100 |
sub tar_check($$$$) { |
sub tar_check($$$$) { |
101 |
my ($host,$share,$num,$filename) = @_; |
my ($host,$share,$num,$filename) = @_; |
102 |
|
|
103 |
|
my $t = time(); |
104 |
|
print curr_time, " check $host:$share#$num -> $filename"; |
105 |
|
|
106 |
sub check_part { |
sub check_part { |
107 |
my ($host, $share, $num, $part_nr, $tar_size, $size, $md5, $items) = @_; |
my ($host, $share, $num, $part_nr, $tar_size, $size, $md5, $items) = @_; |
108 |
my $backup_id = get_backup_id($host, $share, $num); |
my $backup_id = get_backup_id($host, $share, $num); |
117 |
|
|
118 |
if (my $row = $sth_md5->fetchrow_hashref) { |
if (my $row = $sth_md5->fetchrow_hashref) { |
119 |
return if ( |
return if ( |
120 |
$row->{tar_size} == $tar_size && |
$row->{tar_size} >= $tar_size && |
121 |
$row->{size} == $size && |
$row->{size} == $size && |
122 |
$row->{md5} eq $md5 && |
$row->{md5} eq $md5 && |
123 |
$row->{items} == $items |
$row->{items} == $items |
124 |
); |
); |
125 |
print STDERR "# deleting invalid row $row->{id}\n" if ($opt{d}); |
print ", deleting invalid backup_parts $row->{id}"; |
126 |
$dbh->do(qq{ delete from backup_parts where id = $row->{id} }); |
$dbh->do(qq{ delete from backup_parts where id = $row->{id} }); |
127 |
} |
} |
128 |
print STDERR "# inserting new backup_part row\n"; |
print ", inserting new"; |
129 |
my $sth_insert = $dbh->prepare(qq{ |
my $sth_insert = $dbh->prepare(qq{ |
130 |
insert into backup_parts ( |
insert into backup_parts ( |
131 |
backup_id, |
backup_id, |
141 |
$dbh->commit; |
$dbh->commit; |
142 |
} |
} |
143 |
|
|
|
if ($debug) { |
|
|
print STDERR " {{ CHECK: ${host}:${share}#${num} and $filename"; |
|
|
} else { |
|
|
print " check"; |
|
|
} |
|
|
|
|
144 |
my @tar_parts; |
my @tar_parts; |
145 |
|
|
146 |
if (-d "$tar_dir/$filename") { |
if (-d "$tar_dir/$filename") { |
147 |
print STDERR " multi-part" if ($opt{d}); |
print ", multi-part"; |
148 |
opendir(my $dir, "$tar_dir/$filename") || die "can't readdir $tar_dir/$filename: $!"; |
opendir(my $dir, "$tar_dir/$filename") || die "can't readdir $tar_dir/$filename: $!"; |
149 |
@tar_parts = map { my $p = $_; $p =~ s#^#${filename}/#; $p } grep { !/^\./ && !/md5/ && -f "$tar_dir/$filename/$_" } readdir($dir); |
@tar_parts = map { my $p = $_; $p =~ s#^#${filename}/#; $p } grep { !/^\./ && !/md5/ && -f "$tar_dir/$filename/$_" } readdir($dir); |
150 |
closedir($dir); |
closedir($dir); |
159 |
|
|
160 |
my $backup_part; |
my $backup_part; |
161 |
|
|
162 |
print " reading"; |
print " reading" if ($opt{d}); |
163 |
|
|
164 |
foreach my $tarfilename (@tar_parts) { |
foreach my $tarfilename (@tar_parts) { |
165 |
|
|
166 |
print STDERR " $tarfilename" if ($debug); |
print "\n\t- $tarfilename"; |
167 |
|
|
168 |
my $path = "$tar_dir/$tarfilename"; |
my $size = (stat( "$tar_dir/$tarfilename" ))[7] || die "can't stat $tar_dir/$tarfilename"; |
|
my $md5_path = $path; |
|
|
$md5_path =~ s/\.tar\.gz$/.md5/ || die "can't create md5 filename from $md5_path"; |
|
|
if (! -e $md5_path) { |
|
|
print ", creating md5"; |
|
|
system( $bin->{md5sum} . " $path > $md5_path") == 0 or die "can't create md5 $path: $!"; |
|
|
} |
|
169 |
|
|
170 |
my $md5 = read_file( $md5_path ) || die "can't read md5sum file $md5_path: $!"; |
if ($size > $Conf{MaxArchiveSize}) { |
171 |
|
print STDERR " part bigger than media $size > $Conf{MaxArchiveSize} }}" if ($debug); |
172 |
|
$same = 0; |
173 |
|
last; |
174 |
|
} |
175 |
|
|
176 |
my $part_nr = 1; |
print ", $size bytes"; |
|
$part_nr = $1 if ($tarfilename =~ m#/(\d+)\.tar\.gz#); |
|
177 |
|
|
178 |
my $size = (stat( "$tar_dir/$tarfilename" ))[7] || die "can't stat $tar_dir/$tarfilename"; |
my $path = "$tar_dir/$tarfilename"; |
179 |
|
|
180 |
open(my $fh, "gzip -cd $tar_dir/$tarfilename |") or die "can't open $tar_dir/$tarfilename: $!"; |
open(my $fh, "gzip -cd $tar_dir/$tarfilename |") or die "can't open $tar_dir/$tarfilename: $!"; |
181 |
binmode($fh); |
binmode($fh); |
188 |
push @tar_files, $entry->name; |
push @tar_files, $entry->name; |
189 |
$items++; |
$items++; |
190 |
$tar_size += $entry->size; |
$tar_size += $entry->size; |
191 |
|
|
192 |
|
if ($tar_size > $Conf{MaxArchiveFileSize}) { |
193 |
|
print ", part $tarfilename is too big $tar_size > $Conf{MaxArchiveFileSize}\n"; |
194 |
|
$same = 0; |
195 |
|
last; |
196 |
|
} |
197 |
|
|
198 |
} |
} |
199 |
|
|
200 |
if ($tar_size > $Conf{MaxArchiveFileSize}) { |
print ", $items items"; |
201 |
print STDERR " part too big $tar_size > $Conf{MaxArchiveFileSize} }}" if ($debug); |
|
202 |
$same = 0; |
# |
203 |
last; |
# check if md5 exists, and if not, create one |
204 |
} elsif ($size > $Conf{MaxArchiveSize}) { |
# |
205 |
print STDERR " part bigger than media $size > $Conf{MaxArchiveSize} }}" if ($debug); |
|
206 |
$same = 0; |
my $md5_path = $path; |
207 |
last; |
$md5_path =~ s/\.tar\.gz$/.md5/ || die "can't create md5 filename from $md5_path"; |
208 |
|
if (! -e $md5_path || -z $md5_path) { |
209 |
|
print ", creating md5"; |
210 |
|
system( $bin->{md5sum} . " $path > $md5_path") == 0 or die "can't create md5 $path: $!"; |
211 |
|
} else { |
212 |
|
## FIXME check if existing md5 is valid |
213 |
} |
} |
214 |
|
|
215 |
|
my $md5 = read_file( $md5_path ) || die "can't read md5sum file $md5_path: $!"; |
216 |
|
$md5 =~ s#\s.*$##; |
217 |
|
|
218 |
|
# extract part number from filename |
219 |
|
my $part_nr = 1; |
220 |
|
$part_nr = $1 if ($tarfilename =~ m#/(\d+)\.tar\.gz#); |
221 |
|
|
222 |
|
# |
223 |
|
# finally, check if backup_parts table in database is valid |
224 |
|
# |
225 |
|
|
226 |
check_part($host, $share, $num, $part_nr, $tar_size, $size, $md5, $items); |
check_part($host, $share, $num, $part_nr, $tar_size, $size, $md5, $items); |
227 |
} |
} |
228 |
|
|
230 |
return $same unless($same); |
return $same unless($same); |
231 |
|
|
232 |
@tar_files = sort @tar_files; |
@tar_files = sort @tar_files; |
233 |
print STDERR " ",($#tar_files + 1), " files" if ($debug); |
print "\n\t",($#tar_files + 1), " tar files"; |
|
|
|
|
print STDERR ", database" if ($debug); |
|
234 |
|
|
235 |
my $sth = $dbh->prepare(qq{ |
my $sth = $dbh->prepare(qq{ |
236 |
SELECT path,type |
SELECT path,type |
249 |
push @db_files, $path; |
push @db_files, $path; |
250 |
} |
} |
251 |
|
|
252 |
print STDERR " ",($#db_files + 1), " files, diff" if ($debug); |
print " ",($#db_files + 1), " database files, diff"; |
253 |
|
|
254 |
@db_files = sort @db_files; |
@db_files = sort @db_files; |
255 |
|
|
256 |
if ($#tar_files != $#db_files) { |
if ($#tar_files != $#db_files) { |
257 |
$same = 0; |
$same = 0; |
258 |
print STDERR " NUMBER" if ($debug); |
print " NUMBER"; |
259 |
} else { |
} else { |
260 |
my $diff = Algorithm::Diff->new(\@tar_files, \@db_files); |
my $diff = Algorithm::Diff->new(\@tar_files, \@db_files); |
261 |
while ( $diff->Next() ) { |
while ( $diff->Next() ) { |
266 |
} |
} |
267 |
} |
} |
268 |
|
|
269 |
print " ",($same ? 'ok' : 'DIFFERENT'); |
print " ",($same ? 'ok' : 'DIFFERENT'), |
270 |
print STDERR " }} " if ($debug); |
", dur: ",fmt_time(time() - $t), "\n"; |
271 |
|
|
272 |
return $same; |
return $same; |
273 |
} |
} |
298 |
my $curr_backup = 1; |
my $curr_backup = 1; |
299 |
|
|
300 |
while (my $row = $sth->fetchrow_hashref) { |
while (my $row = $sth->fetchrow_hashref) { |
301 |
|
|
302 |
|
$curr_backup++; |
303 |
|
|
304 |
my $tar_file = BackupPC::SearchLib::getGzipName($row->{'host'}, $row->{'share'}, $row->{'num'}); |
my $tar_file = BackupPC::SearchLib::getGzipName($row->{'host'}, $row->{'share'}, $row->{'num'}); |
305 |
|
|
306 |
# this will return -1 if file doesn't exist |
# this will return -1 if file doesn't exist |
308 |
|
|
309 |
print "# size: $size backup.size: ", $row->{inc_size},"\n" if ($opt{d}); |
print "# size: $size backup.size: ", $row->{inc_size},"\n" if ($opt{d}); |
310 |
|
|
311 |
if ( $row->{'inc_size'} != -1 && $size != -1 && $row->{'inc_size'} == $size) { |
if ( $row->{'inc_size'} != -1 && $size != -1 && $row->{'inc_size'} >= $size) { |
312 |
if ($check) { |
if ($check) { |
313 |
tar_check($row->{'host'}, $row->{'share'}, $row->{'num'}, $tar_file) && next; |
tar_check($row->{'host'}, $row->{'share'}, $row->{'num'}, $tar_file) && next; |
314 |
} else { |
} else { |
316 |
} |
} |
317 |
} |
} |
318 |
|
|
319 |
print curr_time, " $curr_backup/$num_backups ", $row->{'host'}, ":", $row->{'share'}, " #", $row->{'num'}, " -> $tar_file"; |
print curr_time, " creating $curr_backup/$num_backups ", $row->{'host'}, ":", $row->{'share'}, " #", $row->{'num'}, " -> $tar_file"; |
|
$curr_backup++; |
|
320 |
|
|
321 |
my $t = time(); |
my $t = time(); |
322 |
|
|