6 |
use DBI; |
use DBI; |
7 |
use BackupPC::Lib; |
use BackupPC::Lib; |
8 |
use BackupPC::View; |
use BackupPC::View; |
9 |
|
use BackupPC::Attrib qw/:all/; |
10 |
use Data::Dumper; |
use Data::Dumper; |
11 |
use Time::HiRes qw/time/; |
use Time::HiRes qw/time/; |
|
use File::Pid; |
|
12 |
use POSIX qw/strftime/; |
use POSIX qw/strftime/; |
13 |
use BackupPC::SearchLib; |
use BackupPC::SearchLib; |
14 |
use Cwd qw/abs_path/; |
use Cwd qw/abs_path/; |
15 |
|
use File::Which; |
16 |
|
use Archive::Tar::Streamed; |
17 |
|
use Algorithm::Diff; |
18 |
|
use Getopt::Std; |
19 |
|
|
20 |
|
# cludge: minimum .tar.gz size |
21 |
|
my $MIN_TAR_SIZE = 80; |
22 |
|
|
23 |
my $path = abs_path($0); |
my $path = abs_path($0); |
24 |
$path =~ s#/[^/]+$#/#; |
$path =~ s#/[^/]+$#/#; |
26 |
|
|
27 |
die "can't find $tarIncCreate: $!\n" unless (-x $tarIncCreate); |
die "can't find $tarIncCreate: $!\n" unless (-x $tarIncCreate); |
28 |
|
|
29 |
my $debug = 0; |
my $bin; |
30 |
|
foreach my $c (qw/gzip split/) { |
31 |
|
$bin->{$c} = which($c) || die "$0 needs $c, install it\n"; |
32 |
|
} |
33 |
|
|
34 |
|
my %opt; |
35 |
|
getopts("cd", \%opt ); |
36 |
|
|
37 |
|
my $debug = $opt{d}; |
38 |
|
my $check = $opt{c} && print STDERR "NOTICE: tar archive check forced\n"; |
39 |
|
|
40 |
$|=1; |
$|=1; |
41 |
|
|
42 |
my $start_t = time(); |
my $start_t = time(); |
73 |
return strftime($t_fmt,localtime()); |
return strftime($t_fmt,localtime()); |
74 |
} |
} |
75 |
|
|
76 |
|
sub tar_join($) { |
77 |
|
my $filename = shift; |
78 |
|
|
79 |
|
my $in = my $out = $filename; |
80 |
|
$out .= '.tmp'; |
81 |
|
|
82 |
|
# FIXME I should really order parts manually! |
83 |
|
system("cat $in/part* > $out && rm -Rf $in && mv $out $in") == 0 or die "can't join $in: $?"; |
84 |
|
|
85 |
|
} |
86 |
|
|
87 |
|
sub tar_check($$$$) { |
88 |
|
my ($host,$share,$num,$filename) = @_; |
89 |
|
|
90 |
|
if ($debug) { |
91 |
|
print STDERR " {{ CHECK: ${host}:${share}#${num} and $filename"; |
92 |
|
} else { |
93 |
|
print " check"; |
94 |
|
} |
95 |
|
|
96 |
|
if (-d $filename) { |
97 |
|
print STDERR ", joining"; |
98 |
|
tar_join($filename); |
99 |
|
} |
100 |
|
|
101 |
|
print STDERR ", opening" if ($debug); |
102 |
|
open(my $fh, "gzip -cd $filename |") or die "can't open $filename: $!"; |
103 |
|
binmode($fh); |
104 |
|
my $tar = Archive::Tar::Streamed->new($fh); |
105 |
|
|
106 |
|
print STDERR ", tar" if ($debug); |
107 |
|
my @tar_files; |
108 |
|
while(my $entry = $tar->next) { |
109 |
|
push @tar_files, $entry->name; |
110 |
|
} |
111 |
|
@tar_files = sort @tar_files; |
112 |
|
print STDERR " ",($#tar_files + 1), " files" if ($debug); |
113 |
|
|
114 |
|
print STDERR ", database" if ($debug); |
115 |
|
|
116 |
|
my $sth = $dbh->prepare(qq{ |
117 |
|
SELECT path,type |
118 |
|
FROM files |
119 |
|
JOIN shares on shares.id = shareid |
120 |
|
JOIN hosts on hosts.id = shares.hostid |
121 |
|
WHERE hosts.name = ? and shares.name = ? and backupnum = ? |
122 |
|
}); |
123 |
|
$sth->execute($host, $share, $num); |
124 |
|
my @db_files; |
125 |
|
while( my $row = $sth->fetchrow_hashref ) { |
126 |
|
|
127 |
|
my $path = $row->{'path'} || die "no path?"; |
128 |
|
$path =~ s#^/#./#; |
129 |
|
$path .= '/' if ($row->{'type'} == BPC_FTYPE_DIR); |
130 |
|
push @db_files, $path; |
131 |
|
} |
132 |
|
|
133 |
|
print STDERR " ",($#db_files + 1), " files, diff" if ($debug); |
134 |
|
|
135 |
|
@db_files = sort @db_files; |
136 |
|
|
137 |
|
my $same = 1; |
138 |
|
if ($#tar_files != $#db_files) { |
139 |
|
$same = 0; |
140 |
|
print STDERR " NUMBER" if ($debug); |
141 |
|
} else { |
142 |
|
my $diff = Algorithm::Diff->new(\@tar_files, \@db_files); |
143 |
|
while ( $diff->Next() ) { |
144 |
|
next if $diff->Same(); |
145 |
|
$same = 0; |
146 |
|
print "< $_\n" for $diff->Items(1); |
147 |
|
print "> $_\n" for $diff->Items(2); |
148 |
|
} |
149 |
|
} |
150 |
|
|
151 |
|
print " ",($same ? 'ok' : 'DIFFERENT'); |
152 |
|
print STDERR " }} " if ($debug); |
153 |
|
|
154 |
|
return $same; |
155 |
|
} |
156 |
|
|
157 |
|
|
158 |
#----- main |
#----- main |
159 |
|
|
160 |
my $sth = $dbh->prepare( qq{ |
my $sth = $dbh->prepare( qq{ |
163 |
backups.id as backup_id, |
backups.id as backup_id, |
164 |
hosts.name as host, |
hosts.name as host, |
165 |
shares.name as share, |
shares.name as share, |
166 |
backups.num as num |
backups.num as num, |
167 |
|
inc_size, |
168 |
|
parts |
169 |
from backups |
from backups |
170 |
join shares on backups.hostid = shares.hostid |
join shares on backups.hostid = shares.hostid |
171 |
and shares.id = backups.shareid |
and shares.id = backups.shareid |
172 |
join hosts on shares.hostid = hosts.id |
join hosts on shares.hostid = hosts.id |
173 |
where inc_size < 0 and not inc_deleted |
where not inc_deleted |
174 |
order by backups.date |
order by backups.date |
175 |
|
|
176 |
} ); |
} ); |
177 |
|
|
178 |
$sth->execute(); |
my $sth_inc_size = $dbh->prepare(qq{ update backups set inc_size = ?, parts = ? where id = ? }); |
|
|
|
|
my $sth_inc_size = $dbh->prepare(qq{ update backups set inc_size = ? where id = ? }); |
|
179 |
my $sth_inc_deleted = $dbh->prepare(qq{ update backups set inc_deleted = ? where id = ? }); |
my $sth_inc_deleted = $dbh->prepare(qq{ update backups set inc_deleted = ? where id = ? }); |
180 |
|
|
181 |
%BackupPC::SearchLib::Conf = %Conf; |
%BackupPC::SearchLib::Conf = %Conf; |
182 |
|
|
183 |
|
$sth->execute(); |
184 |
|
my $num_backups = $sth->rows; |
185 |
|
my $curr_backup = 1; |
186 |
|
|
187 |
while (my $row = $sth->fetchrow_hashref) { |
while (my $row = $sth->fetchrow_hashref) { |
188 |
my $tar_file = BackupPC::SearchLib::getGzipName($row->{'host'}, $row->{'share'}, $row->{'num'}); |
my $tar_file = BackupPC::SearchLib::getGzipName($row->{'host'}, $row->{'share'}, $row->{'num'}); |
189 |
print curr_time, sprintf(" %s:%s %-3d ", $row->{'host'}, $row->{'share'}, $row->{'num'}), " -> $tar_file "; |
|
190 |
|
# this will return -1 if file doesn't exist |
191 |
|
my $size = BackupPC::SearchLib::get_tgz_size_by_name($tar_file); |
192 |
|
|
193 |
|
print curr_time, " $curr_backup/$num_backups ", $row->{'host'}, ":", $row->{'share'}, " #", $row->{'num'}, " -> $tar_file"; |
194 |
|
$curr_backup++; |
195 |
|
|
196 |
my $t = time(); |
my $t = time(); |
197 |
|
|
198 |
my $cmd = qq{$tarIncCreate -h "$row->{'host'}" -s "$row->{'share'}" -n $row->{'num'} | gzip -9 > $tar_dir/$tar_file}; |
# re-create archive? |
199 |
print STDERR "## $cmd\n" if ($debug); |
if ($row->{'inc_size'} == -1 || $size == -1 || |
200 |
|
$row->{'inc_size'} != $size || |
201 |
|
$check && ! tar_check($row->{'host'}, $row->{'share'}, $row->{'num'}, "$tar_dir/$tar_file") |
202 |
|
) { |
203 |
|
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}; |
204 |
|
print STDERR "## $cmd\n" if ($debug); |
205 |
|
|
206 |
|
system($cmd) == 0 or die "failed: $?"; |
207 |
|
|
208 |
system($cmd) == 0 or die "failed: $?"; |
rename("${tar_dir}/${tar_file}.tmp", "$tar_dir/$tar_file") or die "can't rename $tar_dir/$tar_file: $!"; |
209 |
|
|
210 |
my $size = (stat( "$tar_dir/$tar_file" ))[7]; |
$size = (stat( "$tar_dir/$tar_file" ))[7]; |
211 |
|
} |
212 |
|
|
213 |
|
if ($size > $MIN_TAR_SIZE) { |
214 |
|
|
215 |
|
my $max_size = $Conf{'MaxArchiveSize'} || die "problem with MaxArchieSize parametar"; |
216 |
|
$max_size *= 1024; # convert to bytes |
217 |
|
|
218 |
print " dur: ",fmt_time(time() - $t)," $size bytes"; |
# maximum file size on ISO image is 4Gb |
219 |
|
# this will require Linux kernel 2.6.8 or newer |
220 |
|
my $max_iso_file_size = 2^32 - 2048; |
221 |
|
if ( $max_size > $max_iso_file_size ) { |
222 |
|
$max_size = $max_iso_file_size; |
223 |
|
} |
224 |
|
|
225 |
|
my $parts = int( ($size + $max_size - 1) / $max_size ); |
226 |
|
|
227 |
|
if (-d "$tar_dir/$tar_file" && $parts != $row->{'parts'}) { |
228 |
|
print " join"; |
229 |
|
tar_join("$tar_dir/$tar_file"); |
230 |
|
} |
231 |
|
|
232 |
|
if ($size > $max_size && ! -d "$tar_dir/$tar_file") { |
233 |
|
print " split/$parts"; |
234 |
|
my $in = my $out = "$tar_dir/$tar_file"; |
235 |
|
$out .= '.tmp'; |
236 |
|
rename $in, $out || die "can't rename $in: $!"; |
237 |
|
mkdir $in || die "can't mkdir $in: $!"; |
238 |
|
|
239 |
|
my $suffix_len = length("$parts"); |
240 |
|
system("$bin->{'split'} -d -b $max_size -a $suffix_len $out $in/part") == 0 or die "can't split $out: $?"; |
241 |
|
unlink $out || die "can't unlink $out: $!"; |
242 |
|
} |
243 |
|
|
244 |
if ($size > 45) { |
$sth_inc_size->execute($size, $parts, $row->{'backup_id'}); |
|
$sth_inc_size->execute($size, $row->{'backup_id'}); |
|
245 |
$sth_inc_deleted->execute(0, $row->{'backup_id'}); |
$sth_inc_deleted->execute(0, $row->{'backup_id'}); |
246 |
|
|
247 |
|
printf(" %1.2f MB", ($size / 1024 / 1024)); |
248 |
|
|
249 |
} else { |
} else { |
250 |
$sth_inc_deleted->execute(1, $row->{'backup_id'}); |
$sth_inc_deleted->execute(1, $row->{'backup_id'}); |
251 |
unlink "$tar_dir/$tar_file" || die "can't delete $tar_dir/$tar_file: $!\n"; |
unlink "$tar_dir/$tar_file" || die "can't delete $tar_dir/$tar_file: $!\n"; |
252 |
print " EMPTY"; |
print " EMPTY"; |
253 |
} |
} |
254 |
print "\n"; |
print ", dur: ",fmt_time(time() - $t), "\n"; |
255 |
|
|
256 |
$dbh->commit; |
$dbh->commit; |
257 |
|
|