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;

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 > 45) {

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