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/) {
        $bin->{$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) = @_;

        return 1;       # FIXME

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

        if (-d $filename) {
                print STDERR ", joining";
                tar_join($filename);
        }

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

        print STDERR ", tar" if ($debug);
        my @tar_files;
        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_dir/$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'} };
        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	253	foreach my $c (qw/gzip/) {
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	dpavlin	253	return 1; # FIXME
79
80	dpavlin	214	if ($debug) {
81			print STDERR " {{ CHECK: ${host}:${share}#${num} and $filename";
82			} else {
83			print " check";
84			}
85
86			if (-d $filename) {
87			print STDERR ", joining";
88			tar_join($filename);
89			}
90
91			print STDERR ", opening" if ($debug);
92	dpavlin	215	open(my $fh, "gzip -cd $filename \|") or die "can't open $filename: $!";
93			binmode($fh);
94			my $tar = Archive::Tar::Streamed->new($fh);
95	dpavlin	214
96			print STDERR ", tar" if ($debug);
97	dpavlin	215	my @tar_files;
98			while(my $entry = $tar->next) {
99			push @tar_files, $entry->name;
100			}
101			@tar_files = sort @tar_files;
102	dpavlin	214	print STDERR " ",($#tar_files + 1), " files" if ($debug);
103
104			print STDERR ", database" if ($debug);
105
106			my $sth = $dbh->prepare(qq{
107			SELECT path,type
108			FROM files
109			JOIN shares on shares.id = shareid
110			JOIN hosts on hosts.id = shares.hostid
111			WHERE hosts.name = ? and shares.name = ? and backupnum = ?
112			});
113			$sth->execute($host, $share, $num);
114			my @db_files;
115			while( my $row = $sth->fetchrow_hashref ) {
116
117			my $path = $row->{'path'} \|\| die "no path?";
118			$path =~ s#^/#./#;
119			$path .= '/' if ($row->{'type'} == BPC_FTYPE_DIR);
120			push @db_files, $path;
121			}
122
123			print STDERR " ",($#db_files + 1), " files, diff" if ($debug);
124
125			@db_files = sort @db_files;
126
127			my $same = 1;
128			if ($#tar_files != $#db_files) {
129			$same = 0;
130			print STDERR " NUMBER" if ($debug);
131			} else {
132			my $diff = Algorithm::Diff->new(\@tar_files, \@db_files);
133			while ( $diff->Next() ) {
134			next if $diff->Same();
135			$same = 0;
136			print "< $_\n" for $diff->Items(1);
137			print "> $_\n" for $diff->Items(2);
138			}
139			}
140
141	dpavlin	216	print " ",($same ? 'ok' : 'DIFFERENT');
142	dpavlin	214	print STDERR " }} " if ($debug);
143
144			return $same;
145			}
146
147
148	dpavlin	157	#----- main
149
150			my $sth = $dbh->prepare( qq{
151
152			select
153	dpavlin	158	backups.id as backup_id,
154	dpavlin	157	hosts.name as host,
155			shares.name as share,
156	dpavlin	192	backups.num as num,
157	dpavlin	196	inc_size,
158			parts
159	dpavlin	157	from backups
160			join shares on backups.hostid = shares.hostid
161			and shares.id = backups.shareid
162			join hosts on shares.hostid = hosts.id
163	dpavlin	192	where not inc_deleted
164	dpavlin	157	order by backups.date
165
166			} );
167
168	dpavlin	213	$sth->execute();
169			my $num_backups = $sth->rows;
170			my $curr_backup = 1;
171
172	dpavlin	157	while (my $row = $sth->fetchrow_hashref) {
173			my $tar_file = BackupPC::SearchLib::getGzipName($row->{'host'}, $row->{'share'}, $row->{'num'});
174	dpavlin	192
175	dpavlin	194	# this will return -1 if file doesn't exist
176			my $size = BackupPC::SearchLib::get_tgz_size_by_name($tar_file);
177	dpavlin	192
178	dpavlin	253	print "# size: $size backup.size: ", $row->{inc_size},"\n" if ($opt{d});
179
180			if ( $row->{'inc_size'} != -1 && $size != -1 && $row->{'inc_size'} == $size && ( $check && tar_check($row->{'host'}, $row->{'share'}, $row->{'num'}, "$tar_dir/$tar_file") \|\| 1) ) {
181			next;
182			}
183
184	dpavlin	213	print curr_time, " $curr_backup/$num_backups ", $row->{'host'}, ":", $row->{'share'}, " #", $row->{'num'}, " -> $tar_file";
185			$curr_backup++;
186	dpavlin	157
187			my $t = time();
188
189	dpavlin	194	# re-create archive?
190	dpavlin	253	my $cmd = qq{ $tarIncCreate -h "$row->{'host'}" -s "$row->{'share'}" -n $row->{'num'} };
191			print STDERR "## $cmd\n" if ($debug);
192	dpavlin	157
193	dpavlin	253	if (system($cmd) != 0) {
194			print STDERR " FAILED";
195	dpavlin	194	}
196	dpavlin	157
197	dpavlin	194	print ", dur: ",fmt_time(time() - $t), "\n";
198	dpavlin	158
199			$dbh->commit;
200
201	dpavlin	157	}
202
203			undef $sth;
204			$dbh->disconnect;