--- foxridge-archiver/archiver.pl 2004/07/08 17:27:59 1.2 +++ foxridge-archiver/archiver.pl 2005/09/23 17:16:17 1.6 @@ -16,16 +16,40 @@ $|=1; # # program version -my $version = "0.5.1 (8.7.2004)"; +my $version = "0.7.1 (ROC 23.9.2005)"; + +# short help +my $help = "MPIWG archiver $version +use: archiver [options] docpath +options: + -debug show debugging info + -premigrate don't delete archived files + -force archive even if already archived +"; + +# read command line parameters +my $args = MPIWGStor::parseargs; +if (! scalar(%$args)) { + print $help, "\n"; + exit 1; +} + +# debug level +$debug = (exists $$args{'debug'}) ? $$args{'debug'} : 0; + +# force archiving +my $force_archive = (exists $$args{'force'}) ? $$args{'force'} : 0; # rewrite XML file (necessary for archive date!) my $fix_xml = 1; my $xml_changed = 0; + # XML namespace (not really implemented!) my $namespace = ""; # archive name (archive-path element, usually == $docdir) my $archname; + # archive storage date (now) my $archdate = stime(time); @@ -35,50 +59,52 @@ my $delete_junk_files = 1; # delete data files after archiving my $delete_data_files = 1; +# don't delete archived files with "-premigrate" +if (exists $$args{'premigrate'}) { + $delete_data_files = not $$args{'premigrate'}; +} +if ($delete_data_files) { + logger('INFO', "going to remove successfully archived files from disk"); +} + ####################################################### # external programs # my $archprog = "/opt/tivoli/tsm/client/ba/bin/dsmc"; if (! -x $archprog) { - print "ABORT: TSM client program '$archprog' missing!!\n"; + logger('ABORT', "TSM client program '$archprog' missing!"); exit 1; } my $checkprog = "/usr/local/mpiwg/archive/metacheck"; if (! -x $checkprog) { - print "ABORT: meta data checking program '$checkprog' missing!!\n"; + logger('ABORT', "meta data checking program '$checkprog' missing!"); exit 1; } # log file for archiver my $log_file = "/var/log/mpiwg-archiver.log"; if (! open LOG, ">>$log_file") { - print "ABORT: unable to write log file '$log_file'!!\n"; + logger('ABORT', "unable to write log file '$log_file'!"); exit 1; } ####################################################### # check parameters that were passed to the program # -if ($#ARGV < 0) { - print "ABORT: no document directory given!\n"; - exit 1; -} -my $docdir = $ARGV[0]; + +my $docdir = $$args{'path'}; # strip double slashes $docdir =~ s/\/\//\//; # strip trailing slashes $docdir =~ s/\/+$//; if (! -d $docdir) { - print "ABORT: document directory \'$docdir\' doesn't exist!\n"; + logger('ABORT', "document directory \'$docdir\' doesn't exist!"); exit 1; } -if (($#ARGV > 0)&&($ARGV[1] eq "-premigrate")) { - $delete_data_files = 0; -} my $metafile = "$docdir/index.meta"; if (! -f $metafile) { - print "ABORT: metadata index file \'$metafile\' doesn't exist!\n"; + logger('ABORT', "metadata index file \'$metafile\' doesn't exist!"); exit 1; } @@ -109,7 +135,7 @@ sub read_resource_meta { # get archive-path $archname = MPIWGStor::sstrip($rootnode->findvalue('child::archive-path')); if (! $archname) { - print "ABORT: archive-name element missing!!\n"; + logger('ABORT', "archive-name element missing!"); exit 1; } @@ -120,7 +146,7 @@ sub read_resource_meta { foreach my $fn (@filenodes) { my $name = MPIWGStor::sstrip($fn->findvalue('child::name')); my $path = MPIWGStor::sstrip($fn->findvalue('child::path')); - print "FILE: ($path)$name\n"; + logger('DEBUG', "FILE ($path)$name"); my $f = ($path) ? "$path/$name" : "$name"; $files{$f} = $name; } @@ -130,7 +156,7 @@ sub read_resource_meta { # my $stordatenode = ($rootnode->find('child::archive-storage-date'))->get_node(1); if ($stordatenode) { - print "WARNING: archive storage date exists! Resource already archived?\n"; + logger('WARNING', "archive storage date exists! Resource already archived?"); $warncnt++; # delete old date $stordatenode->removeChildNodes; @@ -173,7 +199,7 @@ sub run_archive { } } } else { - print "ABORT: unable to start archive command '$archcmd'!!\n"; + logger('ABORT', "unable to start archive command '$archcmd'!!"); exit 1; } @@ -196,10 +222,10 @@ sub check_files { my $fp = "$docdir/$ft"; #print " fp: $fp\n"; if ($$archived{$fp}) { - print "DEBUG: $ft archived OK\n"; + logger('DEBUG', "$ft archived OK"); $$archived{$fp} = "OK"; } else { - print "ERROR: file '$ft' missing from archive!\n"; + logger('ERROR', "file '$ft' missing from archive!"); $errcnt++; } } @@ -208,19 +234,19 @@ sub check_files { if ($$archived{$fa} ne "OK") { my ($fn, $fp) = MPIWGStor::split_file_path($fa); if ($MPIWGStor::index_files{$fn}) { - print "DEBUG: $fa ignored\n"; + logger('DEBUG', "$fa ignored"); $na--; } else { - print "WARNING: $fa archived but not in list!\n"; + logger('WARNING', "$fa archived but not in list!"); $warncnt++; } } } if ($nt > $na) { - print "WARNING: less files were archived ($na vs. $nt)!\n"; + logger('WARNING', "less files were archived ($na vs. $nt)!"); } elsif ($na > $nt) { - print "WARNING: more files were archived ($na vs. $nt)!\n"; + logger('WARNING', "more files were archived ($na vs. $nt)!"); } } @@ -248,9 +274,9 @@ sub delete_files { next unless (-f $f); # delete files if (unlink $f) { - print "INFO: remove $f ($fn)\n"; + logger('DEBUG', "remove $f ($fn)"); } else { - print "ERROR: unable to delete $f!\n"; + logger('ERROR', "unable to delete $f!"); $errcnt++; } } @@ -262,7 +288,7 @@ sub delete_files { # dont't remove document dir (shouldn't be empty anyway) next if ($d eq $docdir); if (-d $d) { - print "INFO: remove dir $d\n"; + logger('DEBUG', "remove dir $d"); rmdir $d; } } @@ -279,7 +305,7 @@ sub delete_all_files { my ($files, $dir) = @_; if (! opendir DIR, $dir) { - print "ERROR: unable to read directory $dir!\n"; + logger('ERROR', "unable to read directory $dir!"); $errcnt++; return; } @@ -293,28 +319,28 @@ sub delete_all_files { if (-f "$dir/$f") { # $f is a file if (unlink "$dir/$f") { - print "INFO: removed $f\n"; + logger('DEBUG', "removed $f"); } else { - print "ERROR: unable to delete $f!\n"; + logger('ERROR', "unable to delete $f!"); $errcnt++; } } elsif (-d _) { # $f is a directory (unlink won't work) if ((system 'rm', '-r', "$dir/$f") == 0) { - print "INFO: removed directory $f\n"; + logger('DEBUG', "removed directory $f"); } else { - print "ERROR: unable to delete directory $f!\n"; + logger('ERROR', "unable to delete directory $f!"); $errcnt++; } } else { - print "ERROR: funny object $dir/$f!\n"; + logger('ERROR', "funny object $dir/$f!"); $errcnt++; } } else { # $f is not in the list if (-d "$dir/$f") { # recurse into directories - print "DEBUG: enter $dir/$f\n"; + logger('DEBUG', "enter $dir/$f"); delete_all_files($files, "$dir/$f"); } } @@ -326,22 +352,54 @@ sub delete_all_files { # main # -print "START: archiver $version at $archdate\n"; +logger('START', "archiver $version at $archdate"); # make shure the right user is running this program my $user = getlogin; -#if (($user ne "archive")||($user ne "root")) { -# logger("ABORT", "you ($user) must be archive or root user to run this program!"); -# exit 1; -#} +if (($user ne "archive")&&($user ne "root")) { + logger("WARNING", "you ($user) should be archive or root user to run this program!"); +} + +# check for .archived file +if (-f "$docdir/.archived") { + if (not $force_archive) { + logger('ABORT', "already archived! (.archived file exists)"); + exit 1; + } else { + logger('WARNING', "resource already archived? (.archived file exists)"); + $warncnt++; + } +} # use metacheck first -if (system("$checkprog $docdir >/dev/null") == 0) { - print "INFO: resource '$docdir' check OK\n"; +if (open CHECK, "$checkprog -add-files $docdir |") { + my @errors; + my $msg; + while () { + chomp; + if (/^ERROR/) { + push @errors, $_; + } + $msg = $_; + } + if ($msg =~ /^DONE/) { + logger('DEBUG', "checking index file: $msg"); + logger('INFO', "resource '$docdir' check OK"); + } else { + logger('DEBUG', "errors checking index file:\n " . join("\n ", @errors) . "\n $msg"); + logger('ABORT', "resource '$docdir' check failed!"); + exit 1; + } } else { - print "ABORT: resource '$docdir' check failed!!\n"; + logger('ABORT', "unable to run $checkprog"); exit 1; } +# if (system("$checkprog $docdir >/dev/null") == 0) { +# logger('INFO', "resource '$docdir' check OK"); +# } else { +# logger('ABORT', "resource '$docdir' check failed!!"); +# exit 1; +# } # read index.meta file my ($document, $rootnode) = MPIWGStor::read_xml($metafile); @@ -349,15 +407,15 @@ my ($document, $rootnode) = MPIWGStor::r # check file and add archive date my $files_to_archive = read_resource_meta($rootnode); -print "INFO: ", scalar keys %$files_to_archive, " files to archive\n"; +logger('INFO', (scalar keys %$files_to_archive) . " files to archive"); -# check for .archived file +# remove .archived file if (-f "$docdir/.archived") { if (unlink "$docdir/.archived") { - print "WARNING: existing .archived file has been removed! Resource already archived?\n"; + logger('WARNING', "existing .archived file has been removed!"); $warncnt++; } else { - print "ERROR: unable to remove existing .archived file!\n"; + logger('ERROR', "unable to remove existing .archived file!"); $errcnt++; } } @@ -369,7 +427,7 @@ if ($delete_junk_files) { # write new index.meta if ($errcnt > 0) { - print "ABORT: there were errors!\n"; + logger('ABORT', "there were errors!"); exit 1; } else { if ($fix_xml) { @@ -378,9 +436,10 @@ if ($errcnt > 0) { } # start archiving -my $archived_files = run_archive; +my $archived_files = run_archive(); +my $num_archfiles = scalar keys %$archived_files; -print "INFO: ", scalar keys %$archived_files, " files archived\n"; +logger('INFO', "$num_archfiles files archived"); # check list of archived files check_files($files_to_archive, $archived_files); @@ -398,12 +457,11 @@ if ($errcnt == 0) { } } -print "INFO: $warncnt warnings\n"; -print "INFO: $errcnt errors\n"; -my $num_archfiles = %$archived_files + 1; +logger('INFO', "$warncnt warnings"); +logger('INFO', "$errcnt errors"); if ($errcnt > 0) { - print "ABORT: there were errors! ($num_archfiles files archived) at ", stime(time), "\n"; + logger('ABORT', "there were errors! ($num_archfiles files archived) at " . stime(time)); exit 1; } else { - print "DONE: $num_archfiles files archived at ", stime(time), "\n"; + logger('DONE', "$num_archfiles files archived at " . stime(time)); }