#!/usr/local/bin/perl -w use strict; use lib "/usr/local/BackupPC/lib"; 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 Cwd qw/abs_path/; use File::Which; use Archive::Tar::Streamed; use Algorithm::Diff; use Getopt::Std; use File::Slurp; use File::Pid; =head1 NAME BackupPC_incPartsUpdate =head1 DESCRIPTION Create C<.tar.gz> increments on disk calling C. Following options are supported (but all are optional): =over 4 =item -h hostname Update parts for just single C =item -c Force check for tar archives which exist on disk =item -d Turn debugging output =back =cut my %opt; getopts("cdh:", \%opt ); my $debug = $opt{d}; my $check = $opt{c} && print STDERR "NOTICE: tar archive check forced\n"; my $pid_path = abs_path($0); $pid_path =~ s/\W+/_/g; my $pidfile = new File::Pid({ file => "/tmp/$pid_path", }); if (my $pid = $pidfile->running ) { die "$0 already running: $pid\n"; } elsif ($pidfile->pid ne $$) { $pidfile->remove; $pidfile = new File::Pid; } print STDERR "$0 using pid ",$pidfile->pid," file ",$pidfile->file,"\n"; $pidfile->write; my $bpc = BackupPC::Lib->new || die "can't create BackupPC::Lib"; my %Conf = $bpc->Conf(); use BackupPC::SearchLib; %BackupPC::SearchLib::Conf = %Conf; my $path = abs_path($0); $path =~ s#/[^/]+$#/#; my $tarIncCreate = $path .= 'BackupPC_tarIncCreate'; die "can't find $tarIncCreate: $!\n" unless (-x $tarIncCreate); my $bin; foreach my $c (qw/gzip md5sum/) { $bin->{$c} = which($c) || die "$0 needs $c, install it\n"; } $|=1; my $start_t = time(); my $t_fmt = '%Y-%m-%d %H:%M:%S'; my $dsn = $Conf{SearchDSN} || die "Need SearchDSN in config.pl\n"; my $user = $Conf{SearchUser} || ''; my $dbh = DBI->connect($dsn, $user, "", { RaiseError => 1, AutoCommit => 0 }); my $tar_dir = $Conf{InstallDir}.'/'.$Conf{GzipTempDir}; die "problem with $tar_dir, check GzipTempDir in configuration\n" unless (-d $tar_dir && -w $tar_dir); #---- subs ---- sub fmt_time { my $t = shift || return; my $out = ""; my ($ss,$mm,$hh) = gmtime($t); $out .= "${hh}h" if ($hh); $out .= sprintf("%02d:%02d", $mm,$ss); return $out; } sub curr_time { return strftime($t_fmt,localtime()); } my $hsn_cache; sub get_backup_id($$$) { my ($host, $share, $num) = @_; my $key = "$host $share $num"; return $hsn_cache->{$key} if ($hsn_cache->{$key}); my $sth = $dbh->prepare(qq{ SELECT backups.id 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 = ? }); $sth->execute($host, $share, $num); my ($id) = $sth->fetchrow_array; $hsn_cache->{"$host $share $num"} = $id; print STDERR "# $host $share $num == $id\n" if ($opt{d}); return $id; } sub backup_inc_deleted($) { my $backup_id = shift; my $sth_inc_deleted = $dbh->prepare(qq{ update backups set inc_deleted = true where id = ? }); $sth_inc_deleted->execute($backup_id); } sub tar_check($$$$) { my ($host,$share,$num,$filename) = @_; my $t = time(); print curr_time, " check $host:$share#$num -> $filename"; # depending on expected returned value this is used like: # my $uncompress_size = get_gzip_size('/full/path/to.gz'); # my ($compress_size, $uncompress_size) = get_gzip_size('/path.gz'); sub get_gzip_size($) { my $filename = shift; die "file $filename problem: $!" unless (-r $filename); open(my $gzip, $bin->{gzip}." -l $filename |") || die "can't gzip -l $filename: $!"; my $line = <$gzip>; chomp($line); $line = <$gzip> if ($line =~ /^\s+compressed/); my ($comp, $uncomp) = (0,0); if ($line =~ m/^\s+(\d+)\s+(\d+)\s+\d+\.\d+/) { if (wantarray) { return [ $1, $2 ]; } else { return $2; } } else { die "can't find size in line: $line"; } } sub check_part { my ($host, $share, $num, $part_nr, $tar_size, $size, $md5, $items) = @_; my $backup_id = get_backup_id($host, $share, $num); my $sth_md5 = $dbh->prepare(qq{ select id, tar_size, size, md5, items from backup_parts where backup_id = ? and part_nr = ? }); $sth_md5->execute($backup_id, $part_nr); if (my $row = $sth_md5->fetchrow_hashref) { return if ( $row->{tar_size} >= $tar_size && $row->{size} == $size && $row->{md5} eq $md5 && $row->{items} == $items ); print ", deleting invalid backup_parts $row->{id}"; $dbh->do(qq{ delete from backup_parts where id = $row->{id} }); } print ", inserting new"; my $sth_insert = $dbh->prepare(qq{ insert into backup_parts ( backup_id, part_nr, tar_size, size, md5, items ) values (?,?,?,?,?,?) }); $sth_insert->execute($backup_id, $part_nr, $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}); my $same = 1; my @tar_files; my $backup_part; print " reading" if ($opt{d}); foreach my $tarfilename (@tar_parts) { print "\n\t- $tarfilename"; my $path = "$tar_dir/$tarfilename"; my $size = (stat( $path ))[7] || die "can't stat $path: $!"; if ($size > $Conf{MaxArchiveSize}) { print ", part bigger than media $size > $Conf{MaxArchiveSize}\n"; return 0; } print ", $size bytes"; open(my $fh, "gzip -cd $path |") or die "can't open $path: $!"; binmode($fh); my $tar = Archive::Tar::Streamed->new($fh); my $tar_size_inarc = 0; my $items = 0; while(my $entry = $tar->next) { push @tar_files, $entry->name; $items++; $tar_size_inarc += $entry->size; if ($tar_size_inarc > $Conf{MaxArchiveFileSize}) { print ", part $tarfilename is too big $tar_size_inarc > $Conf{MaxArchiveFileSize}\n"; return 0; } } close($fh); print ", $items items"; if ($tar_size_inarc == 0 && $items == 0) { print ", EMPTY tar\n"; my $backup_id = get_backup_id($host, $share, $num); backup_inc_deleted( $backup_id ); $dbh->commit; return 1; } my $tar_size = get_gzip_size( $path ); # real tar size is bigger because of padding if ($tar_size_inarc > $tar_size) { print ", size of files in tar ($tar_size_inarc) bigger than whole tar ($tar_size)!\n"; return 0; } # # check if md5 exists, and if not, create one # 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#); # # finally, check if backup_parts table in database is valid # check_part($host, $share, $num, $part_nr, $tar_size, $size, $md5, $items); } # short-cut and exit; return $same unless($same); @tar_files = sort @tar_files; print "\n\t",($#tar_files + 1), " tar files"; 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, $path; } print " ",($#db_files + 1), " database files, diff"; @db_files = sort @db_files; if ($#tar_files != $#db_files) { $same = 0; print " NUMBER"; } 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'), ", dur: ",fmt_time(time() - $t), "\n"; return $same; } #----- main my $sth = $dbh->prepare( qq{ select backups.id as backup_id, hosts.name as host, shares.name as share, backups.num as num, backups.date, inc_size, parts, count(backup_parts.backup_id) as backup_parts from backups join shares on backups.hostid = shares.hostid and shares.id = backups.shareid join hosts on shares.hostid = hosts.id full outer join backup_parts on backups.id = backup_parts.backup_id where not inc_deleted and backups.size > 0 group by backups.id, hosts.name, shares.name, backups.num, backups.date, inc_size, parts, backup_parts.backup_id order by backups.date } ); $sth->execute(); my $num_backups = $sth->rows; my $curr_backup = 1; if ($opt{h}) { warn "making increments just for host $opt{h}\n"; } while (my $row = $sth->fetchrow_hashref) { if ($opt{h} && $row->{host} ne $opt{h}) { warn "skipped $row->{host}\n" if ($debug); next; } $curr_backup++; 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 "# host: ".$row->{host}.", share: ".$row->{'share'}.", backup_num:".$row->{num}." size: $size backup.size: ", $row->{inc_size},"\n" if ($opt{d}); if ( $row->{'inc_size'} != -1 && $size != -1 && $row->{'inc_size'} >= $size && $row->{parts} == $row->{backup_parts}) { if ($check) { tar_check($row->{'host'}, $row->{'share'}, $row->{'num'}, $tar_file) && next; } else { next; } } print curr_time, " creating $curr_backup/$num_backups ", $row->{host}, ":", $row->{share}, " #", $row->{num}, " ", strftime('%Y-%m-%d', localtime($row->{date})), " -> $tar_file"; my $t = time(); # re-create archive? my $cmd = qq[ $tarIncCreate -h "$row->{host}" -s "$row->{share}" -n $row->{num} -f ]; print STDERR "## $cmd\n" if ($debug); if (system($cmd) != 0) { print STDERR " FAILED, marking this backup deleted"; backup_inc_deleted( $row->{backup_id} ); } print ", dur: ",fmt_time(time() - $t), "\n"; $dbh->commit; } undef $sth; $dbh->disconnect;