use BackupPC::Lib;
use BackupPC::View;
use BackupPC::Attrib qw/:all/;
-#use Data::Dumper;
use Data::Dump qw(dump);
use Time::HiRes qw/time/;
use POSIX qw/strftime/;
use Cwd qw/abs_path/;
-#use Archive::Tar::Streamed;
-#use Algorithm::Diff;
-use Getopt::Std;
+use Archive::Tar::Streamed;
+use Algorithm::Diff;
use File::Slurp;
+use Getopt::Long::Descriptive;
=head1 NAME
=head1 DESCRIPTION
- # /etc/BackupPC/pc/dvd_tar.pl
+ # /etc/BackupPC/pc/_search_archive.pl
=cut
-warn "## ARGV=",dump @ARGV;
-
-my $args;
-my $name;
-foreach ( @ARGV ) {
- my $v = $_;
- if ( m/(\w+)=(.+)/ ) {
- $name = $1;
- $v = $2;
- }
- if ( $name =~ m/List/ ) {
- push @{ $args->{$name} }, $v;
- } else {
- $args->{$name} = $v;
- }
-}
-
-warn "args = ",dump($args);
-
-# FIXME
-my $debug = $ENV{DEBUG} || 1;
-my $check = 0;
-
my $bpc = BackupPC::Lib->new || die "can't create BackupPC::Lib";
+$bpc->ConfigRead('_search_archive'); # read our configuration
my %Conf = $bpc->Conf();
use BackupPC::Search;
%BackupPC::Search::Conf = %Conf;
-my $path = abs_path($0);
-$path =~ s{/[^/]+$}{/}; # FIXME remove?
+my ($opt,$usage) = describe_options(
+"%c %o",
+[ 'host|h=s@', "import just host(s)" ],
+[ 'num|n=s@', "import just backup number(s)" ],
+[ 'check|c', "check archives on disk and sync", { default => 1 } ],
+[ 'debug|d', "debug", { default => 1 } ],
+[ 'help', "show help" ],
+);
+
+print($usage->text), exit if $opt->help;
$|=1;
my $t_fmt = '%Y-%m-%d %H:%M:%S';
-my $dsn = $Conf{SearchDSN} || die "Need SearchDSN in config.pl\n";
-my $user = $Conf{SearchUser} || '';
+#warn "## Conf = ",dump( \%Conf );
my $dbh = DBI->connect($Conf{SearchDSN}, $Conf{SearchUser}, "", { RaiseError => 1, AutoCommit => 0 });
return $out;
}
-=for later
-
my $hsn_cache;
-sub get_backup_id($$$) {
- my ($host, $share, $num) = @_;
+sub get_backup_id($$) {
+ my ($host, $num) = @_;
- my $key = "$host $share $num";
+ my $key = "$host $num";
return $hsn_cache->{$key} if ($hsn_cache->{$key});
my $sth = $dbh->prepare(qq{
FROM backups
INNER JOIN shares ON backups.shareID=shares.ID
INNER JOIN hosts ON backups.hostID = hosts.ID
- WHERE hosts.name = ? and shares.name = ? and backups.num = ?
+ WHERE hosts.name = ? and backups.num = ?
});
- $sth->execute($host, $share, $num);
+ $sth->execute($host, $num);
my ($id) = $sth->fetchrow_array;
- $hsn_cache->{"$host $share $num"} = $id;
+ $hsn_cache->{"$host $num"} = $id;
- print STDERR "# $host $share $num == $id\n" if ($opt{d});
+ print STDERR "# $host $num == $id\n" if $opt->debug;
return $id;
}
$sth_inc_deleted->execute($backup_id);
}
-sub tar_check($$$$) {
- my ($host,$share,$num,$filename) = @_;
+sub system_ok {
+ warn "## system_ok @_\n";
+ system(@_) == 0 || die "system @_:$!";
+}
+
+my $sth_inc_size = $dbh->prepare(qq{
+ update backups set
+ inc_size = ?,
+ parts = ?,
+ inc_deleted = false
+ where id = ?
+});
+
+sub check_archive {
+ my ($host,$num) = @_;
+ warn "# check_archive $host $num";
my $t = time();
- print curr_time, " check $host:$share#$num -> $filename";
+
+ my $glob = "$Conf{ArchiveDest}/$host.$num.*";
+
+ my @tar_parts = sort map { s/^\Q$Conf{ArchiveDest}\E\/*//; $_ } glob $glob ;
+
+ if ( ! @tar_parts ) {
+ warn "ERROR: no files for $glob";
+ return;
+ }
+
+ print curr_time, " check $host $num";
+
+ my $md5_path = "$Conf{ArchiveDest}/$host.$num.md5";
+ unlink $md5_path if -s $md5_path == 0; # fix empty
+
+ if ( ! -e $md5_path ) {
+ system_ok "cd $Conf{ArchiveDest} && /usr/bin/md5sum $host.$num.* > $md5_path";
+ } else {
+ system_ok "cd $Conf{ArchiveDest} && /usr/bin/md5sum -c $md5_path" if $opt->check;
+ }
+
+ my $md5sum;
+ foreach ( split(/\n/, read_file "$Conf{ArchiveDest}/$host.$num.md5" ) ) {
+ my ( $md5, $path ) = split(/\s+/,$_);
+ $md5sum->{$path} = $md5;
+ }
# depending on expected returned value this is used like:
# my $uncompress_size = get_gzip_size('/full/path/to.gz');
sub get_gzip_size($) {
my $filename = shift;
die "file $filename problem: $!" unless (-r $filename);
- open(my $gzip, $bin->{gzip}." -l $filename |") || die "can't gzip -l $filename: $!";
+
+ if ( $filename !~ m/\.gz$/ ) {
+ return -s $filename;
+ }
+
+ open(my $gzip, $Conf{GzipPath}." -l $filename |") || die "can't gzip -l $filename: $!";
+ local $/ = undef;
my $line = <$gzip>;
- chomp($line);
- $line = <$gzip> if ($line =~ /^\s+compressed/);
+ close($gzip);
my ($comp, $uncomp) = (0,0);
- if ($line =~ m/^\s+(\d+)\s+(\d+)\s+\d+\.\d+/) {
+ if ($line =~ m/\s+(\d+)\s+(\d+)\s+\d+\.\d+/s) {
if (wantarray) {
return [ $1, $2 ];
} else {
return $2;
}
} else {
- die "can't find size in line: $line";
+ warn "ERROR can't parse: $line";
+ return -s $filename;
}
}
sub check_part {
- my ($host, $share, $num, $part_nr, $tar_size, $size, $md5, $items) = @_;
- my $backup_id = get_backup_id($host, $share, $num);
+ my ($host, $num, $part_nr, $tar_size, $size, $md5, $items) = @_;
+ my $backup_id = get_backup_id($host, $num);
my $sth_md5 = $dbh->prepare(qq{
select
id, tar_size, size, md5, items
$dbh->commit;
}
- my @tar_parts;
-
- if (-d "$tar_dir/$filename") {
- print ", multi-part";
- opendir(my $dir, "$tar_dir/$filename") || die "can't readdir $tar_dir/$filename: $!";
- @tar_parts = map { my $p = $_; $p =~ s#^#${filename}/#; $p } grep { !/^\./ && !/md5/ && -f "$tar_dir/$filename/$_" } readdir($dir);
- closedir($dir);
- } else {
- push @tar_parts, "${filename}.tar.gz";
- }
-
- print " [parts: ",join(", ", @tar_parts),"]" if ($opt{d});
+ print " [parts: ",join(", ", @tar_parts),"]" if $opt->debug;
- my $same = 1;
my @tar_files;
my $backup_part;
- print " reading" if ($opt{d});
+ print " reading" if $opt->debug;
+
+ my $part_nr = 0;
+ my $inc_size = 0;
- foreach my $tarfilename (@tar_parts) {
+ foreach my $filename (@tar_parts) {
- print "\n\t- $tarfilename";
+ next if $filename eq "$host.$num.md5";
- my $path = "$tar_dir/$tarfilename";
+ print "\n\t- $filename";
+
+ my $path = "$Conf{ArchiveDest}/$filename";
+ $path =~ s{//+}{/}g;
my $size = (stat( $path ))[7] || die "can't stat $path: $!";
- if ($size > $Conf{MaxArchiveSize}) {
- print ", part bigger than media $size > $Conf{MaxArchiveSize}\n";
+ if ($size > $Conf{ArchiveMediaSize}) {
+ print ", part bigger than media $size > $Conf{ArchiveMediaSize}\n";
return 0;
}
print ", $size bytes";
+=for later
open(my $fh, "gzip -cd $path |") or die "can't open $path: $!";
binmode($fh);
$items++;
$tar_size_inarc += $entry->size;
- if ($tar_size_inarc > $Conf{MaxArchiveFileSize}) {
- print ", part $tarfilename is too big $tar_size_inarc > $Conf{MaxArchiveFileSize}\n";
+ if ($tar_size_inarc > $Conf{ArchiveChunkSize}) {
+ print ", part $filename is too big $tar_size_inarc > $Conf{ArchiveChunkSize}\n";
return 0;
}
return 1;
}
- my $tar_size = get_gzip_size( $path );
+=cut
- # real tar size is bigger because of padding
- if ($tar_size_inarc > $tar_size) {
- print ", size of files in tar ($tar_size_inarc) bigger than whole tar ($tar_size)!\n";
- return 0;
- }
+ # FIXME
+ my $tar_size = get_gzip_size( $path );
#
- # check if md5 exists, and if not, create one
+ # finally, check if backup_parts table in database is valid
#
- my $md5_path = $path;
- $md5_path =~ s/\.tar\.gz$/.md5/ || die "can't create md5 filename from $md5_path";
- if (! -e $md5_path || -z $md5_path) {
- print ", creating md5";
- system( $bin->{md5sum} . " $path > $md5_path") == 0 or die "can't create md5 $path: $!";
- } else {
- ## FIXME check if existing md5 is valid
- }
-
- my $md5 = read_file( $md5_path ) || die "can't read md5sum file $md5_path: $!";
- $md5 =~ s#\s.*$##;
-
- # extract part number from filename
- my $part_nr = 1;
- $part_nr = $1 if ($tarfilename =~ m#/(\d+)\.tar\.gz#);
+ my $md5 = $md5sum->{$filename} || die "no md5sum for $filename in ",dump($md5sum);
+ my $items = 1;
+ $part_nr++;
- #
- # finally, check if backup_parts table in database is valid
- #
+ check_part($host, $num, $part_nr, $tar_size, $size, $md5, $items);
- check_part($host, $share, $num, $part_nr, $tar_size, $size, $md5, $items);
+ # round increment size to 2k block size
+ $inc_size += int(($size + 2048) / 2048);
}
- # short-cut and exit;
- return $same unless($same);
+ $sth_inc_size->execute(
+ $inc_size,
+ $part_nr,
+ get_backup_id($host, $num),
+ );
+ $dbh->commit;
@tar_files = sort @tar_files;
print "\n\t",($#tar_files + 1), " tar files";
FROM files
JOIN shares on shares.id = shareid
JOIN hosts on hosts.id = shares.hostid
- WHERE hosts.name = ? and shares.name = ? and backupnum = ?
+ WHERE hosts.name = ? and backupnum = ?
});
- $sth->execute($host, $share, $num);
+ $sth->execute($host, $num);
my @db_files;
while( my $row = $sth->fetchrow_hashref ) {
@db_files = sort @db_files;
+ my $same = 1;
+
if ($#tar_files != $#db_files) {
$same = 0;
print " NUMBER";
return $same;
}
-=cut
#----- main
+foreach ( 0 .. $#{ $opt->host } ) {
+
+ my $host = $opt->host->[$_];
+ my $num = $opt->num->[$_];
+
+ check_archive $host => $num;
+
+}
+
+exit; # FIXME
+
my $sth = $dbh->prepare( qq{
select
while (my $row = $sth->fetchrow_hashref) {
- my $found;
- foreach ( 0 .. $#{ $args->{HostList} } ) {
- if ( $args->{'HostList'}->[$_] eq $row->{host}
- && $args->{'BackupList'}->[$_] eq $row->{num} ) {
- $found = 1;
- }
- }
-
- if ( ! $found ) {
- warn "skipped ",dump($row);
- next;
- }
-
$curr_backup++;
my $tar_file = BackupPC::Search::getGzipName($row->{'host'}, $row->{'share'}, $row->{'num'});
# this will return -1 if file doesn't exist
my $size = BackupPC::Search::get_tgz_size_by_name($tar_file);
- print "# host: ".$row->{host}.", share: ".$row->{'share'}.", backup_num:".$row->{num}." size: $size backup.size: ", $row->{inc_size},"\n" if $debug;
+ print "# host: ".$row->{host}.", share: ".$row->{'share'}.", backup_num:".$row->{num}." size: $size backup.size: ", $row->{inc_size},"\n" if $opt->debug;
if ( $row->{'inc_size'} != -1 && $size != -1 && $row->{'inc_size'} >= $size && $row->{parts} == $row->{backup_parts}) {
- if ($check) {
+ if ($opt->check) {
tar_check($row->{'host'}, $row->{'share'}, $row->{'num'}, $tar_file) && next;
} else {
next;
=for later
# re-create archive?
my $cmd = qq[ $tarIncCreate -h "$row->{host}" -s "$row->{share}" -n $row->{num} -f ];
- print STDERR "## $cmd\n" if ($debug);
+ print STDERR "## $cmd\n" if ($opt->debug);
if (system($cmd) != 0) {
print STDERR " FAILED, marking this backup deleted";
projects / BackupPC.git / blobdiff