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;

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

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

# cludge: minimum .tar.gz size
my $MIN_TAR_SIZE = 80;

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());
}

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

        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";
        }

        my @tar_files;

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

        print " reading";

        foreach my $tarfilename (@tar_parts) {

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

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

                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);

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

        @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;

        my $same = 1;
        if ($#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 && ( $check && tar_check($row->{'host'}, $row->{'share'}, $row->{'num'}, $tar_file) || 1) ) {
                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	157
19	dpavlin	230	my $bpc = BackupPC::Lib->new \|\| die "can't create BackupPC::Lib";
20			my %Conf = $bpc->Conf();
21
22			use BackupPC::SearchLib;
23			%BackupPC::SearchLib::Conf = %Conf;
24
25	dpavlin	218	# cludge: minimum .tar.gz size
26			my $MIN_TAR_SIZE = 80;
27
28	dpavlin	157	my $path = abs_path($0);
29			$path =~ s#/[^/]+$#/#;
30			my $tarIncCreate = $path .= 'BackupPC_tarIncCreate';
31
32			die "can't find $tarIncCreate: $!\n" unless (-x $tarIncCreate);
33
34	dpavlin	201	my $bin;
35	dpavlin	254	foreach my $c (qw/gzip md5sum/) {
36	dpavlin	201	$bin->{$c} = which($c) \|\| die "$0 needs $c, install it\n";
37			}
38
39	dpavlin	214	my %opt;
40			getopts("cd", \%opt );
41	dpavlin	201
42	dpavlin	214	my $debug = $opt{d};
43			my $check = $opt{c} && print STDERR "NOTICE: tar archive check forced\n";
44
45	dpavlin	157	$\|=1;
46
47			my $start_t = time();
48
49			my $t_fmt = '%Y-%m-%d %H:%M:%S';
50
51			my $dsn = $Conf{SearchDSN} \|\| die "Need SearchDSN in config.pl\n";
52			my $user = $Conf{SearchUser} \|\| '';
53
54			my $dbh = DBI->connect($dsn, $user, "", { RaiseError => 1, AutoCommit => 0 });
55
56			my $tar_dir = $Conf{InstallDir}.'/'.$Conf{GzipTempDir};
57
58			die "problem with $tar_dir, check GzipTempDir in configuration\n" unless (-d $tar_dir && -w $tar_dir);
59
60			#---- subs ----
61
62			sub fmt_time {
63			my $t = shift \|\| return;
64			my $out = "";
65			my ($ss,$mm,$hh) = gmtime($t);
66			$out .= "${hh}h" if ($hh);
67			$out .= sprintf("%02d:%02d", $mm,$ss);
68			return $out;
69			}
70
71			sub curr_time {
72			return strftime($t_fmt,localtime());
73			}
74
75	dpavlin	214	sub tar_check($$$$) {
76			my ($host,$share,$num,$filename) = @_;
77
78			if ($debug) {
79			print STDERR " {{ CHECK: ${host}:${share}#${num} and $filename";
80			} else {
81			print " check";
82			}
83
84	dpavlin	254	my @tar_parts;
85
86			if (-d "$tar_dir/$filename") {
87			print STDERR " multi-part" if ($opt{d});
88			opendir(my $dir, "$tar_dir/$filename") \|\| die "can't readdir $tar_dir/$filename: $!";
89			@tar_parts = map { my $p = $_; $p =~ s#^#${filename}/#; $p } grep { !/^\./ && !/md5/ && -f "$tar_dir/$filename/$_" } readdir($dir);
90			closedir($dir);
91			} else {
92			push @tar_parts, "${filename}.tar.gz";
93	dpavlin	214	}
94
95	dpavlin	254	my @tar_files;
96	dpavlin	214
97	dpavlin	254	print " [parts: ",join(", ", @tar_parts),"]" if ($opt{d});
98
99			print " reading";
100
101			foreach my $tarfilename (@tar_parts) {
102
103			print STDERR " $tarfilename" if ($debug);
104
105			my $path = "$tar_dir/$tarfilename";
106			my $md5 = $path;
107			$md5 =~ s/\.tar\.gz$/.md5/ \|\| die "can't create md5 filename from $md5";
108			if (! -e $md5) {
109			print ", creating md5";
110			system( $bin->{md5sum} . " $path > $md5") == 0 or die "can't create md5 $path: $!";
111			}
112
113			open(my $fh, "gzip -cd $tar_dir/$tarfilename \|") or die "can't open $tar_dir/$tarfilename: $!";
114			binmode($fh);
115			my $tar = Archive::Tar::Streamed->new($fh);
116
117			while(my $entry = $tar->next) {
118			push @tar_files, $entry->name;
119			}
120	dpavlin	215	}
121	dpavlin	254
122	dpavlin	215	@tar_files = sort @tar_files;
123	dpavlin	214	print STDERR " ",($#tar_files + 1), " files" if ($debug);
124
125			print STDERR ", database" if ($debug);
126
127			my $sth = $dbh->prepare(qq{
128			SELECT path,type
129			FROM files
130			JOIN shares on shares.id = shareid
131			JOIN hosts on hosts.id = shares.hostid
132			WHERE hosts.name = ? and shares.name = ? and backupnum = ?
133			});
134			$sth->execute($host, $share, $num);
135			my @db_files;
136			while( my $row = $sth->fetchrow_hashref ) {
137
138			my $path = $row->{'path'} \|\| die "no path?";
139			$path =~ s#^/#./#;
140			$path .= '/' if ($row->{'type'} == BPC_FTYPE_DIR);
141			push @db_files, $path;
142			}
143
144			print STDERR " ",($#db_files + 1), " files, diff" if ($debug);
145
146			@db_files = sort @db_files;
147
148			my $same = 1;
149			if ($#tar_files != $#db_files) {
150			$same = 0;
151			print STDERR " NUMBER" if ($debug);
152			} else {
153			my $diff = Algorithm::Diff->new(\@tar_files, \@db_files);
154			while ( $diff->Next() ) {
155			next if $diff->Same();
156			$same = 0;
157			print "< $_\n" for $diff->Items(1);
158			print "> $_\n" for $diff->Items(2);
159			}
160			}
161
162	dpavlin	216	print " ",($same ? 'ok' : 'DIFFERENT');
163	dpavlin	214	print STDERR " }} " if ($debug);
164
165			return $same;
166			}
167
168
169	dpavlin	157	#----- main
170
171			my $sth = $dbh->prepare( qq{
172
173			select
174	dpavlin	158	backups.id as backup_id,
175	dpavlin	157	hosts.name as host,
176			shares.name as share,
177	dpavlin	192	backups.num as num,
178	dpavlin	196	inc_size,
179			parts
180	dpavlin	157	from backups
181			join shares on backups.hostid = shares.hostid
182			and shares.id = backups.shareid
183			join hosts on shares.hostid = hosts.id
184	dpavlin	192	where not inc_deleted
185	dpavlin	157	order by backups.date
186
187			} );
188
189	dpavlin	213	$sth->execute();
190			my $num_backups = $sth->rows;
191			my $curr_backup = 1;
192
193	dpavlin	157	while (my $row = $sth->fetchrow_hashref) {
194			my $tar_file = BackupPC::SearchLib::getGzipName($row->{'host'}, $row->{'share'}, $row->{'num'});
195	dpavlin	192
196	dpavlin	194	# this will return -1 if file doesn't exist
197			my $size = BackupPC::SearchLib::get_tgz_size_by_name($tar_file);
198	dpavlin	192
199	dpavlin	253	print "# size: $size backup.size: ", $row->{inc_size},"\n" if ($opt{d});
200
201	dpavlin	254	if ( $row->{'inc_size'} != -1 && $size != -1 && $row->{'inc_size'} == $size && ( $check && tar_check($row->{'host'}, $row->{'share'}, $row->{'num'}, $tar_file) \|\| 1) ) {
202	dpavlin	253	next;
203			}
204
205	dpavlin	213	print curr_time, " $curr_backup/$num_backups ", $row->{'host'}, ":", $row->{'share'}, " #", $row->{'num'}, " -> $tar_file";
206			$curr_backup++;
207	dpavlin	157
208			my $t = time();
209
210	dpavlin	194	# re-create archive?
211	dpavlin	254	my $cmd = qq{ $tarIncCreate -h "$row->{'host'}" -s "$row->{'share'}" -n $row->{'num'} -f };
212	dpavlin	253	print STDERR "## $cmd\n" if ($debug);
213	dpavlin	157
214	dpavlin	253	if (system($cmd) != 0) {
215			print STDERR " FAILED";
216	dpavlin	194	}
217	dpavlin	157
218	dpavlin	194	print ", dur: ",fmt_time(time() - $t), "\n";
219	dpavlin	158
220			$dbh->commit;
221
222	dpavlin	157	}
223
224			undef $sth;
225			$dbh->disconnect;