15 |
use Archive::Tar::Streamed; |
use Archive::Tar::Streamed; |
16 |
use Algorithm::Diff; |
use Algorithm::Diff; |
17 |
use Getopt::Std; |
use Getopt::Std; |
18 |
|
use File::Slurp; |
19 |
|
|
20 |
my $bpc = BackupPC::Lib->new || die "can't create BackupPC::Lib"; |
my $bpc = BackupPC::Lib->new || die "can't create BackupPC::Lib"; |
21 |
my %Conf = $bpc->Conf(); |
my %Conf = $bpc->Conf(); |
23 |
use BackupPC::SearchLib; |
use BackupPC::SearchLib; |
24 |
%BackupPC::SearchLib::Conf = %Conf; |
%BackupPC::SearchLib::Conf = %Conf; |
25 |
|
|
|
# cludge: minimum .tar.gz size |
|
|
my $MIN_TAR_SIZE = 80; |
|
|
|
|
26 |
my $path = abs_path($0); |
my $path = abs_path($0); |
27 |
$path =~ s#/[^/]+$#/#; |
$path =~ s#/[^/]+$#/#; |
28 |
my $tarIncCreate = $path .= 'BackupPC_tarIncCreate'; |
my $tarIncCreate = $path .= 'BackupPC_tarIncCreate'; |
30 |
die "can't find $tarIncCreate: $!\n" unless (-x $tarIncCreate); |
die "can't find $tarIncCreate: $!\n" unless (-x $tarIncCreate); |
31 |
|
|
32 |
my $bin; |
my $bin; |
33 |
foreach my $c (qw/gzip split/) { |
foreach my $c (qw/gzip md5sum/) { |
34 |
$bin->{$c} = which($c) || die "$0 needs $c, install it\n"; |
$bin->{$c} = which($c) || die "$0 needs $c, install it\n"; |
35 |
} |
} |
36 |
|
|
70 |
return strftime($t_fmt,localtime()); |
return strftime($t_fmt,localtime()); |
71 |
} |
} |
72 |
|
|
73 |
sub tar_join($) { |
my $hsn_cache; |
74 |
my $filename = shift; |
|
75 |
|
sub get_backup_id($$$) { |
76 |
|
my ($host, $share, $num) = @_; |
77 |
|
|
78 |
|
my $key = "$host $share $num"; |
79 |
|
return $hsn_cache->{$key} if ($hsn_cache->{$key}); |
80 |
|
|
81 |
|
my $sth = $dbh->prepare(qq{ |
82 |
|
SELECT |
83 |
|
backups.id |
84 |
|
FROM backups |
85 |
|
INNER JOIN shares ON backups.shareID=shares.ID |
86 |
|
INNER JOIN hosts ON backups.hostID = hosts.ID |
87 |
|
where hosts.name = ? and shares.name = ? and backups.num = ? |
88 |
|
}); |
89 |
|
$sth->execute($host, $share, $num); |
90 |
|
my ($id) = $sth->fetchrow_array; |
91 |
|
|
92 |
my $in = my $out = $filename; |
$hsn_cache->{"$host $share $num"} = $id; |
|
$out .= '.tmp'; |
|
93 |
|
|
94 |
# FIXME I should really order parts manually! |
print STDERR "# $host $share $num == $id\n" if ($opt{d}); |
|
system("cat $in/part* > $out && rm -Rf $in && mv $out $in") == 0 or die "can't join $in: $?"; |
|
95 |
|
|
96 |
|
return $id; |
97 |
} |
} |
98 |
|
|
99 |
|
|
100 |
sub tar_check($$$$) { |
sub tar_check($$$$) { |
101 |
my ($host,$share,$num,$filename) = @_; |
my ($host,$share,$num,$filename) = @_; |
102 |
|
|
103 |
|
sub check_part { |
104 |
|
my ($host, $share, $num, $part_nr, $tar_size, $size, $md5, $items) = @_; |
105 |
|
my $backup_id = get_backup_id($host, $share, $num); |
106 |
|
my $sth_md5 = $dbh->prepare(qq{ |
107 |
|
select |
108 |
|
id, tar_size, size, md5, items |
109 |
|
from backup_parts |
110 |
|
where backup_id = ? and part_nr = ? |
111 |
|
}); |
112 |
|
|
113 |
|
$sth_md5->execute($backup_id, $part_nr); |
114 |
|
|
115 |
|
if (my $row = $sth_md5->fetchrow_hashref) { |
116 |
|
return if ( |
117 |
|
$row->{tar_size} >= $tar_size && |
118 |
|
$row->{size} == $size && |
119 |
|
$row->{md5} eq $md5 && |
120 |
|
$row->{items} == $items |
121 |
|
); |
122 |
|
print STDERR "# deleting invalid row $row->{id}\n" if ($opt{d}); |
123 |
|
$dbh->do(qq{ delete from backup_parts where id = $row->{id} }); |
124 |
|
} |
125 |
|
print STDERR "# inserting new backup_part row\n"; |
126 |
|
my $sth_insert = $dbh->prepare(qq{ |
127 |
|
insert into backup_parts ( |
128 |
|
backup_id, |
129 |
|
part_nr, |
130 |
|
tar_size, |
131 |
|
size, |
132 |
|
md5, |
133 |
|
items |
134 |
|
) values (?,?,?,?,?,?) |
135 |
|
}); |
136 |
|
|
137 |
|
$sth_insert->execute($backup_id, $part_nr, $tar_size, $size, $md5, $items); |
138 |
|
$dbh->commit; |
139 |
|
} |
140 |
|
|
141 |
if ($debug) { |
if ($debug) { |
142 |
print STDERR " {{ CHECK: ${host}:${share}#${num} and $filename"; |
print STDERR " {{ CHECK: ${host}:${share}#${num} and $filename"; |
143 |
} else { |
} else { |
144 |
print " check"; |
print " check $filename"; |
145 |
} |
} |
146 |
|
|
147 |
if (-d $filename) { |
my @tar_parts; |
148 |
print STDERR ", joining"; |
|
149 |
tar_join($filename); |
if (-d "$tar_dir/$filename") { |
150 |
|
print STDERR " multi-part" if ($opt{d}); |
151 |
|
opendir(my $dir, "$tar_dir/$filename") || die "can't readdir $tar_dir/$filename: $!"; |
152 |
|
@tar_parts = map { my $p = $_; $p =~ s#^#${filename}/#; $p } grep { !/^\./ && !/md5/ && -f "$tar_dir/$filename/$_" } readdir($dir); |
153 |
|
closedir($dir); |
154 |
|
} else { |
155 |
|
push @tar_parts, "${filename}.tar.gz"; |
156 |
} |
} |
157 |
|
|
158 |
print STDERR ", opening" if ($debug); |
print " [parts: ",join(", ", @tar_parts),"]" if ($opt{d}); |
|
open(my $fh, "gzip -cd $filename |") or die "can't open $filename: $!"; |
|
|
binmode($fh); |
|
|
my $tar = Archive::Tar::Streamed->new($fh); |
|
159 |
|
|
160 |
print STDERR ", tar" if ($debug); |
my $same = 1; |
161 |
my @tar_files; |
my @tar_files; |
162 |
while(my $entry = $tar->next) { |
|
163 |
push @tar_files, $entry->name; |
my $backup_part; |
164 |
|
|
165 |
|
print " reading" if ($opt{d}); |
166 |
|
|
167 |
|
foreach my $tarfilename (@tar_parts) { |
168 |
|
|
169 |
|
print STDERR " $tarfilename" if ($debug); |
170 |
|
|
171 |
|
my $path = "$tar_dir/$tarfilename"; |
172 |
|
my $md5_path = $path; |
173 |
|
$md5_path =~ s/\.tar\.gz$/.md5/ || die "can't create md5 filename from $md5_path"; |
174 |
|
if (! -e $md5_path || -z $md5_path) { |
175 |
|
print ", creating md5"; |
176 |
|
system( $bin->{md5sum} . " $path > $md5_path") == 0 or die "can't create md5 $path: $!"; |
177 |
|
} |
178 |
|
|
179 |
|
my $md5 = read_file( $md5_path ) || die "can't read md5sum file $md5_path: $!"; |
180 |
|
$md5 =~ s#\s.*$##; |
181 |
|
|
182 |
|
my $part_nr = 1; |
183 |
|
$part_nr = $1 if ($tarfilename =~ m#/(\d+)\.tar\.gz#); |
184 |
|
|
185 |
|
my $size = (stat( "$tar_dir/$tarfilename" ))[7] || die "can't stat $tar_dir/$tarfilename"; |
186 |
|
|
187 |
|
open(my $fh, "gzip -cd $tar_dir/$tarfilename |") or die "can't open $tar_dir/$tarfilename: $!"; |
188 |
|
binmode($fh); |
189 |
|
my $tar = Archive::Tar::Streamed->new($fh); |
190 |
|
|
191 |
|
my $tar_size = 0; |
192 |
|
my $items = 0; |
193 |
|
|
194 |
|
while(my $entry = $tar->next) { |
195 |
|
push @tar_files, $entry->name; |
196 |
|
$items++; |
197 |
|
$tar_size += $entry->size; |
198 |
|
} |
199 |
|
|
200 |
|
if ($tar_size > $Conf{MaxArchiveFileSize}) { |
201 |
|
print STDERR " part too big $tar_size > $Conf{MaxArchiveFileSize} }}" if ($debug); |
202 |
|
$same = 0; |
203 |
|
last; |
204 |
|
} elsif ($size > $Conf{MaxArchiveSize}) { |
205 |
|
print STDERR " part bigger than media $size > $Conf{MaxArchiveSize} }}" if ($debug); |
206 |
|
$same = 0; |
207 |
|
last; |
208 |
|
} |
209 |
|
|
210 |
|
check_part($host, $share, $num, $part_nr, $tar_size, $size, $md5, $items); |
211 |
} |
} |
212 |
|
|
213 |
|
# short-cut and exit; |
214 |
|
return $same unless($same); |
215 |
|
|
216 |
@tar_files = sort @tar_files; |
@tar_files = sort @tar_files; |
217 |
print STDERR " ",($#tar_files + 1), " files" if ($debug); |
print STDERR " ",($#tar_files + 1), " files" if ($debug); |
218 |
|
|
239 |
|
|
240 |
@db_files = sort @db_files; |
@db_files = sort @db_files; |
241 |
|
|
|
my $same = 1; |
|
242 |
if ($#tar_files != $#db_files) { |
if ($#tar_files != $#db_files) { |
243 |
$same = 0; |
$same = 0; |
244 |
print STDERR " NUMBER" if ($debug); |
print STDERR " NUMBER" if ($debug); |
279 |
|
|
280 |
} ); |
} ); |
281 |
|
|
|
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 = ? }); |
|
|
|
|
|
|
|
282 |
$sth->execute(); |
$sth->execute(); |
283 |
my $num_backups = $sth->rows; |
my $num_backups = $sth->rows; |
284 |
my $curr_backup = 1; |
my $curr_backup = 1; |
289 |
# this will return -1 if file doesn't exist |
# this will return -1 if file doesn't exist |
290 |
my $size = BackupPC::SearchLib::get_tgz_size_by_name($tar_file); |
my $size = BackupPC::SearchLib::get_tgz_size_by_name($tar_file); |
291 |
|
|
292 |
|
print "# size: $size backup.size: ", $row->{inc_size},"\n" if ($opt{d}); |
293 |
|
|
294 |
|
if ( $row->{'inc_size'} != -1 && $size != -1 && $row->{'inc_size'} >= $size) { |
295 |
|
if ($check) { |
296 |
|
tar_check($row->{'host'}, $row->{'share'}, $row->{'num'}, $tar_file) && next; |
297 |
|
} else { |
298 |
|
next; |
299 |
|
} |
300 |
|
} |
301 |
|
|
302 |
print curr_time, " $curr_backup/$num_backups ", $row->{'host'}, ":", $row->{'share'}, " #", $row->{'num'}, " -> $tar_file"; |
print curr_time, " $curr_backup/$num_backups ", $row->{'host'}, ":", $row->{'share'}, " #", $row->{'num'}, " -> $tar_file"; |
303 |
$curr_backup++; |
$curr_backup++; |
304 |
|
|
305 |
my $t = time(); |
my $t = time(); |
306 |
|
|
307 |
# re-create archive? |
# re-create archive? |
308 |
if ($row->{'inc_size'} == -1 || $size == -1 || |
my $cmd = qq{ $tarIncCreate -h "$row->{'host'}" -s "$row->{'share'}" -n $row->{'num'} -f }; |
309 |
$row->{'inc_size'} != $size || |
print STDERR "## $cmd\n" if ($debug); |
|
$check && ! tar_check($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: $?"; |
|
310 |
|
|
311 |
rename("${tar_dir}/${tar_file}.tmp", "$tar_dir/$tar_file") or die "can't rename $tar_dir/$tar_file: $!"; |
if (system($cmd) != 0) { |
312 |
|
print STDERR " FAILED"; |
|
$size = (stat( "$tar_dir/$tar_file" ))[7]; |
|
313 |
} |
} |
314 |
|
|
|
if ($size > $MIN_TAR_SIZE) { |
|
|
|
|
|
my $max_size = $Conf{'MaxArchiveSize'} || die "problem with MaxArchiveSize parametar"; |
|
|
$max_size *= 1024; # convert to bytes |
|
|
|
|
|
my $max_file_size = $Conf{'MaxArchiveFileSize'} || die "problem with MaxArchiveFileSize parametar"; |
|
|
$max_file_size *= 1024; # bytes |
|
|
|
|
|
if ($max_file_size > $max_size) { |
|
|
warn "MaxArchiveFileSize ($max_file_size) is bigger than MaxArchiveSize ($max_size)\n"; |
|
|
$max_file_size = $max_size; |
|
|
} |
|
|
|
|
|
# maximum file size on ISO image is 4Gb |
|
|
# this will require Linux kernel 2.6.8 or newer |
|
|
if ( $max_size > $max_file_size ) { |
|
|
$max_size = $max_file_size; |
|
|
} |
|
|
|
|
|
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"; |
|
|
} |
|
315 |
print ", dur: ",fmt_time(time() - $t), "\n"; |
print ", dur: ",fmt_time(time() - $t), "\n"; |
316 |
|
|
317 |
$dbh->commit; |
$dbh->commit; |