1 |
#!/usr/local/bin/perl -w |
2 |
|
3 |
use strict; |
4 |
use lib "__INSTALLDIR__/lib"; |
5 |
|
6 |
use DBI; |
7 |
use BackupPC::Lib; |
8 |
use BackupPC::View; |
9 |
use BackupPC::Attrib qw/:all/; |
10 |
use Data::Dumper; |
11 |
use Time::HiRes qw/time/; |
12 |
use POSIX qw/strftime/; |
13 |
use Cwd qw/abs_path/; |
14 |
use File::Which; |
15 |
use Archive::Tar::Streamed; |
16 |
use Algorithm::Diff; |
17 |
use Getopt::Std; |
18 |
|
19 |
my $bpc = BackupPC::Lib->new || die "can't create BackupPC::Lib"; |
20 |
my %Conf = $bpc->Conf(); |
21 |
|
22 |
use BackupPC::SearchLib; |
23 |
%BackupPC::SearchLib::Conf = %Conf; |
24 |
|
25 |
# cludge: minimum .tar.gz size |
26 |
my $MIN_TAR_SIZE = 80; |
27 |
|
28 |
my $path = abs_path($0); |
29 |
$path =~ s#/[^/]+$#/#; |
30 |
my $tarIncCreate = $path .= 'BackupPC_tarIncCreate'; |
31 |
|
32 |
die "can't find $tarIncCreate: $!\n" unless (-x $tarIncCreate); |
33 |
|
34 |
my $bin; |
35 |
foreach my $c (qw/gzip split/) { |
36 |
$bin->{$c} = which($c) || die "$0 needs $c, install it\n"; |
37 |
} |
38 |
|
39 |
my %opt; |
40 |
getopts("cd", \%opt ); |
41 |
|
42 |
my $debug = $opt{d}; |
43 |
my $check = $opt{c} && print STDERR "NOTICE: tar archive check forced\n"; |
44 |
|
45 |
$|=1; |
46 |
|
47 |
my $start_t = time(); |
48 |
|
49 |
my $t_fmt = '%Y-%m-%d %H:%M:%S'; |
50 |
|
51 |
my $dsn = $Conf{SearchDSN} || die "Need SearchDSN in config.pl\n"; |
52 |
my $user = $Conf{SearchUser} || ''; |
53 |
|
54 |
my $dbh = DBI->connect($dsn, $user, "", { RaiseError => 1, AutoCommit => 0 }); |
55 |
|
56 |
my $tar_dir = $Conf{InstallDir}.'/'.$Conf{GzipTempDir}; |
57 |
|
58 |
die "problem with $tar_dir, check GzipTempDir in configuration\n" unless (-d $tar_dir && -w $tar_dir); |
59 |
|
60 |
#---- subs ---- |
61 |
|
62 |
sub fmt_time { |
63 |
my $t = shift || return; |
64 |
my $out = ""; |
65 |
my ($ss,$mm,$hh) = gmtime($t); |
66 |
$out .= "${hh}h" if ($hh); |
67 |
$out .= sprintf("%02d:%02d", $mm,$ss); |
68 |
return $out; |
69 |
} |
70 |
|
71 |
sub curr_time { |
72 |
return strftime($t_fmt,localtime()); |
73 |
} |
74 |
|
75 |
sub tar_join($) { |
76 |
my $filename = shift; |
77 |
|
78 |
my $in = my $out = $filename; |
79 |
$out .= '.tmp'; |
80 |
|
81 |
# FIXME I should really order parts manually! |
82 |
system("cat $in/part* > $out && rm -Rf $in && mv $out $in") == 0 or die "can't join $in: $?"; |
83 |
|
84 |
} |
85 |
|
86 |
sub tar_check($$$$) { |
87 |
my ($host,$share,$num,$filename) = @_; |
88 |
|
89 |
if ($debug) { |
90 |
print STDERR " {{ CHECK: ${host}:${share}#${num} and $filename"; |
91 |
} else { |
92 |
print " check"; |
93 |
} |
94 |
|
95 |
if (-d $filename) { |
96 |
print STDERR ", joining"; |
97 |
tar_join($filename); |
98 |
} |
99 |
|
100 |
print STDERR ", opening" if ($debug); |
101 |
open(my $fh, "gzip -cd $filename |") or die "can't open $filename: $!"; |
102 |
binmode($fh); |
103 |
my $tar = Archive::Tar::Streamed->new($fh); |
104 |
|
105 |
print STDERR ", tar" if ($debug); |
106 |
my @tar_files; |
107 |
while(my $entry = $tar->next) { |
108 |
push @tar_files, $entry->name; |
109 |
} |
110 |
@tar_files = sort @tar_files; |
111 |
print STDERR " ",($#tar_files + 1), " files" if ($debug); |
112 |
|
113 |
print STDERR ", database" if ($debug); |
114 |
|
115 |
my $sth = $dbh->prepare(qq{ |
116 |
SELECT path,type |
117 |
FROM files |
118 |
JOIN shares on shares.id = shareid |
119 |
JOIN hosts on hosts.id = shares.hostid |
120 |
WHERE hosts.name = ? and shares.name = ? and backupnum = ? |
121 |
}); |
122 |
$sth->execute($host, $share, $num); |
123 |
my @db_files; |
124 |
while( my $row = $sth->fetchrow_hashref ) { |
125 |
|
126 |
my $path = $row->{'path'} || die "no path?"; |
127 |
$path =~ s#^/#./#; |
128 |
$path .= '/' if ($row->{'type'} == BPC_FTYPE_DIR); |
129 |
push @db_files, $path; |
130 |
} |
131 |
|
132 |
print STDERR " ",($#db_files + 1), " files, diff" if ($debug); |
133 |
|
134 |
@db_files = sort @db_files; |
135 |
|
136 |
my $same = 1; |
137 |
if ($#tar_files != $#db_files) { |
138 |
$same = 0; |
139 |
print STDERR " NUMBER" if ($debug); |
140 |
} else { |
141 |
my $diff = Algorithm::Diff->new(\@tar_files, \@db_files); |
142 |
while ( $diff->Next() ) { |
143 |
next if $diff->Same(); |
144 |
$same = 0; |
145 |
print "< $_\n" for $diff->Items(1); |
146 |
print "> $_\n" for $diff->Items(2); |
147 |
} |
148 |
} |
149 |
|
150 |
print " ",($same ? 'ok' : 'DIFFERENT'); |
151 |
print STDERR " }} " if ($debug); |
152 |
|
153 |
return $same; |
154 |
} |
155 |
|
156 |
|
157 |
#----- main |
158 |
|
159 |
my $sth = $dbh->prepare( qq{ |
160 |
|
161 |
select |
162 |
backups.id as backup_id, |
163 |
hosts.name as host, |
164 |
shares.name as share, |
165 |
backups.num as num, |
166 |
inc_size, |
167 |
parts |
168 |
from backups |
169 |
join shares on backups.hostid = shares.hostid |
170 |
and shares.id = backups.shareid |
171 |
join hosts on shares.hostid = hosts.id |
172 |
where not inc_deleted |
173 |
order by backups.date |
174 |
|
175 |
} ); |
176 |
|
177 |
my $sth_inc_size = $dbh->prepare(qq{ update backups set inc_size = ?, parts = ? where id = ? }); |
178 |
my $sth_inc_deleted = $dbh->prepare(qq{ update backups set inc_deleted = ? where id = ? }); |
179 |
|
180 |
|
181 |
$sth->execute(); |
182 |
my $num_backups = $sth->rows; |
183 |
my $curr_backup = 1; |
184 |
|
185 |
while (my $row = $sth->fetchrow_hashref) { |
186 |
my $tar_file = BackupPC::SearchLib::getGzipName($row->{'host'}, $row->{'share'}, $row->{'num'}); |
187 |
|
188 |
# this will return -1 if file doesn't exist |
189 |
my $size = BackupPC::SearchLib::get_tgz_size_by_name($tar_file); |
190 |
|
191 |
print curr_time, " $curr_backup/$num_backups ", $row->{'host'}, ":", $row->{'share'}, " #", $row->{'num'}, " -> $tar_file"; |
192 |
$curr_backup++; |
193 |
|
194 |
my $t = time(); |
195 |
|
196 |
# re-create archive? |
197 |
if ($row->{'inc_size'} == -1 || $size == -1 || |
198 |
$row->{'inc_size'} != $size || |
199 |
$check && ! tar_check($row->{'host'}, $row->{'share'}, $row->{'num'}, "$tar_dir/$tar_file") |
200 |
) { |
201 |
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}; |
202 |
print STDERR "## $cmd\n" if ($debug); |
203 |
|
204 |
system($cmd) == 0 or die "failed: $?"; |
205 |
|
206 |
rename("${tar_dir}/${tar_file}.tmp", "$tar_dir/$tar_file") or die "can't rename $tar_dir/$tar_file: $!"; |
207 |
|
208 |
$size = (stat( "$tar_dir/$tar_file" ))[7]; |
209 |
} |
210 |
|
211 |
if ($size > $MIN_TAR_SIZE) { |
212 |
|
213 |
my $max_size = $Conf{'MaxArchiveSize'} || die "problem with MaxArchiveSize parametar"; |
214 |
$max_size *= 1024; # convert to bytes |
215 |
|
216 |
my $max_file_size = $Conf{'MaxArchiveFileSize'} || die "problem with MaxArchiveFileSize parametar"; |
217 |
$max_file_size *= 1024; # bytes |
218 |
|
219 |
if ($max_file_size > $max_size) { |
220 |
warn "MaxArchiveFileSize ($max_file_size) is bigger than MaxArchiveSize ($max_size)\n"; |
221 |
$max_file_size = $max_size; |
222 |
} |
223 |
|
224 |
# maximum file size on ISO image is 4Gb |
225 |
# this will require Linux kernel 2.6.8 or newer |
226 |
if ( $max_size > $max_file_size ) { |
227 |
$max_size = $max_file_size; |
228 |
} |
229 |
|
230 |
my $parts = int( ($size + $max_size - 1) / $max_size ); |
231 |
|
232 |
if (-d "$tar_dir/$tar_file" && $parts != $row->{'parts'}) { |
233 |
print " join"; |
234 |
tar_join("$tar_dir/$tar_file"); |
235 |
} |
236 |
|
237 |
if ($size > $max_size && ! -d "$tar_dir/$tar_file") { |
238 |
print " split/$parts"; |
239 |
my $in = my $out = "$tar_dir/$tar_file"; |
240 |
$out .= '.tmp'; |
241 |
rename $in, $out || die "can't rename $in: $!"; |
242 |
mkdir $in || die "can't mkdir $in: $!"; |
243 |
|
244 |
my $suffix_len = length("$parts"); |
245 |
system("$bin->{'split'} -d -b $max_size -a $suffix_len $out $in/part") == 0 or die "can't split $out: $?"; |
246 |
unlink $out || die "can't unlink $out: $!"; |
247 |
} |
248 |
|
249 |
$sth_inc_size->execute($size, $parts, $row->{'backup_id'}); |
250 |
$sth_inc_deleted->execute(0, $row->{'backup_id'}); |
251 |
|
252 |
printf(" %1.2f MB", ($size / 1024 / 1024)); |
253 |
|
254 |
} else { |
255 |
$sth_inc_deleted->execute(1, $row->{'backup_id'}); |
256 |
unlink "$tar_dir/$tar_file" || die "can't delete $tar_dir/$tar_file: $!\n"; |
257 |
print " EMPTY"; |
258 |
} |
259 |
print ", dur: ",fmt_time(time() - $t), "\n"; |
260 |
|
261 |
$dbh->commit; |
262 |
|
263 |
} |
264 |
|
265 |
undef $sth; |
266 |
$dbh->disconnect; |