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;

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";
        }

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

        my $same = 1;
        my @tar_files;

        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);

                my $total_size = 0;

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

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

        # 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	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	157	my $path = abs_path($0);
26			$path =~ s#/[^/]+$#/#;
27			my $tarIncCreate = $path .= 'BackupPC_tarIncCreate';
28
29			die "can't find $tarIncCreate: $!\n" unless (-x $tarIncCreate);
30
31	dpavlin	201	my $bin;
32	dpavlin	254	foreach my $c (qw/gzip md5sum/) {
33	dpavlin	201	$bin->{$c} = which($c) \|\| die "$0 needs $c, install it\n";
34			}
35
36	dpavlin	214	my %opt;
37			getopts("cd", \%opt );
38	dpavlin	201
39	dpavlin	214	my $debug = $opt{d};
40			my $check = $opt{c} && print STDERR "NOTICE: tar archive check forced\n";
41
42	dpavlin	157	$\|=1;
43
44			my $start_t = time();
45
46			my $t_fmt = '%Y-%m-%d %H:%M:%S';
47
48			my $dsn = $Conf{SearchDSN} \|\| die "Need SearchDSN in config.pl\n";
49			my $user = $Conf{SearchUser} \|\| '';
50
51			my $dbh = DBI->connect($dsn, $user, "", { RaiseError => 1, AutoCommit => 0 });
52
53			my $tar_dir = $Conf{InstallDir}.'/'.$Conf{GzipTempDir};
54
55			die "problem with $tar_dir, check GzipTempDir in configuration\n" unless (-d $tar_dir && -w $tar_dir);
56
57			#---- subs ----
58
59			sub fmt_time {
60			my $t = shift \|\| return;
61			my $out = "";
62			my ($ss,$mm,$hh) = gmtime($t);
63			$out .= "${hh}h" if ($hh);
64			$out .= sprintf("%02d:%02d", $mm,$ss);
65			return $out;
66			}
67
68			sub curr_time {
69			return strftime($t_fmt,localtime());
70			}
71
72	dpavlin	214	sub tar_check($$$$) {
73			my ($host,$share,$num,$filename) = @_;
74
75			if ($debug) {
76			print STDERR " {{ CHECK: ${host}:${share}#${num} and $filename";
77			} else {
78			print " check";
79			}
80
81	dpavlin	254	my @tar_parts;
82
83			if (-d "$tar_dir/$filename") {
84			print STDERR " multi-part" if ($opt{d});
85			opendir(my $dir, "$tar_dir/$filename") \|\| die "can't readdir $tar_dir/$filename: $!";
86			@tar_parts = map { my $p = $_; $p =~ s#^#${filename}/#; $p } grep { !/^\./ && !/md5/ && -f "$tar_dir/$filename/$_" } readdir($dir);
87			closedir($dir);
88			} else {
89			push @tar_parts, "${filename}.tar.gz";
90	dpavlin	214	}
91
92	dpavlin	256	print " [parts: ",join(", ", @tar_parts),"]" if ($opt{d});
93
94			my $same = 1;
95	dpavlin	254	my @tar_files;
96	dpavlin	214
97	dpavlin	254	print " reading";
98
99			foreach my $tarfilename (@tar_parts) {
100
101			print STDERR " $tarfilename" if ($debug);
102
103			my $path = "$tar_dir/$tarfilename";
104			my $md5 = $path;
105			$md5 =~ s/\.tar\.gz$/.md5/ \|\| die "can't create md5 filename from $md5";
106			if (! -e $md5) {
107			print ", creating md5";
108			system( $bin->{md5sum} . " $path > $md5") == 0 or die "can't create md5 $path: $!";
109			}
110
111			open(my $fh, "gzip -cd $tar_dir/$tarfilename \|") or die "can't open $tar_dir/$tarfilename: $!";
112			binmode($fh);
113			my $tar = Archive::Tar::Streamed->new($fh);
114
115	dpavlin	256	my $total_size = 0;
116
117	dpavlin	254	while(my $entry = $tar->next) {
118			push @tar_files, $entry->name;
119	dpavlin	256	$total_size += $entry->size;
120	dpavlin	254	}
121	dpavlin	256
122			if ($total_size > $Conf{MaxArchiveFileSize}) {
123			print STDERR " part too big $total_size > $Conf{MaxArchiveFileSize} }}" if ($debug);
124			$same = 0;
125			last;
126			} elsif ($total_size > $Conf{MaxArchiveSize}) {
127			print STDERR " part bigger than media $total_size > $Conf{MaxArchiveSize} }}" if ($debug);
128			$same = 0;
129			last;
130			}
131	dpavlin	215	}
132	dpavlin	254
133	dpavlin	256	# short-cut and exit;
134			return $same unless($same);
135
136	dpavlin	215	@tar_files = sort @tar_files;
137	dpavlin	214	print STDERR " ",($#tar_files + 1), " files" if ($debug);
138
139			print STDERR ", database" if ($debug);
140
141			my $sth = $dbh->prepare(qq{
142			SELECT path,type
143			FROM files
144			JOIN shares on shares.id = shareid
145			JOIN hosts on hosts.id = shares.hostid
146			WHERE hosts.name = ? and shares.name = ? and backupnum = ?
147			});
148			$sth->execute($host, $share, $num);
149			my @db_files;
150			while( my $row = $sth->fetchrow_hashref ) {
151
152			my $path = $row->{'path'} \|\| die "no path?";
153			$path =~ s#^/#./#;
154			$path .= '/' if ($row->{'type'} == BPC_FTYPE_DIR);
155			push @db_files, $path;
156			}
157
158			print STDERR " ",($#db_files + 1), " files, diff" if ($debug);
159
160			@db_files = sort @db_files;
161
162			if ($#tar_files != $#db_files) {
163			$same = 0;
164			print STDERR " NUMBER" if ($debug);
165			} else {
166			my $diff = Algorithm::Diff->new(\@tar_files, \@db_files);
167			while ( $diff->Next() ) {
168			next if $diff->Same();
169			$same = 0;
170			print "< $_\n" for $diff->Items(1);
171			print "> $_\n" for $diff->Items(2);
172			}
173			}
174
175	dpavlin	216	print " ",($same ? 'ok' : 'DIFFERENT');
176	dpavlin	214	print STDERR " }} " if ($debug);
177
178			return $same;
179			}
180
181
182	dpavlin	157	#----- main
183
184			my $sth = $dbh->prepare( qq{
185
186			select
187	dpavlin	158	backups.id as backup_id,
188	dpavlin	157	hosts.name as host,
189			shares.name as share,
190	dpavlin	192	backups.num as num,
191	dpavlin	196	inc_size,
192			parts
193	dpavlin	157	from backups
194			join shares on backups.hostid = shares.hostid
195			and shares.id = backups.shareid
196			join hosts on shares.hostid = hosts.id
197	dpavlin	192	where not inc_deleted
198	dpavlin	157	order by backups.date
199
200			} );
201
202	dpavlin	213	$sth->execute();
203			my $num_backups = $sth->rows;
204			my $curr_backup = 1;
205
206	dpavlin	157	while (my $row = $sth->fetchrow_hashref) {
207			my $tar_file = BackupPC::SearchLib::getGzipName($row->{'host'}, $row->{'share'}, $row->{'num'});
208	dpavlin	192
209	dpavlin	194	# this will return -1 if file doesn't exist
210			my $size = BackupPC::SearchLib::get_tgz_size_by_name($tar_file);
211	dpavlin	192
212	dpavlin	253	print "# size: $size backup.size: ", $row->{inc_size},"\n" if ($opt{d});
213
214	dpavlin	256	if ( $row->{'inc_size'} != -1 && $size != -1 && $row->{'inc_size'} == $size) {
215			if ($check) {
216			tar_check($row->{'host'}, $row->{'share'}, $row->{'num'}, $tar_file) && next;
217			} else {
218			next;
219			}
220	dpavlin	253	}
221
222	dpavlin	213	print curr_time, " $curr_backup/$num_backups ", $row->{'host'}, ":", $row->{'share'}, " #", $row->{'num'}, " -> $tar_file";
223			$curr_backup++;
224	dpavlin	157
225			my $t = time();
226
227	dpavlin	194	# re-create archive?
228	dpavlin	254	my $cmd = qq{ $tarIncCreate -h "$row->{'host'}" -s "$row->{'share'}" -n $row->{'num'} -f };
229	dpavlin	253	print STDERR "## $cmd\n" if ($debug);
230	dpavlin	157
231	dpavlin	253	if (system($cmd) != 0) {
232			print STDERR " FAILED";
233	dpavlin	194	}
234	dpavlin	157
235	dpavlin	194	print ", dur: ",fmt_time(time() - $t), "\n";
236	dpavlin	158
237			$dbh->commit;
238
239	dpavlin	157	}
240
241			undef $sth;
242			$dbh->disconnect;