use DBI;
use BackupPC::Lib;
use BackupPC::View;
+use BackupPC::Attrib qw/:all/;
use Data::Dumper;
use Time::HiRes qw/time/;
use POSIX qw/strftime/;
use BackupPC::SearchLib;
use Cwd qw/abs_path/;
use File::Which;
+use Archive::Tar::Streamed;
+use Algorithm::Diff;
+use Getopt::Std;
+
+# cludge: minimum .tar.gz size
+my $MIN_TAR_SIZE = 80;
my $path = abs_path($0);
$path =~ s#/[^/]+$#/#;
$bin->{$c} = which($c) || die "$0 needs $c, install it\n";
}
+my %opt;
+getopts("cd", \%opt );
+
+my $debug = $opt{d};
+my $check = $opt{c} && print STDERR "NOTICE: tar archive check forced\n";
-my $debug = 0;
$|=1;
my $start_t = time();
return strftime($t_fmt,localtime());
}
+sub tar_join($) {
+ my $filename = shift;
+
+ my $in = my $out = $filename;
+ $out .= '.tmp';
+
+ # FIXME I should really order parts manually!
+ system("cat $in/part* > $out && rm -Rf $in && mv $out $in") == 0 or die "can't join $in: $?";
+
+}
+
+sub tar_check($$$$) {
+ my ($host,$share,$num,$filename) = @_;
+
+ if ($debug) {
+ print STDERR " {{ CHECK: ${host}:${share}#${num} and $filename";
+ } else {
+ print " check";
+ }
+
+ if (-d $filename) {
+ print STDERR ", joining";
+ tar_join($filename);
+ }
+
+ print STDERR ", opening" if ($debug);
+ open(my $fh, "gzip -cd $filename |") or die "can't open $filename: $!";
+ binmode($fh);
+ my $tar = Archive::Tar::Streamed->new($fh);
+
+ print STDERR ", tar" if ($debug);
+ my @tar_files;
+ while(my $entry = $tar->next) {
+ push @tar_files, $entry->name;
+ }
+ @tar_files = sort @tar_files;
+ print STDERR " ",($#tar_files + 1), " files" if ($debug);
+
+ print STDERR ", database" if ($debug);
+
+ my $sth = $dbh->prepare(qq{
+ SELECT path,type
+ FROM files
+ JOIN shares on shares.id = shareid
+ JOIN hosts on hosts.id = shares.hostid
+ WHERE hosts.name = ? and shares.name = ? and backupnum = ?
+ });
+ $sth->execute($host, $share, $num);
+ my @db_files;
+ while( my $row = $sth->fetchrow_hashref ) {
+
+ my $path = $row->{'path'} || die "no path?";
+ $path =~ s#^/#./#;
+ $path .= '/' if ($row->{'type'} == BPC_FTYPE_DIR);
+ push @db_files, $path;
+ }
+
+ print STDERR " ",($#db_files + 1), " files, diff" if ($debug);
+
+ @db_files = sort @db_files;
+
+ my $same = 1;
+ if ($#tar_files != $#db_files) {
+ $same = 0;
+ print STDERR " NUMBER" if ($debug);
+ } else {
+ my $diff = Algorithm::Diff->new(\@tar_files, \@db_files);
+ while ( $diff->Next() ) {
+ next if $diff->Same();
+ $same = 0;
+ print "< $_\n" for $diff->Items(1);
+ print "> $_\n" for $diff->Items(2);
+ }
+ }
+
+ print " ",($same ? 'ok' : 'DIFFERENT');
+ print STDERR " }} " if ($debug);
+
+ return $same;
+}
+
+
#----- main
my $sth = $dbh->prepare( qq{
} );
-$sth->execute();
-
my $sth_inc_size = $dbh->prepare(qq{ update backups set inc_size = ?, parts = ? where id = ? });
my $sth_inc_deleted = $dbh->prepare(qq{ update backups set inc_deleted = ? where id = ? });
%BackupPC::SearchLib::Conf = %Conf;
+$sth->execute();
+my $num_backups = $sth->rows;
+my $curr_backup = 1;
+
while (my $row = $sth->fetchrow_hashref) {
my $tar_file = BackupPC::SearchLib::getGzipName($row->{'host'}, $row->{'share'}, $row->{'num'});
# this will return -1 if file doesn't exist
my $size = BackupPC::SearchLib::get_tgz_size_by_name($tar_file);
- print curr_time, " ", $row->{'host'}, ":", $row->{'share'}, " #", $row->{'num'}, " -> $tar_file";
+ print curr_time, " $curr_backup/$num_backups ", $row->{'host'}, ":", $row->{'share'}, " #", $row->{'num'}, " -> $tar_file";
+ $curr_backup++;
my $t = time();
# re-create archive?
- if ($row->{'inc_size'} == -1 || $size == -1 || $row->{'inc_size'} != $size) {
- my $cmd = qq{rm -Rf $tar_dir/$tar_file && $tarIncCreate -h "$row->{'host'}" -s "$row->{'share'}" -n $row->{'num'} | $bin->{'gzip'} -9 > $tar_dir/$tar_file};
+ if ($row->{'inc_size'} == -1 || $size == -1 ||
+ $row->{'inc_size'} != $size ||
+ $check && ! tar_check($row->{'host'}, $row->{'share'}, $row->{'num'}, "$tar_dir/$tar_file")
+ ) {
+ 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};
print STDERR "## $cmd\n" if ($debug);
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: $!";
+
$size = (stat( "$tar_dir/$tar_file" ))[7];
}
- if ($size > 45) {
+ if ($size > $MIN_TAR_SIZE) {
my $max_size = $Conf{'MaxArchiveSize'} || die "problem with MaxArchieSize parametar";
$max_size *= 1024; # convert to bytes
if (-d "$tar_dir/$tar_file" && $parts != $row->{'parts'}) {
print " join";
-
- my $in = my $out = "$tar_dir/$tar_file";
- $out .= '.tmp';
-
- # FIXME I should really order parts manually!
- system("cat $in/part* > $out && rm -Rf $in && mv $out $in") == 0 or die "can't join $in: $?";
+ tar_join("$tar_dir/$tar_file");
}
if ($size > $max_size && ! -d "$tar_dir/$tar_file") {