			my $t = shift \|\| return;
62			my $out = "";
63			my ($ss,$mm,$hh) = gmtime($t);
64			$out .= "${hh}h" if ($hh);
65			$out .= sprintf("%02d:%02d", $mm,$ss);
66			return $out;
67			}
68
69			sub curr_time {
70			return strftime($t_fmt,localtime());
71			}
72
73	dpavlin	264	my $hsn_cache;
74
75			sub get_backup_id($$$) {
76			my ($host, $share, $num) = @_;
77
78			my $key = "$host $share $num";
79			return $hsn_cache->{$key} if ($hsn_cache->{$key});
80
81			my $sth = $dbh->prepare(qq{
82			SELECT
83			backups.id
84			FROM backups
85			INNER JOIN shares ON backups.shareID=shares.ID
86			INNER JOIN hosts ON backups.hostID = hosts.ID
87			where hosts.name = ? and shares.name = ? and backups.num = ?
88			});
89			$sth->execute($host, $share, $num);
90			my ($id) = $sth->fetchrow_array;
91
92			$hsn_cache->{"$host $share $num"} = $id;
93
94			print STDERR "# $host $share $num == $id\n" if ($opt{d});
95
96			return $id;
97			}
98
99
100	dpavlin	214	sub tar_check($$$$) {
101			my ($host,$share,$num,$filename) = @_;
102
103	dpavlin	264	sub check_part {
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{
107			select
108			id, tar_size, size, md5, items
109			from backup_parts
110			where backup_id = ? and part_nr = ?
111			});
112
113			$sth_md5->execute($backup_id, $part_nr);
114
115			if (my $row = $sth_md5->fetchrow_hashref) {
116			return if (
117	dpavlin	265	$row->{tar_size} >= $tar_size &&
118	dpavlin	264	$row->{size} == $size &&
119			$row->{md5} eq $md5 &&
120			$row->{items} == $items
121			);
122			print STDERR "# deleting invalid row $row->{id}\n" if ($opt{d});
123			$dbh->do(qq{ delete from backup_parts where id = $row->{id} });
124			}
125			print STDERR "# inserting new backup_part row\n";
126			my $sth_insert = $dbh->prepare(qq{
127			insert into backup_parts (
128			backup_id,
129			part_nr,
130			tar_size,
131			size,
132			md5,
133			items
134			) values (?,?,?,?,?,?)
135			});
136
137			$sth_insert->execute($backup_id, $part_nr, $tar_size, $size, $md5, $items);
138			$dbh->commit;
139			}
140
141	dpavlin	214	if ($debug) {
142			print STDERR " {{ CHECK: ${host}:${share}#${num} and $filename";
143			} else {
144			print " check";
145			}
146
147	dpavlin	254	my @tar_parts;
148
149			if (-d "$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($dir);
153			closedir($dir);
154			} else {
155			push @tar_parts, "${filename}.tar.gz";
156	dpavlin	214	}
157
158	dpavlin	256	print " [parts: ",join(", ", @tar_parts),"]" if ($opt{d});
159
160			my $same = 1;
161	dpavlin	254	my @tar_files;
162	dpavlin	214
163	dpavlin	264	my $backup_part;
164
165	dpavlin	254	print " reading";
166
167			foreach my $tarfilename (@tar_parts) {
168
169			print STDERR " $tarfilename" if ($debug);
170
171			my $path = "$tar_dir/$tarfilename";
172	dpavlin	264	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	dpavlin	254	print ", creating md5";
176	dpavlin	264	system( $bin->{md5sum} . " $path > $md5_path") == 0 or die "can't create md5 $path: $!";
177	dpavlin	254	}
178
179	dpavlin	264	my $md5 = read_file( $md5_path ) \|\| die "can't read md5sum file $md5_path: $!";
180	dpavlin	265	$md5 =~ s#\s.*$##;
181	dpavlin	264
182			my $part_nr = 1;
183			$part_nr = $1 if ($tarfilename =~ m#/(\d+)\.tar\.gz#);
184
185			my $size = (stat( "$tar_dir/$tarfilename" ))[7] \|\| die "can't stat $tar_dir/$tarfilename";
186
187	dpavlin	254	open(my $fh, "gzip -cd $tar_dir/$tarfilename \|") or die "can't open $tar_dir/$tarfilename: $!";
188			binmode($fh);
189			my $tar = Archive::Tar::Streamed->new($fh);
190
191	dpavlin	264	my $tar_size = 0;
192			my $items = 0;
193	dpavlin	256
194	dpavlin	254	while(my $entry = $tar->next) {
195			push @tar_files, $entry->name;
196	dpavlin	264	$items++;
197			$tar_size += $entry->size;
198	dpavlin	254	}
199	dpavlin	256
200	dpavlin	264	if ($tar_size > $Conf{MaxArchiveFileSize}) {
201			print STDERR " part too big $tar_size > $Conf{MaxArchiveFileSize} }}" if ($debug);
202	dpavlin	256	$same = 0;
203			last;
204	dpavlin	264	} elsif ($size > $Conf{MaxArchiveSize}) {
205			print STDERR " part bigger than media $size > $Conf{MaxArchiveSize} }}" if ($debug);
206	dpavlin	256	$same = 0;
207			last;
208			}
209	dpavlin	264
210			check_part($host, $share, $num, $part_nr, $tar_size, $size, $md5, $items);
211	dpavlin	215	}
212	dpavlin	254
213	dpavlin	256	# short-cut and exit;
214			return $same unless($same);
215
216	dpavlin	215	@tar_files = sort @tar_files;
217	dpavlin	214	print STDERR " ",($#tar_files + 1), " files" if ($debug);
218
219			print STDERR ", database" if ($debug);
220
221			my $sth = $dbh->prepare(qq{
222			SELECT path,type
223			FROM files
224			JOIN shares on shares.id = shareid
225			JOIN hosts on hosts.id = shares.hostid
226			WHERE hosts.name = ? and shares.name = ? and backupnum = ?
227			});
228			$sth->execute($host, $share, $num);
229			my @db_files;
230			while( my $row = $sth->fetchrow_hashref ) {
231
232			my $path = $row->{'path'} \|\| die "no path?";
233			$path =~ s#^/#./#;
234			$path .= '/' if ($row->{'type'} == BPC_FTYPE_DIR);
235			push @db_files, $path;
236			}
237
238			print STDERR " ",($#db_files + 1), " files, diff" if ($debug);
239
240			@db_files = sort @db_files;
241
242			if ($#tar_files != $#db_files) {
243			$same = 0;
244			print STDERR " NUMBER" if ($debug);
245			} else {
246			my $diff = Algorithm::Diff->new(\@tar_files, \@db_files);
247			while ( $diff->Next() ) {
248			next if $diff->Same();
249			$same = 0;
250			print "< $_\n" for $diff->Items(1);
251			print "> $_\n" for $diff->Items(2);
252			}
253			}
254
255	dpavlin	216	print " ",($same ? 'ok' : 'DIFFERENT');
256	dpavlin	214	print STDERR " }} " if ($debug);
257
258			return $same;
259			}
260
261
262	dpavlin	157	#----- main
263
264			my $sth = $dbh->prepare( qq{
265
266			select
267	dpavlin	158	backups.id as backup_id,
268	dpavlin	157	hosts.name as host,
269			shares.name as share,
270	dpavlin	192	backups.num as num,
271	dpavlin	196	inc_size,
272			parts
273	dpavlin	157	from backups
274			join shares on backups.hostid = shares.hostid
275			and shares.id = backups.shareid
276			join hosts on shares.hostid = hosts.id
277	dpavlin	192	where not inc_deleted
278	dpavlin	157	order by backups.date
279
280			} );
281
282	dpavlin	213	$sth->execute();
283			my $num_backups = $sth->rows;
284			my $curr_backup = 1;
285
286	dpavlin	157	while (my $row = $sth->fetchrow_hashref) {
287			my $tar_file = BackupPC::SearchLib::getGzipName($row->{'host'}, $row->{'share'}, $row->{'num'});
288	dpavlin	192
289	dpavlin	194	# this will return -1 if file doesn't exist
290			my $size = BackupPC::SearchLib::get_tgz_size_by_name($tar_file);
291	dpavlin	192
292	dpavlin	253	print "# size: $size backup.size: ", $row->{inc_size},"\n" if ($opt{d});
293
294	dpavlin	256	if ( $row->{'inc_size'} != -1 && $size != -1 && $row->{'inc_size'} == $size) {
295			if ($check) {
296			tar_check($row->{'host'}, $row->{'share'}, $row->{'num'}, $tar_file) && next;
297			} else {
298			next;
299			}
300	dpavlin	253	}
301
302	dpavlin	213	print curr_time, " $curr_backup/$num_backups ", $row->{'host'}, ":", $row->{'share'}, " #", $row->{'num'}, " -> $tar_file";
303			$curr_backup++;
304	dpavlin	157
305			my $t = time();
306
307	dpavlin	194	# re-create archive?
308	dpavlin	254	my $cmd = qq{ $tarIncCreate -h "$row->{'host'}" -s "$row->{'share'}" -n $row->{'num'} -f };
309	dpavlin	253	print STDERR "## $cmd\n" if ($debug);
310	dpavlin	157
311	dpavlin	253	if (system($cmd) != 0) {
312			print STDERR " FAILED";
313	dpavlin	194	}
314	dpavlin	157
315	dpavlin	194	print ", dur: ",fmt_time(time() - $t), "\n";
316	dpavlin	158
317			$dbh->commit;
318
319	dpavlin	157	}
320
321			undef $sth;
322			$dbh->disconnect;