1 #============================================================= -*-perl-*-
3 # BackupPC::PoolWrite package
7 # This library defines a BackupPC::PoolWrite class for writing
8 # files to disk that are candidates for pooling. One instance
9 # of this class is used to write each file. The following steps
12 # - As the incoming data arrives, the first 1MB is buffered
13 # in memory so the MD5 digest can be computed.
15 # - A running comparison against all the candidate pool files
16 # (ie: those with the same MD5 digest, usually at most a single
17 # file) is done as new incoming data arrives. Up to $MaxFiles
18 # simultaneous files can be compared in parallel. This
19 # involves reading and uncompressing one or more pool files.
21 # - When a pool file no longer matches it is discarded from
22 # the search. If there are more than $MaxFiles candidates, one of
23 # the new candidates is added to the search, first checking
24 # that it matches up to the current point (this requires
25 # re-reading one of the other pool files).
27 # - When or if no pool files match then the new file is written
28 # to disk. This could occur many MB into the file. We don't
29 # need to buffer all this data in memory since we can copy it
30 # from the last matching pool file, up to the point where it
33 # - When all the new data is complete, if a pool file exactly
34 # matches then the file is simply created as a hardlink to
38 # Craig Barratt <cbarratt@users.sourceforge.net>
41 # Copyright (C) 2001-2003 Craig Barratt
43 # This program is free software; you can redistribute it and/or modify
44 # it under the terms of the GNU General Public License as published by
45 # the Free Software Foundation; either version 2 of the License, or
46 # (at your option) any later version.
48 # This program is distributed in the hope that it will be useful,
49 # but WITHOUT ANY WARRANTY; without even the implied warranty of
50 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
51 # GNU General Public License for more details.
53 # You should have received a copy of the GNU General Public License
54 # along with this program; if not, write to the Free Software
55 # Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
57 #========================================================================
59 # Version 3.0.0beta2, released 11 Nov 2006.
61 # See http://backuppc.sourceforge.net.
63 #========================================================================
65 package BackupPC::PoolWrite;
71 use BackupPC::FileZIO;
75 my($class, $bpc, $fileName, $fileSize, $compress) = @_;
78 fileName => $fileName,
79 fileSize => $fileSize,
81 compress => $compress,
92 $self->{hardLinkMax} = $bpc->ConfValue("HardLinkMax");
95 # Always unlink any current file in case it is already linked
97 unlink($fileName) if ( -f $fileName );
101 my $BufSize = 1048576; # 1MB or 2^20
102 my $MaxFiles = 20; # max number of compare files open at one time
106 my($a, $dataRef) = @_;
108 return if ( $a->{eof} );
109 $a->{data} .= $$dataRef if ( defined($dataRef) );
110 return if ( length($a->{data}) < $BufSize && defined($dataRef) );
113 # Correct the fileSize if it is wrong (rsync might transfer
114 # a file whose length is different to the length sent with the
115 # file list if the file changes between the file list sending
116 # and the file sending). Here we only catch the case where
117 # we haven't computed the digest (ie: we have written no more
118 # than $BufSize). We catch the big file case below.
120 if ( !defined($dataRef) && !defined($a->{digest})
121 && $a->{fileSize} != length($a->{data}) ) {
122 #my $newSize = length($a->{data});
123 #print("Fixing file size from $a->{fileSize} to $newSize\n");
124 $a->{fileSize} = length($a->{data});
127 if ( !defined($a->{digest}) && length($a->{data}) > 0 ) {
129 # build a list of all the candidate matching files
131 my $md5 = Digest::MD5->new;
132 $a->{fileSize} = length($a->{data})
133 if ( $a->{fileSize} < length($a->{data}) );
134 $a->{digest} = $a->{bpc}->Buffer2MD5($md5, $a->{fileSize}, \$a->{data});
135 if ( !defined($a->{base} = $a->{bpc}->MD52Path($a->{digest},
137 push(@{$a->{errors}}, "Unable to get path from '$a->{digest}'"
138 . " for $a->{fileName}\n");
140 while ( @{$a->{files}} < $MaxFiles ) {
142 my $fileName = $a->{fileCnt} < 0 ? $a->{base}
143 : "$a->{base}_$a->{fileCnt}";
144 last if ( !-f $fileName );
146 # Don't attempt to match pool files that already
147 # have too many hardlinks. Also, don't match pool
148 # files with only one link since starting in
149 # BackupPC v3.0, BackupPC_nightly could be running
150 # in parallel (and removing those files). This doesn't
151 # eliminate all possible race conditions, but just
152 # reduces the odds. Other design steps eliminate
153 # the remaining race conditions of linking vs
156 if ( (stat(_))[3] >= $a->{hardLinkMax}
158 || !defined($fh = BackupPC::FileZIO->open($fileName, 0,
163 push(@{$a->{files}}, {
171 # if there are no candidate files then we must write
172 # the new file to disk
174 if ( !@{$a->{files}} ) {
175 $a->{fhOut} = BackupPC::FileZIO->open($a->{fileName},
177 if ( !defined($a->{fhOut}) ) {
178 push(@{$a->{errors}}, "Unable to open $a->{fileName}"
183 my $dataLen = length($a->{data});
184 if ( !defined($a->{fhOut}) && length($a->{data}) > 0 ) {
186 # See if the new chunk of data continues to match the
189 for ( my $i = 0 ; $i < @{$a->{files}} ; $i++ ) {
191 my $fileName = $a->{fileCnt} < 0 ? $a->{base}
192 : "$a->{base}_$a->{fileCnt}";
193 if ( $dataLen > 0 ) {
194 # verify next $dataLen bytes from candidate file
195 my $n = $a->{files}[$i]->{fh}->read(\$d, $dataLen);
196 next if ( $n == $dataLen && $d eq $a->{data} );
198 # verify candidate file is at EOF
199 my $n = $a->{files}[$i]->{fh}->read(\$d, 100);
202 #print(" File $a->{files}[$i]->{name} doesn't match\n");
204 # this candidate file didn't match. Replace it
205 # with a new candidate file. We have to qualify
206 # any new candidate file by making sure that its
207 # first $a->{nWrite} bytes match, plus the next $dataLen
208 # bytes match $a->{data}.
210 while ( -f $fileName ) {
212 if ( (stat(_))[3] >= $a->{hardLinkMax}
213 || !defined($fh = BackupPC::FileZIO->open($fileName, 0,
216 #print(" Discarding $fileName (open failed)\n");
217 $fileName = "$a->{base}_$a->{fileCnt}";
220 if ( !$a->{files}[$i]->{fh}->rewind() ) {
221 push(@{$a->{errors}},
222 "Unable to rewind $a->{files}[$i]->{name}"
225 $match = $a->filePartialCompare($a->{files}[$i]->{fh}, $fh,
226 $a->{nWrite}, $dataLen, \$a->{data});
228 $a->{files}[$i]->{fh}->close();
229 $a->{files}[$i]->{fh} = $fh,
230 $a->{files}[$i]->{name} = $fileName;
231 #print(" Found new candidate $fileName\n");
235 #print(" Discarding $fileName (no match)\n");
239 $fileName = "$a->{base}_$a->{fileCnt}";
243 # We couldn't find another candidate file
245 if ( @{$a->{files}} == 1 ) {
246 #print(" Exhausted matches, now writing\n");
247 $a->{fhOut} = BackupPC::FileZIO->open($a->{fileName},
249 if ( !defined($a->{fhOut}) ) {
250 push(@{$a->{errors}},
251 "Unable to open $a->{fileName}"
254 if ( !$a->{files}[$i]->{fh}->rewind() ) {
255 push(@{$a->{errors}},
257 . " $a->{files}[$i]->{name} for copy\n");
259 $a->filePartialCopy($a->{files}[$i]->{fh}, $a->{fhOut},
263 $a->{files}[$i]->{fh}->close();
264 splice(@{$a->{files}}, $i, 1);
269 if ( defined($a->{fhOut}) && $dataLen > 0 ) {
271 # if we are in writing mode then just write the data
273 my $n = $a->{fhOut}->write(\$a->{data});
274 if ( $n != $dataLen ) {
275 push(@{$a->{errors}}, "Unable to write $dataLen bytes to"
276 . " $a->{fileName} (got $n)\n");
279 $a->{nWrite} += $dataLen;
281 return if ( defined($dataRef) );
284 # We are at EOF, so finish up
289 # Make sure the fileSize was correct. See above for comments about
292 if ( $a->{nWrite} != $a->{fileSize} ) {
294 # Oops, fileSize was wrong, so our MD5 digest was wrong and our
295 # effort to match files likely failed. This is ugly, but our
296 # only choice at this point is to re-write the entire file with
297 # the correct length. We need to rename the file, open it for
298 # reading, and then re-write the file with the correct length.
301 #print("Doing big file fixup ($a->{fileSize} != $a->{nWrite})\n");
304 $a->{fileSize} = $a->{nWrite};
306 if ( defined($a->{fhOut}) ) {
307 if ( $a->{fileName} =~ /(.*)\// ) {
313 # Find a unique target temporary file name
316 while ( -f "$fileName/t$$.$i" ) {
319 $fileName = "$fileName/t$$.$i";
320 $a->{fhOut}->close();
321 if ( !rename($a->{fileName}, $fileName)
322 || !defined($fh = BackupPC::FileZIO->open($fileName, 0,
324 push(@{$a->{errors}}, "Can't rename $a->{fileName} -> $fileName"
325 . " or open during size fixup\n");
327 #print("Using temporary name $fileName\n");
328 } elsif ( defined($a->{files}) && defined($a->{files}[0]) ) {
330 # We haven't written anything yet, so just use the
331 # compare file to copy from.
333 $fh = $a->{files}[0]->{fh};
335 #print("Using compare file $a->{files}[0]->{name}\n");
337 if ( defined($fh) ) {
338 my $poolWrite = BackupPC::PoolWrite->new($a->{bpc}, $a->{fileName},
339 $a->{fileSize}, $a->{compress});
342 while ( $nRead < $a->{fileSize} ) {
343 my $thisRead = $a->{fileSize} - $nRead < $BufSize
344 ? $a->{fileSize} - $nRead : $BufSize;
346 my $n = $fh->read(\$data, $thisRead);
347 if ( $n != $thisRead ) {
348 push(@{$a->{errors}},
349 "Unable to read $thisRead bytes during resize"
350 . " from temp $fileName (got $n)\n");
353 $poolWrite->write(\$data);
357 unlink($fileName) if ( defined($fileName) );
358 if ( @{$a->{errors}} ) {
360 return (0, $a->{digest}, -s $a->{fileName}, $a->{errors});
362 return $poolWrite->close;
367 if ( $a->{fileSize} == 0 ) {
369 # Simply create an empty file
372 if ( !open(OUT, ">", $a->{fileName}) ) {
373 push(@{$a->{errors}}, "Can't open $a->{fileName} for empty"
379 # Close the compare files
381 foreach my $f ( @{$a->{files}} ) {
384 return (1, $a->{digest}, -s $a->{fileName}, $a->{errors});
385 } elsif ( defined($a->{fhOut}) ) {
386 $a->{fhOut}->close();
388 # Close the compare files
390 foreach my $f ( @{$a->{files}} ) {
393 return (0, $a->{digest}, -s $a->{fileName}, $a->{errors});
395 if ( @{$a->{files}} == 0 ) {
396 push(@{$a->{errors}}, "Botch, no matches on $a->{fileName}"
397 . " ($a->{digest})\n");
398 } elsif ( @{$a->{files}} > 1 ) {
400 # This is no longer a real error because $Conf{HardLinkMax}
401 # could be hit, thereby creating identical pool files
403 #my $str = "Unexpected multiple matches on"
404 # . " $a->{fileName} ($a->{digest})\n";
405 #for ( my $i = 0 ; $i < @{$a->{files}} ; $i++ ) {
406 # $str .= " -> $a->{files}[$i]->{name}\n";
408 #push(@{$a->{errors}}, $str);
410 for ( my $i = 0 ; $i < @{$a->{files}} ; $i++ ) {
411 if ( link($a->{files}[$i]->{name}, $a->{fileName}) ) {
412 #print(" Linked $a->{fileName} to $a->{files}[$i]->{name}\n");
414 # Close the compare files
416 foreach my $f ( @{$a->{files}} ) {
419 return (1, $a->{digest}, -s $a->{fileName}, $a->{errors});
423 # We were unable to link to the pool. Either we're at the
424 # hardlink max, or the pool file got deleted. Recover by
425 # writing the matching file, since we still have an open
428 for ( my $i = 0 ; $i < @{$a->{files}} ; $i++ ) {
429 if ( !$a->{files}[$i]->{fh}->rewind() ) {
430 push(@{$a->{errors}},
431 "Unable to rewind $a->{files}[$i]->{name}"
432 . " for copy after link fail\n");
435 $a->{fhOut} = BackupPC::FileZIO->open($a->{fileName},
437 if ( !defined($a->{fhOut}) ) {
438 push(@{$a->{errors}},
439 "Unable to open $a->{fileName}"
440 . " for writing after link fail\n");
442 $a->filePartialCopy($a->{files}[$i]->{fh}, $a->{fhOut},
449 # Close the compare files
451 foreach my $f ( @{$a->{files}} ) {
454 return (0, $a->{digest}, -s $a->{fileName}, $a->{errors});
459 # Finish writing: pass undef dataRef to write so it can do all
460 # the work. Returns a 4 element array:
462 # (existingFlag, digestString, outputFileLength, errorList)
468 return $a->write(undef);
478 if ( defined($a->{fhOut}) ) {
479 $a->{fhOut}->close();
480 unlink($a->{fileName});
482 foreach my $f ( @{$a->{files}} ) {
489 # Copy $nBytes from files $fhIn to $fhOut.
493 my($a, $fhIn, $fhOut, $nBytes) = @_;
496 while ( $nRead < $nBytes ) {
497 my $thisRead = $nBytes - $nRead < $BufSize
498 ? $nBytes - $nRead : $BufSize;
500 my $n = $fhIn->read(\$data, $thisRead);
501 if ( $n != $thisRead ) {
502 push(@{$a->{errors}},
503 "Unable to read $thisRead bytes from "
504 . $fhIn->name . " (got $n)\n");
507 $n = $fhOut->write(\$data, $thisRead);
508 if ( $n != $thisRead ) {
509 push(@{$a->{errors}},
510 "Unable to write $thisRead bytes to "
511 . $fhOut->name . " (got $n)\n");
519 # Compare $nBytes from files $fh0 and $fh1, and also compare additional
520 # $extra bytes from $fh1 to $$extraData.
522 sub filePartialCompare
524 my($a, $fh0, $fh1, $nBytes, $extra, $extraData) = @_;
528 while ( $nRead < $nBytes ) {
529 my $thisRead = $nBytes - $nRead < $BufSize
530 ? $nBytes - $nRead : $BufSize;
531 $n = $fh0->read(\$data0, $thisRead);
532 if ( $n != $thisRead ) {
533 push(@{$a->{errors}}, "Unable to read $thisRead bytes from "
534 . $fh0->name . " (got $n)\n");
537 $n = $fh1->read(\$data1, $thisRead);
538 return 0 if ( $n < $thisRead || $data0 ne $data1 );
542 # verify additional bytes
543 $n = $fh1->read(\$data1, $extra);
544 return 0 if ( $n != $extra || $data1 ne $$extraData );
547 $n = $fh1->read(\$data1, 100);
548 return 0 if ( $n != 0 );
554 # LinkOrCopy() does a hardlink from oldFile to newFile.
556 # If that fails (because there are too many links on oldFile)
557 # then oldFile is copied to newFile, and the pool stats are
558 # returned to be added to the new file list. That allows
559 # BackupPC_link to try again, and to create a new pool file
564 my($bpc, $oldFile, $oldFileComp, $newFile, $newFileComp) = @_;
567 unlink($newFile) if ( -f $newFile );
569 # Try to link if hardlink limit is ok, and compression types
572 return (1, undef) if ( (stat($oldFile))[3] < $bpc->{Conf}{HardLinkMax}
573 && !$oldFileComp == !$newFileComp
574 && link($oldFile, $newFile) );
576 # There are too many links on oldFile, or compression
577 # type if different, so now we have to copy it.
579 # We need to compute the file size, which is expensive
580 # since we need to read the file twice. That's probably
581 # ok since the hardlink limit is rarely hit.
583 my $readFd = BackupPC::FileZIO->open($oldFile, 0, $oldFileComp);
584 if ( !defined($readFd) ) {
585 return (0, undef, undef, undef, ["LinkOrCopy: can't open $oldFile"]);
587 while ( $readFd->read(\$data, $BufSize) > 0 ) {
588 $nRead += length($data);
592 my $poolWrite = BackupPC::PoolWrite->new($bpc, $newFile,
593 $nRead, $newFileComp);
594 while ( $readFd->read(\$data, $BufSize) > 0 ) {
595 $poolWrite->write(\$data);
597 my($exists, $digest, $outSize, $errs) = $poolWrite->close;
599 return ($exists, $digest, $nRead, $outSize, $errs);