12 |
use File::Pid; |
use File::Pid; |
13 |
use POSIX qw/strftime/; |
use POSIX qw/strftime/; |
14 |
use BackupPC::SearchLib; |
use BackupPC::SearchLib; |
15 |
|
use Cwd qw/abs_path/; |
16 |
|
|
17 |
use constant BPC_FTYPE_DIR => 5; |
use constant BPC_FTYPE_DIR => 5; |
18 |
use constant EST_CHUNK => 100000; |
use constant EST_CHUNK => 4096; |
19 |
|
|
20 |
|
# daylight saving time change offset for 1h |
21 |
|
my $dst_offset = 60 * 60; |
22 |
|
|
23 |
my $debug = 0; |
my $debug = 0; |
24 |
$|=1; |
$|=1; |
25 |
|
|
26 |
my $start_t = time(); |
my $start_t = time(); |
27 |
|
|
28 |
my $pidfile = new File::Pid; |
my $pid_path = abs_path($0); |
29 |
|
$pid_path =~ s/\W+/_/g; |
30 |
|
|
31 |
|
my $pidfile = new File::Pid({ |
32 |
|
file => "/tmp/$pid_path", |
33 |
|
}); |
34 |
|
|
35 |
if (my $pid = $pidfile->running ) { |
if (my $pid = $pidfile->running ) { |
36 |
die "$0 already running: $pid\n"; |
die "$0 already running: $pid\n"; |
38 |
$pidfile->remove; |
$pidfile->remove; |
39 |
$pidfile = new File::Pid; |
$pidfile = new File::Pid; |
40 |
} |
} |
|
$pidfile->write; |
|
41 |
print STDERR "$0 using pid ",$pidfile->pid," file ",$pidfile->file,"\n"; |
print STDERR "$0 using pid ",$pidfile->pid," file ",$pidfile->file,"\n"; |
42 |
|
$pidfile->write; |
43 |
|
|
44 |
my $t_fmt = '%Y-%m-%d %H:%M:%S'; |
my $t_fmt = '%Y-%m-%d %H:%M:%S'; |
45 |
|
|
52 |
my $dsn = $Conf{SearchDSN} || die "Need SearchDSN in config.pl\n"; |
my $dsn = $Conf{SearchDSN} || die "Need SearchDSN in config.pl\n"; |
53 |
my $user = $Conf{SearchUser} || ''; |
my $user = $Conf{SearchUser} || ''; |
54 |
|
|
55 |
my $use_hest = $Conf{HyperEstraierIndex}; |
my $index_node_url = $Conf{HyperEstraierIndex}; |
|
my ($index_path, $index_node_url) = BackupPC::SearchLib::getHyperEstraier_url($use_hest); |
|
56 |
|
|
57 |
my $dbh = DBI->connect($dsn, $user, "", { RaiseError => 1, AutoCommit => 0 }); |
my $dbh = DBI->connect($dsn, $user, "", { RaiseError => 1, AutoCommit => 0 }); |
58 |
|
|
59 |
my %opt; |
my %opt; |
60 |
|
|
61 |
if ( !getopts("cdm:v:ij", \%opt ) ) { |
if ( !getopts("cdm:v:ijfq", \%opt ) ) { |
62 |
print STDERR <<EOF; |
print STDERR <<EOF; |
63 |
usage: $0 [-c|-d] [-m num] [-v|-v level] [-i] |
usage: $0 [-c|-d] [-m num] [-v|-v level] [-i|-j|-f] |
64 |
|
|
65 |
Options: |
Options: |
66 |
-c create database on first use |
-c create database on first use |
69 |
-v num set verbosity (debug) level (default $debug) |
-v num set verbosity (debug) level (default $debug) |
70 |
-i update Hyper Estraier full text index |
-i update Hyper Estraier full text index |
71 |
-j update full text, don't check existing files |
-j update full text, don't check existing files |
72 |
|
-f don't do anything with full text index |
73 |
|
-q be quiet for hosts without changes |
74 |
|
|
75 |
Option -j is variation on -i. It will allow faster initial creation |
Option -j is variation on -i. It will allow faster initial creation |
76 |
of full-text index from existing database. |
of full-text index from existing database. |
77 |
|
|
78 |
|
Option -f will create database which is out of sync with full text index. You |
79 |
|
will have to re-run $0 with -i to fix it. |
80 |
|
|
81 |
EOF |
EOF |
82 |
exit 1; |
exit 1; |
83 |
} |
} |
85 |
if ($opt{v}) { |
if ($opt{v}) { |
86 |
print "Debug level at $opt{v}\n"; |
print "Debug level at $opt{v}\n"; |
87 |
$debug = $opt{v}; |
$debug = $opt{v}; |
88 |
|
} elsif ($opt{f}) { |
89 |
|
print "WARNING: disabling full-text index update. You need to re-run $0 -j !\n"; |
90 |
|
$index_node_url = undef; |
91 |
} |
} |
92 |
|
|
93 |
#---- subs ---- |
#---- subs ---- |
105 |
return strftime($t_fmt,localtime()); |
return strftime($t_fmt,localtime()); |
106 |
} |
} |
107 |
|
|
|
my $hest_db; |
|
108 |
my $hest_node; |
my $hest_node; |
109 |
|
|
|
sub signal { |
|
|
my($sig) = @_; |
|
|
if ($hest_db) { |
|
|
print "\nCaught a SIG$sig--syncing database and shutting down\n"; |
|
|
$hest_db->sync(); |
|
|
$hest_db->close(); |
|
|
} |
|
|
exit(0); |
|
|
} |
|
|
|
|
|
$SIG{'INT'} = \&signal; |
|
|
$SIG{'QUIT'} = \&signal; |
|
|
|
|
110 |
sub hest_update { |
sub hest_update { |
111 |
|
|
112 |
my ($host_id, $share_id, $num) = @_; |
my ($host_id, $share_id, $num) = @_; |
113 |
|
|
114 |
my $skip_check = $opt{j} && print STDERR "Skipping check for existing files -- this should be used only with initital import\n"; |
my $skip_check = $opt{j} && print STDERR "Skipping check for existing files -- this should be used only with initital import\n"; |
115 |
|
|
116 |
unless ($use_hest) { |
unless (defined($index_node_url)) { |
117 |
print STDERR "HyperEstraier support not enabled in configuration\n"; |
print STDERR "HyperEstraier support not enabled in configuration\n"; |
118 |
|
$index_node_url = 0; |
119 |
return; |
return; |
120 |
} |
} |
121 |
|
|
122 |
print curr_time," updating HyperEstraier:"; |
print curr_time," updating Hyper Estraier:"; |
123 |
|
|
124 |
my $t = time(); |
my $t = time(); |
125 |
|
|
126 |
my $offset = 0; |
my $offset = 0; |
127 |
my $added = 0; |
my $added = 0; |
128 |
|
|
129 |
print " opening index $use_hest"; |
if ($index_node_url) { |
130 |
if ($index_path) { |
print " opening index $index_node_url"; |
131 |
$hest_db = HyperEstraier::Database->new(); |
$hest_node ||= Search::Estraier::Node->new( |
132 |
$hest_db->open($TopDir . $index_path, $HyperEstraier::Database::DBWRITER | $HyperEstraier::Database::DBCREAT); |
url => $index_node_url, |
133 |
print " directly"; |
user => 'admin', |
134 |
} elsif ($index_node_url) { |
passwd => 'admin', |
135 |
$hest_node ||= HyperEstraier::Node->new($index_node_url); |
croak_on_error => 1, |
136 |
$hest_node->set_auth('admin', 'admin'); |
); |
137 |
print " via node URL"; |
print " via node URL"; |
|
} else { |
|
|
die "don't know how to use HyperEstraier Index $use_hest"; |
|
138 |
} |
} |
|
print " increment is " . EST_CHUNK . " files:"; |
|
139 |
|
|
140 |
my $results = 0; |
my $results = 0; |
141 |
|
|
182 |
|
|
183 |
if ($results == 0) { |
if ($results == 0) { |
184 |
print " - no new files\n"; |
print " - no new files\n"; |
185 |
last; |
return; |
186 |
|
} else { |
187 |
|
print "..."; |
188 |
} |
} |
189 |
|
|
190 |
sub fmt_date { |
sub fmt_date { |
196 |
|
|
197 |
while (my $row = $sth->fetchrow_hashref()) { |
while (my $row = $sth->fetchrow_hashref()) { |
198 |
|
|
199 |
my $fid = $row->{'fid'} || die "no fid?"; |
my $uri = $row->{hname} . ':' . $row->{sname} . '#' . $row->{backupnum} . ' ' . $row->{filepath}; |
200 |
my $uri = 'file:///' . $fid; |
unless ($skip_check && $hest_node) { |
201 |
|
my $id = $hest_node->uri_to_id($uri); |
202 |
unless ($skip_check) { |
next if ($id && $id == -1); |
|
my $id = ($hest_db || $hest_node)->uri_to_id($uri); |
|
|
next unless ($id == -1); |
|
203 |
} |
} |
204 |
|
|
205 |
# create a document object |
# create a document object |
206 |
my $doc = HyperEstraier::Document->new; |
my $doc = Search::Estraier::Document->new; |
207 |
|
|
208 |
# add attributes to the document object |
# add attributes to the document object |
209 |
$doc->add_attr('@uri', $uri); |
$doc->add_attr('@uri', $uri); |
210 |
|
|
211 |
foreach my $c (@{ $sth->{NAME} }) { |
foreach my $c (@{ $sth->{NAME} }) { |
212 |
$doc->add_attr($c, $row->{$c}) if ($row->{$c}); |
print STDERR "attr $c = $row->{$c}\n" if ($debug > 2); |
213 |
|
$doc->add_attr($c, $row->{$c}) if (defined($row->{$c})); |
214 |
} |
} |
215 |
|
|
216 |
#$doc->add_attr('@cdate', fmt_date($row->{'date'})); |
#$doc->add_attr('@cdate', fmt_date($row->{'date'})); |
224 |
print STDERR $doc->dump_draft,"\n" if ($debug > 1); |
print STDERR $doc->dump_draft,"\n" if ($debug > 1); |
225 |
|
|
226 |
# register the document object to the database |
# register the document object to the database |
227 |
if ($hest_db) { |
if ($hest_node) { |
|
$hest_db->put_doc($doc, $HyperEstraier::Database::PDCLEAN); |
|
|
} elsif ($hest_node) { |
|
228 |
$hest_node->put_doc($doc); |
$hest_node->put_doc($doc); |
229 |
} else { |
} else { |
230 |
die "not supported"; |
die "not supported"; |
232 |
$added++; |
$added++; |
233 |
} |
} |
234 |
|
|
235 |
print " $added"; |
print "$added"; |
|
$hest_db->sync() if ($index_path); |
|
236 |
|
|
237 |
$offset += EST_CHUNK; |
$offset += EST_CHUNK; |
238 |
|
|
239 |
} while ($results == EST_CHUNK); |
} while ($results == EST_CHUNK); |
240 |
|
|
|
if ($index_path) { |
|
|
print ", close"; |
|
|
$hest_db->close(); |
|
|
} |
|
|
|
|
241 |
my $dur = (time() - $t) || 1; |
my $dur = (time() - $t) || 1; |
242 |
printf(" [%.2f/s dur: %s]\n", |
printf(" [%.2f/s dur: %s]\n", |
243 |
( $added / $dur ), |
( $added / $dur ), |
249 |
|
|
250 |
|
|
251 |
## update index ## |
## update index ## |
252 |
if (($opt{i} || $opt{j} || ($index_path && ! -e $index_path)) && !$opt{c}) { |
if ( ( $opt{i} || $opt{j} ) && !$opt{c} ) { |
253 |
# update all |
# update all |
254 |
print "force update of HyperEstraier index "; |
print "force update of Hyper Estraier index "; |
|
print "importing existing data" unless (-e $index_path); |
|
255 |
print "by -i flag" if ($opt{i}); |
print "by -i flag" if ($opt{i}); |
256 |
print "by -j flag" if ($opt{j}); |
print "by -j flag" if ($opt{j}); |
257 |
print "\n"; |
print "\n"; |
302 |
size bigint not null, |
size bigint not null, |
303 |
inc_size bigint not null default -1, |
inc_size bigint not null default -1, |
304 |
inc_deleted boolean default false, |
inc_deleted boolean default false, |
305 |
parts integer not null default 1, |
parts integer not null default 0, |
306 |
PRIMARY KEY(id) |
PRIMARY KEY(id) |
307 |
); |
); |
308 |
|
|
336 |
|
|
337 |
create table archive_burned ( |
create table archive_burned ( |
338 |
archive_id int references archive(id), |
archive_id int references archive(id), |
339 |
date date default now(), |
date timestamp default now(), |
340 |
part int not null default 1, |
part int not null default 1, |
341 |
|
copy int not null default 1, |
342 |
iso_size bigint default -1 |
iso_size bigint default -1 |
343 |
); |
); |
344 |
|
|
345 |
|
create table backup_parts ( |
346 |
|
id serial, |
347 |
|
backup_id int references backups(id), |
348 |
|
part_nr int not null check (part_nr > 0), |
349 |
|
tar_size bigint not null check (tar_size > 0), |
350 |
|
size bigint not null check (size > 0), |
351 |
|
md5 text not null, |
352 |
|
items int not null check (items > 0), |
353 |
|
date timestamp default now(), |
354 |
|
primary key(id) |
355 |
|
); |
356 |
}); |
}); |
357 |
|
|
358 |
print "creating indexes: "; |
print "creating indexes: "; |
371 |
files:size |
files:size |
372 |
archive:dvd_nr |
archive:dvd_nr |
373 |
archive_burned:archive_id |
archive_burned:archive_id |
374 |
|
backup_parts:backup_id,part_nr |
375 |
)) { |
)) { |
376 |
do_index($index); |
do_index($index); |
377 |
} |
} |
382 |
$dbh->do( qq{ CREATE SEQUENCE $seq } ); |
$dbh->do( qq{ CREATE SEQUENCE $seq } ); |
383 |
} |
} |
384 |
|
|
385 |
|
print " creating triggers "; |
386 |
|
$dbh->do( qq{ |
387 |
|
create or replace function backup_parts_check() returns trigger as ' |
388 |
|
declare |
389 |
|
b_parts integer; |
390 |
|
b_counted integer; |
391 |
|
b_id integer; |
392 |
|
begin |
393 |
|
if (TG_OP=''UPDATE'') then |
394 |
|
b_id := old.id; |
395 |
|
b_parts := old.parts; |
396 |
|
elsif (TG_OP = ''INSERT'') then |
397 |
|
b_id := new.id; |
398 |
|
b_parts := new.parts; |
399 |
|
end if; |
400 |
|
b_counted := (select count(*) from backup_parts where backup_id = b_id); |
401 |
|
if ( b_parts != b_counted ) then |
402 |
|
raise exception ''Update of backup % aborted, requested % parts and there are really % parts'', b_id, b_parts, b_counted; |
403 |
|
end if; |
404 |
|
return null; |
405 |
|
end; |
406 |
|
' language plpgsql; |
407 |
|
|
408 |
|
create trigger do_backup_parts_check |
409 |
|
after insert or update or delete on backups |
410 |
|
for each row execute procedure backup_parts_check(); |
411 |
|
}); |
412 |
|
|
413 |
print "...\n"; |
print "...\n"; |
414 |
|
|
453 |
|
|
454 |
$sth->{insert_backups} = $dbh->prepare(qq{ |
$sth->{insert_backups} = $dbh->prepare(qq{ |
455 |
INSERT INTO backups (hostID, num, date, type, shareid, size) |
INSERT INTO backups (hostID, num, date, type, shareid, size) |
456 |
VALUES (?,?,?,?,?,?) |
VALUES (?,?,?,?,?,-1) |
457 |
|
}); |
458 |
|
|
459 |
|
$sth->{update_backups_size} = $dbh->prepare(qq{ |
460 |
|
UPDATE backups SET size = ? |
461 |
|
WHERE hostID = ? and num = ? and date = ? and type =? and shareid = ? |
462 |
}); |
}); |
463 |
|
|
464 |
$sth->{insert_files} = $dbh->prepare(qq{ |
$sth->{insert_files} = $dbh->prepare(qq{ |
467 |
VALUES (?,?,?,?,?,?,?) |
VALUES (?,?,?,?,?,?,?) |
468 |
}); |
}); |
469 |
|
|
470 |
foreach my $host_key (keys %{$hosts}) { |
my @hosts = keys %{$hosts}; |
471 |
|
my $host_nr = 0; |
472 |
|
|
473 |
|
foreach my $host_key (@hosts) { |
474 |
|
|
475 |
my $hostname = $hosts->{$host_key}->{'host'} || die "can't find host for $host_key"; |
my $hostname = $hosts->{$host_key}->{'host'} || die "can't find host for $host_key"; |
476 |
|
|
485 |
$hostID = $dbh->last_insert_id(undef,undef,'hosts',undef); |
$hostID = $dbh->last_insert_id(undef,undef,'hosts',undef); |
486 |
} |
} |
487 |
|
|
488 |
print "host ".$hosts->{$host_key}->{'host'}.": "; |
$host_nr++; |
|
|
|
489 |
# get backups for a host |
# get backups for a host |
490 |
my @backups = $bpc->BackupInfoRead($hostname); |
my @backups = $bpc->BackupInfoRead($hostname); |
491 |
my $incs = scalar @backups; |
my $incs = scalar @backups; |
|
print "$incs increments\n"; |
|
492 |
|
|
493 |
|
my $host_header = sprintf("host %s [%d/%d]: %d increments\n", |
494 |
|
$hosts->{$host_key}->{'host'}, |
495 |
|
$host_nr, |
496 |
|
($#hosts + 1), |
497 |
|
$incs |
498 |
|
); |
499 |
|
print $host_header unless ($opt{q}); |
500 |
|
|
501 |
my $inc_nr = 0; |
my $inc_nr = 0; |
502 |
$beenThere = {}; |
$beenThere = {}; |
503 |
|
|
509 |
my $backupNum = $backup->{'num'}; |
my $backupNum = $backup->{'num'}; |
510 |
my @backupShares = (); |
my @backupShares = (); |
511 |
|
|
512 |
printf("%-10s %2d/%-2d #%-2d %s %5s/%5s files (date: %s dur: %s)\n", |
my $share_header = sprintf("%-10s %2d/%-2d #%-2d %s %5s/%5s files (date: %s dur: %s)\n", |
513 |
$hosts->{$host_key}->{'host'}, |
$hosts->{$host_key}->{'host'}, |
514 |
$inc_nr, $incs, $backupNum, |
$inc_nr, $incs, $backupNum, |
515 |
$backup->{type} || '?', |
$backup->{type} || '?', |
517 |
strftime($t_fmt,localtime($backup->{startTime})), |
strftime($t_fmt,localtime($backup->{startTime})), |
518 |
fmt_time($backup->{endTime} - $backup->{startTime}) |
fmt_time($backup->{endTime} - $backup->{startTime}) |
519 |
); |
); |
520 |
|
print $share_header unless ($opt{q}); |
521 |
|
|
522 |
my $files = BackupPC::View->new($bpc, $hostname, \@backups, 1); |
my $files = BackupPC::View->new($bpc, $hostname, \@backups, 1); |
523 |
foreach my $share ($files->shareList($backupNum)) { |
foreach my $share ($files->shareList($backupNum)) { |
531 |
# skip if allready in database! |
# skip if allready in database! |
532 |
next if ($count > 0); |
next if ($count > 0); |
533 |
|
|
534 |
|
# dump host and share header for -q |
535 |
|
if ($opt{q}) { |
536 |
|
if ($host_header) { |
537 |
|
print $host_header; |
538 |
|
$host_header = undef; |
539 |
|
} |
540 |
|
print $share_header; |
541 |
|
} |
542 |
|
|
543 |
# dump some log |
# dump some log |
544 |
print curr_time," ", $share; |
print curr_time," ", $share; |
545 |
|
|
|
my ($f, $nf, $d, $nd, $size) = recurseDir($bpc, $hostname, $files, $backupNum, $share, "", $shareID); |
|
|
|
|
546 |
$sth->{insert_backups}->execute( |
$sth->{insert_backups}->execute( |
547 |
$hostID, |
$hostID, |
548 |
$backupNum, |
$backupNum, |
549 |
$backup->{'endTime'}, |
$backup->{'endTime'}, |
550 |
substr($backup->{'type'},0,4), |
substr($backup->{'type'},0,4), |
551 |
$shareID, |
$shareID, |
|
$size, |
|
552 |
); |
); |
553 |
|
|
554 |
print " commit"; |
my ($f, $nf, $d, $nd, $size) = recurseDir($bpc, $hostname, $files, $backupNum, $share, "", $shareID); |
555 |
$dbh->commit(); |
|
556 |
|
eval { |
557 |
|
$sth->{update_backups_size}->execute( |
558 |
|
$size, |
559 |
|
$hostID, |
560 |
|
$backupNum, |
561 |
|
$backup->{'endTime'}, |
562 |
|
substr($backup->{'type'},0,4), |
563 |
|
$shareID, |
564 |
|
); |
565 |
|
print " commit"; |
566 |
|
$dbh->commit(); |
567 |
|
}; |
568 |
|
if ($@) { |
569 |
|
print " rollback"; |
570 |
|
$dbh->rollback(); |
571 |
|
} |
572 |
|
|
573 |
my $dur = (time() - $t) || 1; |
my $dur = (time() - $t) || 1; |
574 |
printf(" %d/%d files %d/%d dirs %0.2f MB [%.2f/s dur: %s]\n", |
printf(" %d/%d files %d/%d dirs %0.2f MB [%.2f/s dur: %s]\n", |
631 |
SELECT 1 FROM files |
SELECT 1 FROM files |
632 |
WHERE shareID = ? and |
WHERE shareID = ? and |
633 |
path = ? and |
path = ? and |
634 |
date = ? and |
size = ? and |
635 |
size = ? |
( date = ? or date = ? or date = ? ) |
636 |
LIMIT 1 |
LIMIT 1 |
637 |
}); |
}); |
638 |
|
|
639 |
my @param = ($shareID,$path,$date,$size); |
my @param = ($shareID,$path,$size,$date, $date-$dst_offset, $date+$dst_offset); |
640 |
$sth->{file_in_db}->execute(@param); |
$sth->{file_in_db}->execute(@param); |
641 |
my $rows = $sth->{file_in_db}->rows; |
my $rows = $sth->{file_in_db}->rows; |
642 |
print STDERR "## found_in_db($shareID,$path,$date,$size) ",( $rows ? '+' : '-' ), join(" ",@param), "\n" if ($debug >= 3); |
print STDERR "## found_in_db($shareID,$path,$date,$size) ",( $rows ? '+' : '-' ), join(" ",@param), "\n" if ($debug >= 3); |
686 |
$filesInBackup->{$path_key}->{'size'} |
$filesInBackup->{$path_key}->{'size'} |
687 |
)); |
)); |
688 |
|
|
689 |
|
my $key_dst_prev = join(" ", ( |
690 |
|
$shareID, |
691 |
|
$dir, |
692 |
|
$path_key, |
693 |
|
$filesInBackup->{$path_key}->{'mtime'} - $dst_offset, |
694 |
|
$filesInBackup->{$path_key}->{'size'} |
695 |
|
)); |
696 |
|
|
697 |
|
my $key_dst_next = join(" ", ( |
698 |
|
$shareID, |
699 |
|
$dir, |
700 |
|
$path_key, |
701 |
|
$filesInBackup->{$path_key}->{'mtime'} + $dst_offset, |
702 |
|
$filesInBackup->{$path_key}->{'size'} |
703 |
|
)); |
704 |
|
|
705 |
my $found; |
my $found; |
706 |
if (! defined($beenThere->{$key}) && ! ($found = found_in_db($key, @data)) ) { |
if ( |
707 |
|
! defined($beenThere->{$key}) && |
708 |
|
! defined($beenThere->{$key_dst_prev}) && |
709 |
|
! defined($beenThere->{$key_dst_next}) && |
710 |
|
! ($found = found_in_db($key, @data)) |
711 |
|
) { |
712 |
print STDERR "# key: $key [", $beenThere->{$key},"]" if ($debug >= 2); |
print STDERR "# key: $key [", $beenThere->{$key},"]" if ($debug >= 2); |
713 |
|
|
714 |
if ($filesInBackup->{$path_key}->{'type'} == BPC_FTYPE_DIR) { |
if ($filesInBackup->{$path_key}->{'type'} == BPC_FTYPE_DIR) { |