16 |
use Algorithm::Diff; |
use Algorithm::Diff; |
17 |
use Getopt::Std; |
use Getopt::Std; |
18 |
use File::Slurp; |
use File::Slurp; |
19 |
|
use File::Pid; |
20 |
|
|
21 |
|
=head1 NAME |
22 |
|
|
23 |
|
BackupPC_incPartsUpdate |
24 |
|
|
25 |
|
=head1 DESCRIPTION |
26 |
|
|
27 |
|
Create C<.tar.gz> increments on disk calling C<BackupPC_tarIncCreate>. |
28 |
|
|
29 |
|
Following options are supported (but all are optional): |
30 |
|
|
31 |
|
=over 4 |
32 |
|
|
33 |
|
=item -h hostname |
34 |
|
|
35 |
|
Update parts for just single C<hostname> |
36 |
|
|
37 |
|
=item -c |
38 |
|
|
39 |
|
Force check for tar archives which exist on disk |
40 |
|
|
41 |
|
=item -d |
42 |
|
|
43 |
|
Turn debugging output |
44 |
|
|
45 |
|
=back |
46 |
|
|
47 |
|
=cut |
48 |
|
|
49 |
|
my %opt; |
50 |
|
getopts("cdh:", \%opt ); |
51 |
|
|
52 |
|
my $debug = $opt{d}; |
53 |
|
my $check = $opt{c} && print STDERR "NOTICE: tar archive check forced\n"; |
54 |
|
|
55 |
|
my $pid_path = abs_path($0); |
56 |
|
$pid_path =~ s/\W+/_/g; |
57 |
|
|
58 |
|
my $pidfile = new File::Pid({ |
59 |
|
file => "/tmp/$pid_path", |
60 |
|
}); |
61 |
|
|
62 |
|
if (my $pid = $pidfile->running ) { |
63 |
|
die "$0 already running: $pid\n"; |
64 |
|
} elsif ($pidfile->pid ne $$) { |
65 |
|
$pidfile->remove; |
66 |
|
$pidfile = new File::Pid; |
67 |
|
} |
68 |
|
|
69 |
|
print STDERR "$0 using pid ",$pidfile->pid," file ",$pidfile->file,"\n"; |
70 |
|
$pidfile->write; |
71 |
|
|
72 |
my $bpc = BackupPC::Lib->new || die "can't create BackupPC::Lib"; |
my $bpc = BackupPC::Lib->new || die "can't create BackupPC::Lib"; |
73 |
my %Conf = $bpc->Conf(); |
my %Conf = $bpc->Conf(); |
86 |
$bin->{$c} = which($c) || die "$0 needs $c, install it\n"; |
$bin->{$c} = which($c) || die "$0 needs $c, install it\n"; |
87 |
} |
} |
88 |
|
|
|
my %opt; |
|
|
getopts("cd", \%opt ); |
|
|
|
|
|
my $debug = $opt{d}; |
|
|
my $check = $opt{c} && print STDERR "NOTICE: tar archive check forced\n"; |
|
|
|
|
89 |
$|=1; |
$|=1; |
90 |
|
|
91 |
my $start_t = time(); |
my $start_t = time(); |
130 |
FROM backups |
FROM backups |
131 |
INNER JOIN shares ON backups.shareID=shares.ID |
INNER JOIN shares ON backups.shareID=shares.ID |
132 |
INNER JOIN hosts ON backups.hostID = hosts.ID |
INNER JOIN hosts ON backups.hostID = hosts.ID |
133 |
where hosts.name = ? and shares.name = ? and backups.num = ? |
WHERE hosts.name = ? and shares.name = ? and backups.num = ? |
134 |
}); |
}); |
135 |
$sth->execute($host, $share, $num); |
$sth->execute($host, $share, $num); |
136 |
my ($id) = $sth->fetchrow_array; |
my ($id) = $sth->fetchrow_array; |
142 |
return $id; |
return $id; |
143 |
} |
} |
144 |
|
|
145 |
|
sub backup_inc_deleted($) { |
146 |
|
my $backup_id = shift; |
147 |
|
my $sth_inc_deleted = $dbh->prepare(qq{ |
148 |
|
update backups set |
149 |
|
inc_deleted = true |
150 |
|
where id = ? |
151 |
|
}); |
152 |
|
$sth_inc_deleted->execute($backup_id); |
153 |
|
} |
154 |
|
|
155 |
sub tar_check($$$$) { |
sub tar_check($$$$) { |
156 |
my ($host,$share,$num,$filename) = @_; |
my ($host,$share,$num,$filename) = @_; |
157 |
|
|
158 |
|
my $t = time(); |
159 |
|
print curr_time, " check $host:$share#$num -> $filename"; |
160 |
|
|
161 |
|
# depending on expected returned value this is used like: |
162 |
|
# my $uncompress_size = get_gzip_size('/full/path/to.gz'); |
163 |
|
# my ($compress_size, $uncompress_size) = get_gzip_size('/path.gz'); |
164 |
|
sub get_gzip_size($) { |
165 |
|
my $filename = shift; |
166 |
|
die "file $filename problem: $!" unless (-r $filename); |
167 |
|
open(my $gzip, $bin->{gzip}." -l $filename |") || die "can't gzip -l $filename: $!"; |
168 |
|
my $line = <$gzip>; |
169 |
|
chomp($line); |
170 |
|
$line = <$gzip> if ($line =~ /^\s+compressed/); |
171 |
|
|
172 |
|
my ($comp, $uncomp) = (0,0); |
173 |
|
|
174 |
|
if ($line =~ m/^\s+(\d+)\s+(\d+)\s+\d+\.\d+/) { |
175 |
|
if (wantarray) { |
176 |
|
return [ $1, $2 ]; |
177 |
|
} else { |
178 |
|
return $2; |
179 |
|
} |
180 |
|
} else { |
181 |
|
die "can't find size in line: $line"; |
182 |
|
} |
183 |
|
} |
184 |
|
|
185 |
sub check_part { |
sub check_part { |
186 |
my ($host, $share, $num, $part_nr, $tar_size, $size, $md5, $items) = @_; |
my ($host, $share, $num, $part_nr, $tar_size, $size, $md5, $items) = @_; |
187 |
my $backup_id = get_backup_id($host, $share, $num); |
my $backup_id = get_backup_id($host, $share, $num); |
201 |
$row->{md5} eq $md5 && |
$row->{md5} eq $md5 && |
202 |
$row->{items} == $items |
$row->{items} == $items |
203 |
); |
); |
204 |
print STDERR "# deleting invalid row $row->{id}\n" if ($opt{d}); |
print ", deleting invalid backup_parts $row->{id}"; |
205 |
$dbh->do(qq{ delete from backup_parts where id = $row->{id} }); |
$dbh->do(qq{ delete from backup_parts where id = $row->{id} }); |
206 |
} |
} |
207 |
print STDERR "# inserting new backup_part row\n"; |
print ", inserting new"; |
208 |
my $sth_insert = $dbh->prepare(qq{ |
my $sth_insert = $dbh->prepare(qq{ |
209 |
insert into backup_parts ( |
insert into backup_parts ( |
210 |
backup_id, |
backup_id, |
220 |
$dbh->commit; |
$dbh->commit; |
221 |
} |
} |
222 |
|
|
|
if ($debug) { |
|
|
print STDERR " {{ CHECK: ${host}:${share}#${num} and $filename"; |
|
|
} else { |
|
|
print " check"; |
|
|
} |
|
|
|
|
223 |
my @tar_parts; |
my @tar_parts; |
224 |
|
|
225 |
if (-d "$tar_dir/$filename") { |
if (-d "$tar_dir/$filename") { |
226 |
print STDERR " multi-part" if ($opt{d}); |
print ", multi-part"; |
227 |
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: $!"; |
228 |
@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($dir); |
229 |
closedir($dir); |
closedir($dir); |
238 |
|
|
239 |
my $backup_part; |
my $backup_part; |
240 |
|
|
241 |
print " reading"; |
print " reading" if ($opt{d}); |
242 |
|
|
243 |
foreach my $tarfilename (@tar_parts) { |
foreach my $tarfilename (@tar_parts) { |
244 |
|
|
245 |
print STDERR " $tarfilename" if ($debug); |
print "\n\t- $tarfilename"; |
246 |
|
|
247 |
my $path = "$tar_dir/$tarfilename"; |
my $path = "$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: $!"; |
|
|
} |
|
248 |
|
|
249 |
my $md5 = read_file( $md5_path ) || die "can't read md5sum file $md5_path: $!"; |
my $size = (stat( $path ))[7] || die "can't stat $path: $!"; |
|
$md5 =~ s#\s.*$##; |
|
250 |
|
|
251 |
my $part_nr = 1; |
if ($size > $Conf{MaxArchiveSize}) { |
252 |
$part_nr = $1 if ($tarfilename =~ m#/(\d+)\.tar\.gz#); |
print ", part bigger than media $size > $Conf{MaxArchiveSize}\n"; |
253 |
|
return 0; |
254 |
|
} |
255 |
|
|
256 |
my $size = (stat( "$tar_dir/$tarfilename" ))[7] || die "can't stat $tar_dir/$tarfilename"; |
print ", $size bytes"; |
257 |
|
|
258 |
open(my $fh, "gzip -cd $tar_dir/$tarfilename |") or die "can't open $tar_dir/$tarfilename: $!"; |
|
259 |
|
open(my $fh, "gzip -cd $path |") or die "can't open $path: $!"; |
260 |
binmode($fh); |
binmode($fh); |
261 |
my $tar = Archive::Tar::Streamed->new($fh); |
my $tar = Archive::Tar::Streamed->new($fh); |
262 |
|
|
263 |
my $tar_size = 0; |
my $tar_size_inarc = 0; |
264 |
my $items = 0; |
my $items = 0; |
265 |
|
|
266 |
while(my $entry = $tar->next) { |
while(my $entry = $tar->next) { |
267 |
push @tar_files, $entry->name; |
push @tar_files, $entry->name; |
268 |
$items++; |
$items++; |
269 |
$tar_size += $entry->size; |
$tar_size_inarc += $entry->size; |
270 |
|
|
271 |
|
if ($tar_size_inarc > $Conf{MaxArchiveFileSize}) { |
272 |
|
print ", part $tarfilename is too big $tar_size_inarc > $Conf{MaxArchiveFileSize}\n"; |
273 |
|
return 0; |
274 |
|
} |
275 |
|
|
276 |
} |
} |
277 |
|
|
278 |
if ($tar_size > $Conf{MaxArchiveFileSize}) { |
close($fh); |
279 |
print STDERR " part too big $tar_size > $Conf{MaxArchiveFileSize} }}" if ($debug); |
|
280 |
$same = 0; |
print ", $items items"; |
281 |
last; |
|
282 |
} elsif ($size > $Conf{MaxArchiveSize}) { |
if ($tar_size_inarc == 0 && $items == 0) { |
283 |
print STDERR " part bigger than media $size > $Conf{MaxArchiveSize} }}" if ($debug); |
print ", EMPTY tar\n"; |
284 |
$same = 0; |
|
285 |
last; |
my $backup_id = get_backup_id($host, $share, $num); |
286 |
|
backup_inc_deleted( $backup_id ); |
287 |
|
|
288 |
|
$dbh->commit; |
289 |
|
|
290 |
|
return 1; |
291 |
} |
} |
292 |
|
|
293 |
|
my $tar_size = get_gzip_size( $path ); |
294 |
|
|
295 |
|
# real tar size is bigger because of padding |
296 |
|
if ($tar_size_inarc > $tar_size) { |
297 |
|
print ", size of files in tar ($tar_size_inarc) bigger than whole tar ($tar_size)!\n"; |
298 |
|
return 0; |
299 |
|
} |
300 |
|
|
301 |
|
# |
302 |
|
# check if md5 exists, and if not, create one |
303 |
|
# |
304 |
|
|
305 |
|
my $md5_path = $path; |
306 |
|
$md5_path =~ s/\.tar\.gz$/.md5/ || die "can't create md5 filename from $md5_path"; |
307 |
|
if (! -e $md5_path || -z $md5_path) { |
308 |
|
print ", creating md5"; |
309 |
|
system( $bin->{md5sum} . " $path > $md5_path") == 0 or die "can't create md5 $path: $!"; |
310 |
|
} else { |
311 |
|
## FIXME check if existing md5 is valid |
312 |
|
} |
313 |
|
|
314 |
|
my $md5 = read_file( $md5_path ) || die "can't read md5sum file $md5_path: $!"; |
315 |
|
$md5 =~ s#\s.*$##; |
316 |
|
|
317 |
|
# extract part number from filename |
318 |
|
my $part_nr = 1; |
319 |
|
$part_nr = $1 if ($tarfilename =~ m#/(\d+)\.tar\.gz#); |
320 |
|
|
321 |
|
# |
322 |
|
# finally, check if backup_parts table in database is valid |
323 |
|
# |
324 |
|
|
325 |
check_part($host, $share, $num, $part_nr, $tar_size, $size, $md5, $items); |
check_part($host, $share, $num, $part_nr, $tar_size, $size, $md5, $items); |
326 |
} |
} |
327 |
|
|
329 |
return $same unless($same); |
return $same unless($same); |
330 |
|
|
331 |
@tar_files = sort @tar_files; |
@tar_files = sort @tar_files; |
332 |
print STDERR " ",($#tar_files + 1), " files" if ($debug); |
print "\n\t",($#tar_files + 1), " tar files"; |
|
|
|
|
print STDERR ", database" if ($debug); |
|
333 |
|
|
334 |
my $sth = $dbh->prepare(qq{ |
my $sth = $dbh->prepare(qq{ |
335 |
SELECT path,type |
SELECT path,type |
348 |
push @db_files, $path; |
push @db_files, $path; |
349 |
} |
} |
350 |
|
|
351 |
print STDERR " ",($#db_files + 1), " files, diff" if ($debug); |
print " ",($#db_files + 1), " database files, diff"; |
352 |
|
|
353 |
@db_files = sort @db_files; |
@db_files = sort @db_files; |
354 |
|
|
355 |
if ($#tar_files != $#db_files) { |
if ($#tar_files != $#db_files) { |
356 |
$same = 0; |
$same = 0; |
357 |
print STDERR " NUMBER" if ($debug); |
print " NUMBER"; |
358 |
} else { |
} else { |
359 |
my $diff = Algorithm::Diff->new(\@tar_files, \@db_files); |
my $diff = Algorithm::Diff->new(\@tar_files, \@db_files); |
360 |
while ( $diff->Next() ) { |
while ( $diff->Next() ) { |
365 |
} |
} |
366 |
} |
} |
367 |
|
|
368 |
print " ",($same ? 'ok' : 'DIFFERENT'); |
print " ",($same ? 'ok' : 'DIFFERENT'), |
369 |
print STDERR " }} " if ($debug); |
", dur: ",fmt_time(time() - $t), "\n"; |
370 |
|
|
371 |
return $same; |
return $same; |
372 |
} |
} |
381 |
hosts.name as host, |
hosts.name as host, |
382 |
shares.name as share, |
shares.name as share, |
383 |
backups.num as num, |
backups.num as num, |
384 |
|
backups.date, |
385 |
inc_size, |
inc_size, |
386 |
parts |
parts, |
387 |
|
count(backup_parts.backup_id) as backup_parts |
388 |
from backups |
from backups |
389 |
join shares on backups.hostid = shares.hostid |
join shares on backups.hostid = shares.hostid |
390 |
and shares.id = backups.shareid |
and shares.id = backups.shareid |
391 |
join hosts on shares.hostid = hosts.id |
join hosts on shares.hostid = hosts.id |
392 |
where not inc_deleted |
full outer join backup_parts on backups.id = backup_parts.backup_id |
393 |
|
where not inc_deleted and backups.size > 0 |
394 |
|
group by backups.id, hosts.name, shares.name, backups.num, backups.date, inc_size, parts, backup_parts.backup_id |
395 |
order by backups.date |
order by backups.date |
396 |
|
|
397 |
} ); |
} ); |
400 |
my $num_backups = $sth->rows; |
my $num_backups = $sth->rows; |
401 |
my $curr_backup = 1; |
my $curr_backup = 1; |
402 |
|
|
403 |
|
if ($opt{h}) { |
404 |
|
warn "making increments just for host $opt{h}\n"; |
405 |
|
} |
406 |
|
|
407 |
while (my $row = $sth->fetchrow_hashref) { |
while (my $row = $sth->fetchrow_hashref) { |
408 |
|
|
409 |
|
if ($opt{h} && $row->{host} ne $opt{h}) { |
410 |
|
warn "skipped $row->{host}\n" if ($debug); |
411 |
|
next; |
412 |
|
} |
413 |
|
|
414 |
|
$curr_backup++; |
415 |
|
|
416 |
my $tar_file = BackupPC::SearchLib::getGzipName($row->{'host'}, $row->{'share'}, $row->{'num'}); |
my $tar_file = BackupPC::SearchLib::getGzipName($row->{'host'}, $row->{'share'}, $row->{'num'}); |
417 |
|
|
418 |
# this will return -1 if file doesn't exist |
# this will return -1 if file doesn't exist |
420 |
|
|
421 |
print "# size: $size backup.size: ", $row->{inc_size},"\n" if ($opt{d}); |
print "# size: $size backup.size: ", $row->{inc_size},"\n" if ($opt{d}); |
422 |
|
|
423 |
if ( $row->{'inc_size'} != -1 && $size != -1 && $row->{'inc_size'} == $size) { |
if ( $row->{'inc_size'} != -1 && $size != -1 && $row->{'inc_size'} >= $size && $row->{parts} == $row->{backup_parts}) { |
424 |
if ($check) { |
if ($check) { |
425 |
tar_check($row->{'host'}, $row->{'share'}, $row->{'num'}, $tar_file) && next; |
tar_check($row->{'host'}, $row->{'share'}, $row->{'num'}, $tar_file) && next; |
426 |
} else { |
} else { |
428 |
} |
} |
429 |
} |
} |
430 |
|
|
431 |
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}, |
432 |
$curr_backup++; |
" ", strftime('%Y-%m-%d', localtime($row->{date})), " -> $tar_file"; |
433 |
|
|
434 |
my $t = time(); |
my $t = time(); |
435 |
|
|
436 |
# re-create archive? |
# re-create archive? |
437 |
my $cmd = qq{ $tarIncCreate -h "$row->{'host'}" -s "$row->{'share'}" -n $row->{'num'} -f }; |
my $cmd = qq[ $tarIncCreate -h "$row->{host}" -s "$row->{share}" -n $row->{num} -f ]; |
438 |
print STDERR "## $cmd\n" if ($debug); |
print STDERR "## $cmd\n" if ($debug); |
439 |
|
|
440 |
if (system($cmd) != 0) { |
if (system($cmd) != 0) { |
441 |
print STDERR " FAILED"; |
print STDERR " FAILED, marking this backup deleted"; |
442 |
|
backup_inc_deleted( $row->{backup_id} ); |
443 |
} |
} |
444 |
|
|
445 |
print ", dur: ",fmt_time(time() - $t), "\n"; |
print ", dur: ",fmt_time(time() - $t), "\n"; |