| 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 BackupPC::SearchLib; |
| 14 |
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); |
| 24 |
$path =~ s#/[^/]+$#/#; |
| 25 |
my $tarIncCreate = $path .= 'BackupPC_tarIncCreate'; |
| 26 |
|
| 27 |
die "can't find $tarIncCreate: $!\n" unless (-x $tarIncCreate); |
| 28 |
|
| 29 |
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; |
| 41 |
|
| 42 |
my $start_t = time(); |
| 43 |
|
| 44 |
my $t_fmt = '%Y-%m-%d %H:%M:%S'; |
| 45 |
|
| 46 |
my $hosts; |
| 47 |
my $bpc = BackupPC::Lib->new || die; |
| 48 |
my %Conf = $bpc->Conf(); |
| 49 |
my $TopDir = $bpc->TopDir(); |
| 50 |
my $beenThere = {}; |
| 51 |
|
| 52 |
my $dsn = $Conf{SearchDSN} || die "Need SearchDSN in config.pl\n"; |
| 53 |
my $user = $Conf{SearchUser} || ''; |
| 54 |
|
| 55 |
my $dbh = DBI->connect($dsn, $user, "", { RaiseError => 1, AutoCommit => 0 }); |
| 56 |
|
| 57 |
my $tar_dir = $Conf{InstallDir}.'/'.$Conf{GzipTempDir}; |
| 58 |
|
| 59 |
die "problem with $tar_dir, check GzipTempDir in configuration\n" unless (-d $tar_dir && -w $tar_dir); |
| 60 |
|
| 61 |
#---- subs ---- |
| 62 |
|
| 63 |
sub fmt_time { |
| 64 |
my $t = shift || return; |
| 65 |
my $out = ""; |
| 66 |
my ($ss,$mm,$hh) = gmtime($t); |
| 67 |
$out .= "${hh}h" if ($hh); |
| 68 |
$out .= sprintf("%02d:%02d", $mm,$ss); |
| 69 |
return $out; |
| 70 |
} |
| 71 |
|
| 72 |
sub curr_time { |
| 73 |
return strftime($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 |
| 159 |
|
| 160 |
my $sth = $dbh->prepare( qq{ |
| 161 |
|
| 162 |
select |
| 163 |
backups.id as backup_id, |
| 164 |
hosts.name as host, |
| 165 |
shares.name as share, |
| 166 |
backups.num as num, |
| 167 |
inc_size, |
| 168 |
parts |
| 169 |
from backups |
| 170 |
join shares on backups.hostid = shares.hostid |
| 171 |
and shares.id = backups.shareid |
| 172 |
join hosts on shares.hostid = hosts.id |
| 173 |
where not inc_deleted |
| 174 |
order by backups.date |
| 175 |
|
| 176 |
} ); |
| 177 |
|
| 178 |
my $sth_inc_size = $dbh->prepare(qq{ update backups set inc_size = ?, parts = ? where id = ? }); |
| 179 |
my $sth_inc_deleted = $dbh->prepare(qq{ update backups set inc_deleted = ? where id = ? }); |
| 180 |
|
| 181 |
%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) { |
| 188 |
my $tar_file = BackupPC::SearchLib::getGzipName($row->{'host'}, $row->{'share'}, $row->{'num'}); |
| 189 |
|
| 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 $t = time(); |
| 197 |
|
| 198 |
# re-create archive? |
| 199 |
if ($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: $?"; |
| 207 |
|
| 208 |
rename("${tar_dir}/${tar_file}.tmp", "$tar_dir/$tar_file") or die "can't rename $tar_dir/$tar_file: $!"; |
| 209 |
|
| 210 |
$size = (stat( "$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 |
my $parts = int( ($size + $max_size - 1) / $max_size ); |
| 219 |
|
| 220 |
if (-d "$tar_dir/$tar_file" && $parts != $row->{'parts'}) { |
| 221 |
print " join"; |
| 222 |
tar_join("$tar_dir/$tar_file"); |
| 223 |
} |
| 224 |
|
| 225 |
if ($size > $max_size && ! -d "$tar_dir/$tar_file") { |
| 226 |
print " split/$parts"; |
| 227 |
my $in = my $out = "$tar_dir/$tar_file"; |
| 228 |
$out .= '.tmp'; |
| 229 |
rename $in, $out || die "can't rename $in: $!"; |
| 230 |
mkdir $in || die "can't mkdir $in: $!"; |
| 231 |
|
| 232 |
my $suffix_len = length("$parts"); |
| 233 |
system("$bin->{'split'} -d -b $max_size -a $suffix_len $out $in/part") == 0 or die "can't split $out: $?"; |
| 234 |
unlink $out || die "can't unlink $out: $!"; |
| 235 |
} |
| 236 |
|
| 237 |
$sth_inc_size->execute($size, $parts, $row->{'backup_id'}); |
| 238 |
$sth_inc_deleted->execute(0, $row->{'backup_id'}); |
| 239 |
|
| 240 |
printf(" %1.2f MB", ($size / 1024 / 1024)); |
| 241 |
|
| 242 |
} else { |
| 243 |
$sth_inc_deleted->execute(1, $row->{'backup_id'}); |
| 244 |
unlink "$tar_dir/$tar_file" || die "can't delete $tar_dir/$tar_file: $!\n"; |
| 245 |
print " EMPTY"; |
| 246 |
} |
| 247 |
print ", dur: ",fmt_time(time() - $t), "\n"; |
| 248 |
|
| 249 |
$dbh->commit; |
| 250 |
|
| 251 |
} |
| 252 |
|
| 253 |
undef $sth; |
| 254 |
$dbh->disconnect; |
Parent Directory
| 
Revision Log