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 BackupPC::SearchLib;
use Cwd qw/abs_path/;
use File::Which;
use Archive::Tar::Streamed;
use Algorithm::Diff;
use Getopt::Std;

# 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 split/) {
        $bin->{$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 $hosts;
my $bpc = BackupPC::Lib->new || die;
my %Conf = $bpc->Conf();
my $TopDir = $bpc->TopDir();
my $beenThere = {};

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_join($) {
        my $filename = shift;

        my $in = my $out = $filename;
        $out .= '.tmp';

        # FIXME I should really order parts manually!
        system("cat $in/part* > $out && rm -Rf $in && mv $out $in") == 0 or die "can't join $in: $?";

}

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

        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

} );

my $sth_inc_size = $dbh->prepare(qq{ update backups set inc_size = ?, parts = ? where id = ? });
my $sth_inc_deleted = $dbh->prepare(qq{ update backups set inc_deleted = ? where id = ? });

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

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

        my $t = time();

        # re-create archive?
        if ($row->{'inc_size'} == -1 || $size == -1 ||
                $row->{'inc_size'} != $size ||
                $check && ! tar_check($row->{'host'}, $row->{'share'}, $row->{'num'}, "$tar_dir/$tar_file")
        ) {
                my $cmd = qq{rm -Rf $tar_dir/$tar_file && $tarIncCreate -h "$row->{'host'}" -s "$row->{'share'}" -n $row->{'num'} | $bin->{'gzip'} $Conf{GzipLevel} > ${tar_dir}/${tar_file}.tmp};
                print STDERR "## $cmd\n" if ($debug);

                system($cmd) == 0 or die "failed: $?";

                rename("${tar_dir}/${tar_file}.tmp", "$tar_dir/$tar_file") or die "can't rename $tar_dir/$tar_file: $!";

                $size = (stat( "$tar_dir/$tar_file" ))[7];
        }

        if ($size > $MIN_TAR_SIZE) {

                my $max_size = $Conf{'MaxArchiveSize'} || die "problem with MaxArchieSize parametar";
                $max_size *= 1024;      # convert to bytes

                my $parts = int( ($size + $max_size - 1) / $max_size );

                if (-d "$tar_dir/$tar_file" && $parts != $row->{'parts'}) {
                        print " join";
                        tar_join("$tar_dir/$tar_file");
                }

                if ($size > $max_size && ! -d "$tar_dir/$tar_file") {
                        print " split/$parts";
                        my $in = my $out = "$tar_dir/$tar_file";
                        $out .= '.tmp';
                        rename $in, $out || die "can't rename $in: $!";
                        mkdir $in || die "can't mkdir $in: $!";

                        my $suffix_len = length("$parts");
                        system("$bin->{'split'} -d -b $max_size -a $suffix_len $out $in/part") == 0 or die "can't split $out: $?";
                        unlink $out || die "can't unlink $out: $!";
                }

                $sth_inc_size->execute($size, $parts, $row->{'backup_id'});
                $sth_inc_deleted->execute(0, $row->{'backup_id'});

                printf(" %1.2f MB", ($size / 1024 / 1024));

        } else {
                $sth_inc_deleted->execute(1, $row->{'backup_id'});
                unlink "$tar_dir/$tar_file" || die "can't delete $tar_dir/$tar_file: $!\n";
                print " EMPTY";
        }
        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 BackupPC::SearchLib;
14			use Cwd qw/abs_path/;
15	dpavlin	201	use File::Which;
16	dpavlin	215	use Archive::Tar::Streamed;
17	dpavlin	214	use Algorithm::Diff;
18			use Getopt::Std;
19	dpavlin	157
20	dpavlin	218	# cludge: minimum .tar.gz size
21			my $MIN_TAR_SIZE = 80;
22
23	dpavlin	157	my $path = abs_path($0);
24			$path =~ s#/[^/]+$#/#;
25			my $tarIncCreate = $path .= 'BackupPC_tarIncCreate';
26
27			die "can't find $tarIncCreate: $!\n" unless (-x $tarIncCreate);
28
29	dpavlin	201	my $bin;
30			foreach my $c (qw/gzip split/) {
31			$bin->{$c} = which($c) \|\| die "$0 needs $c, install it\n";
32			}
33
34	dpavlin	214	my %opt;
35			getopts("cd", \%opt );
36	dpavlin	201
37	dpavlin	214	my $debug = $opt{d};
38			my $check = $opt{c} && print STDERR "NOTICE: tar archive check forced\n";
39
40	dpavlin	157	$\|=1;
41
42			my $start_t = time();
43
44			my $t_fmt = '%Y-%m-%d %H:%M:%S';
45
46			my $hosts;
47			my $bpc = BackupPC::Lib->new \|\| die;
48			my %Conf = $bpc->Conf();
49			my $TopDir = $bpc->TopDir();
50			my $beenThere = {};
51
52			my $dsn = $Conf{SearchDSN} \|\| die "Need SearchDSN in config.pl\n";
53			my $user = $Conf{SearchUser} \|\| '';
54
55			my $dbh = DBI->connect($dsn, $user, "", { RaiseError => 1, AutoCommit => 0 });
56
57			my $tar_dir = $Conf{InstallDir}.'/'.$Conf{GzipTempDir};
58
59			die "problem with $tar_dir, check GzipTempDir in configuration\n" unless (-d $tar_dir && -w $tar_dir);
60
61			#---- subs ----
62
63			sub fmt_time {
64			my $t = shift \|\| return;
65			my $out = "";
66			my ($ss,$mm,$hh) = gmtime($t);
67			$out .= "${hh}h" if ($hh);
68			$out .= sprintf("%02d:%02d", $mm,$ss);
69			return $out;
70			}
71
72			sub curr_time {
73			return strftime($t_fmt,localtime());
74			}
75
76	dpavlin	214	sub tar_join($) {
77			my $filename = shift;
78
79			my $in = my $out = $filename;
80			$out .= '.tmp';
81
82			# FIXME I should really order parts manually!
83			system("cat $in/part* > $out && rm -Rf $in && mv $out $in") == 0 or die "can't join $in: $?";
84
85			}
86
87			sub tar_check($$$$) {
88			my ($host,$share,$num,$filename) = @_;
89
90			if ($debug) {
91			print STDERR " {{ CHECK: ${host}:${share}#${num} and $filename";
92			} else {
93			print " check";
94			}
95
96			if (-d $filename) {
97			print STDERR ", joining";
98			tar_join($filename);
99			}
100
101			print STDERR ", opening" if ($debug);
102	dpavlin	215	open(my $fh, "gzip -cd $filename \|") or die "can't open $filename: $!";
103			binmode($fh);
104			my $tar = Archive::Tar::Streamed->new($fh);
105	dpavlin	214
106			print STDERR ", tar" if ($debug);
107	dpavlin	215	my @tar_files;
108			while(my $entry = $tar->next) {
109			push @tar_files, $entry->name;
110			}
111			@tar_files = sort @tar_files;
112	dpavlin	214	print STDERR " ",($#tar_files + 1), " files" if ($debug);
113
114			print STDERR ", database" if ($debug);
115
116			my $sth = $dbh->prepare(qq{
117			SELECT path,type
118			FROM files
119			JOIN shares on shares.id = shareid
120			JOIN hosts on hosts.id = shares.hostid
121			WHERE hosts.name = ? and shares.name = ? and backupnum = ?
122			});
123			$sth->execute($host, $share, $num);
124			my @db_files;
125			while( my $row = $sth->fetchrow_hashref ) {
126
127			my $path = $row->{'path'} \|\| die "no path?";
128			$path =~ s#^/#./#;
129			$path .= '/' if ($row->{'type'} == BPC_FTYPE_DIR);
130			push @db_files, $path;
131			}
132
133			print STDERR " ",($#db_files + 1), " files, diff" if ($debug);
134
135			@db_files = sort @db_files;
136
137			my $same = 1;
138			if ($#tar_files != $#db_files) {
139			$same = 0;
140			print STDERR " NUMBER" if ($debug);
141			} else {
142			my $diff = Algorithm::Diff->new(\@tar_files, \@db_files);
143			while ( $diff->Next() ) {
144			next if $diff->Same();
145			$same = 0;
146			print "< $_\n" for $diff->Items(1);
147			print "> $_\n" for $diff->Items(2);
148			}
149			}
150
151	dpavlin	216	print " ",($same ? 'ok' : 'DIFFERENT');
152	dpavlin	214	print STDERR " }} " if ($debug);
153
154			return $same;
155			}
156
157
158	dpavlin	157	#----- main
159
160			my $sth = $dbh->prepare( qq{
161
162			select
163	dpavlin	158	backups.id as backup_id,
164	dpavlin	157	hosts.name as host,
165			shares.name as share,
166	dpavlin	192	backups.num as num,
167	dpavlin	196	inc_size,
168			parts
169	dpavlin	157	from backups
170			join shares on backups.hostid = shares.hostid
171			and shares.id = backups.shareid
172			join hosts on shares.hostid = hosts.id
173	dpavlin	192	where not inc_deleted
174	dpavlin	157	order by backups.date
175
176			} );
177
178	dpavlin	196	my $sth_inc_size = $dbh->prepare(qq{ update backups set inc_size = ?, parts = ? where id = ? });
179	dpavlin	163	my $sth_inc_deleted = $dbh->prepare(qq{ update backups set inc_deleted = ? where id = ? });
180	dpavlin	158
181	dpavlin	157	%BackupPC::SearchLib::Conf = %Conf;
182
183	dpavlin	213	$sth->execute();
184			my $num_backups = $sth->rows;
185			my $curr_backup = 1;
186
187	dpavlin	157	while (my $row = $sth->fetchrow_hashref) {
188			my $tar_file = BackupPC::SearchLib::getGzipName($row->{'host'}, $row->{'share'}, $row->{'num'});
189	dpavlin	192
190	dpavlin	194	# this will return -1 if file doesn't exist
191			my $size = BackupPC::SearchLib::get_tgz_size_by_name($tar_file);
192	dpavlin	192
193	dpavlin	213	print curr_time, " $curr_backup/$num_backups ", $row->{'host'}, ":", $row->{'share'}, " #", $row->{'num'}, " -> $tar_file";
194			$curr_backup++;
195	dpavlin	157
196			my $t = time();
197
198	dpavlin	194	# re-create archive?
199	dpavlin	214	if ($row->{'inc_size'} == -1 \|\| $size == -1 \|\|
200			$row->{'inc_size'} != $size \|\|
201			$check && ! tar_check($row->{'host'}, $row->{'share'}, $row->{'num'}, "$tar_dir/$tar_file")
202			) {
203	dpavlin	212	my $cmd = qq{rm -Rf $tar_dir/$tar_file && $tarIncCreate -h "$row->{'host'}" -s "$row->{'share'}" -n $row->{'num'} \| $bin->{'gzip'} $Conf{GzipLevel} > ${tar_dir}/${tar_file}.tmp};
204	dpavlin	194	print STDERR "## $cmd\n" if ($debug);
205	dpavlin	157
206	dpavlin	194	system($cmd) == 0 or die "failed: $?";
207	dpavlin	212
208	dpavlin	213	rename("${tar_dir}/${tar_file}.tmp", "$tar_dir/$tar_file") or die "can't rename $tar_dir/$tar_file: $!";
209	dpavlin	212
210	dpavlin	194	$size = (stat( "$tar_dir/$tar_file" ))[7];
211			}
212	dpavlin	157
213	dpavlin	218	if ($size > $MIN_TAR_SIZE) {
214	dpavlin	157
215	dpavlin	194	my $max_size = $Conf{'MaxArchiveSize'} \|\| die "problem with MaxArchieSize parametar";
216	dpavlin	196	$max_size *= 1024; # convert to bytes
217	dpavlin	158
218	dpavlin	196	my $parts = int( ($size + $max_size - 1) / $max_size );
219
220			if (-d "$tar_dir/$tar_file" && $parts != $row->{'parts'}) {
221			print " join";
222	dpavlin	214	tar_join("$tar_dir/$tar_file");
223	dpavlin	196	}
224
225	dpavlin	194	if ($size > $max_size && ! -d "$tar_dir/$tar_file") {
226	dpavlin	196	print " split/$parts";
227	dpavlin	194	my $in = my $out = "$tar_dir/$tar_file";
228			$out .= '.tmp';
229			rename $in, $out \|\| die "can't rename $in: $!";
230			mkdir $in \|\| die "can't mkdir $in: $!";
231	dpavlin	196
232			my $suffix_len = length("$parts");
233	dpavlin	201	system("$bin->{'split'} -d -b $max_size -a $suffix_len $out $in/part") == 0 or die "can't split $out: $?";
234	dpavlin	194	unlink $out \|\| die "can't unlink $out: $!";
235			}
236
237	dpavlin	196	$sth_inc_size->execute($size, $parts, $row->{'backup_id'});
238	dpavlin	168	$sth_inc_deleted->execute(0, $row->{'backup_id'});
239	dpavlin	194
240			printf(" %1.2f MB", ($size / 1024 / 1024));
241
242	dpavlin	163	} else {
243			$sth_inc_deleted->execute(1, $row->{'backup_id'});
244			unlink "$tar_dir/$tar_file" \|\| die "can't delete $tar_dir/$tar_file: $!\n";
245			print " EMPTY";
246			}
247	dpavlin	194	print ", dur: ",fmt_time(time() - $t), "\n";
248	dpavlin	158
249			$dbh->commit;
250
251	dpavlin	157	}
252
253			undef $sth;
254			$dbh->disconnect;