--- loncom/loncron 2019/08/01 18:21:36 1.103.2.5 +++ loncom/loncron 2019/03/18 00:37:37 1.118 @@ -2,7 +2,7 @@ # Housekeeping program, started by cron, loncontrol and loncron.pl # -# $Id: loncron,v 1.103.2.5 2019/08/01 18:21:36 raeburn Exp $ +# $Id: loncron,v 1.118 2019/03/18 00:37:37 raeburn Exp $ # # Copyright Michigan State University Board of Trustees # @@ -34,6 +34,7 @@ use lib '/home/httpd/lib/perl/'; use LONCAPA::Configuration; use LONCAPA::Checksumming; use LONCAPA; +use LONCAPA::LWPReq; use Apache::lonnet; use Apache::loncommon; @@ -41,6 +42,9 @@ use IO::File; use IO::Socket; use HTML::Entities; use Getopt::Long; +use GDBM_File; +use Storable qw(thaw); +use File::ReadBackwards; use File::Copy; #globals use vars qw (%perlvar %simplestatus $errors $warnings $notices $totalcount); @@ -115,7 +119,7 @@ sub checkon_daemon { if ($fh) { if (-e "$perlvar{'lonDaemons'}/logs/$daemon.log"){ if (open(DFH,"tail -n25 $perlvar{'lonDaemons'}/logs/$daemon.log|")) { - while (my $line=) { + while (my $line=) { &log($fh,"$line"); if ($line=~/INFO/) { $notices++; } if ($line=~/WARNING/) { $notices++; } @@ -126,7 +130,7 @@ sub checkon_daemon { } &log($fh,"

"); } - + my $pidfile="$perlvar{'lonDaemons'}/logs/$daemon.pid"; my $restartflag=1; @@ -167,6 +171,9 @@ sub checkon_daemon { &log($fh,unlink($pidfile).' - '. `killall -9 $kadaemon 2>&1`. '
'); + if ($kadaemon eq 'loncnew') { + &clean_lonc_childpids(); + } &log($fh,"

$daemon not running, trying to start

"); if (&start_daemon($fh,$daemon,$pidfile,$args)) { @@ -205,8 +212,8 @@ sub checkon_daemon { close (DFH); } &log($fh,"

"); - } - } + } + } } my $fname="$perlvar{'lonDaemons'}/logs/$daemon.log"; @@ -670,35 +677,199 @@ sub test_connections { # ------------------------------------------------------------ Delayed messages sub check_delayed_msg { - my ($fh)=@_; + my ($fh,$weightsref,$exclusionsref)=@_; &log($fh,'

Delayed Messages

'); print "Checking buffers.\n"; &log($fh,'

Scanning Permanent Log

'); my $unsend=0; + my $ignored=0; my %hostname = &Apache::lonnet::all_hostnames(); my $numhosts = scalar(keys(%hostname)); - - my $dfh=IO::File->new("$perlvar{'lonDaemons'}/logs/lonnet.perm.log"); - while (my $line=<$dfh>) { - my ($time,$sdf,$dserv,$dcmd)=split(/:/,$line); - if ($numhosts) { - next unless ($hostname{$dserv}); - } - if ($sdf eq 'F') { - my $local=localtime($time); - &log($fh,"Failed: $time, $dserv, $dcmd
"); - $warnings++; - } - if ($sdf eq 'S') { $unsend--; } - if ($sdf eq 'D') { $unsend++; } + my $checkbackwards = 0; + my $checkfrom = 0; + my $checkexcluded = 0; + my (%bymachine,%weights,%exclusions,%serverhomes); + if (ref($weightsref) eq 'HASH') { + %weights = %{$weightsref}; + } + if (ref($exclusionsref) eq 'HASH') { + %exclusions = %{$exclusionsref}; + if (keys(%exclusions)) { + $checkexcluded = 1; + %serverhomes = &read_serverhomeIDs(); + } } - &log($fh,"

Total unsend messages: $unsend

\n"); - if ($unsend > 0) { - $warnings=$warnings+5*$unsend; +# +# For LON-CAPA 1.2.0 to 2.1.3 (release dates: 8/31/2004 and 3/31/2006) any +# entry logged in lonnet.perm.log for completion of a delayed (critical) +# transaction lacked the hostID for the remote node to which the command +# to be completed was sent. +# +# Because of this, exclusion of items in lonnet.perm.log for nodes which are +# no longer part of the cluster from adding to the overall "unsend" count +# needs additional effort besides the changes made in loncron rev. 1.105. +# +# For "S" (completion) events logging in LON-CAPA 1.2.0 through 2.1.3 included +# "LondTransaction=HASH(hexadecimal)->getClient() :$cmd, where the hexadecimal +# is a memory location, and $cmd is the command sent to the remote node. +# +# Starting with 2.2.0 (released 8/21/2006) logging for "S" (completion) events +# had sethost:$host_id:$cmd after LondTransaction=HASH(hexadecimal)->getClient() +# +# Starting with 2.4.1 (released 6/13/2007) logging for "S" replaced echoing the +# getClient() call with the result of the Transaction->getClient() call itself +# undef for completion of delivery of a delayed message. +# +# The net effect of these changes is that lonnet.perm.log is now accessed three +# times: (a) oldest record is checked, if earlier than release date for 2.5.0 +# then (b) file is read backwards, with timestamp recorded for most recent +# instance of logged "S" event for "update" command without "sethost:$host_id:" +# then (c) file is read forward with records ignored which predate the timestamp +# recorded in (b), if one was found. +# +# In (c), when calculating the unsend total, i.e., the difference between delayed +# transactions ("D") and sent transactions ("S"), transactions are ignored if the +# target node is no longer in the cluster, and also (for "update" commands), if +# the target node is in the list of nodes excluded from the count, in the domain +# configuration for this machine's default domain. The idea here is to remove +# delayed "update" commands for nodes for which inbound access to port 5663, +# is blocked, but are still part of the LON-CAPA network, (i.e., they can still +# replicate content from other nodes). +# + + my $dfh=IO::File->new("$perlvar{'lonDaemons'}/logs/lonnet.perm.log","r"); + if (defined($dfh)) { + while (my $line=<$dfh>) { + my ($time,$sdf,$rest)=split(/:/,$line,3); + if ($time < 1541185772) { + $checkbackwards = 1; + } + last; + } + undef $dfh; + } + + if ($checkbackwards) { + if (tie *BW, 'File::ReadBackwards', "$perlvar{'lonDaemons'}/logs/lonnet.perm.log") { + while(my $line=) { + if ($line =~ /\QLondTransaction=HASH\E[^:]+:update:/) { + ($checkfrom) = split(/:/,$line,2); + last; + } + } + close(BW); + } + } + $dfh=IO::File->new("$perlvar{'lonDaemons'}/logs/lonnet.perm.log","r"); + if (defined($dfh)) { + while (my $line=<$dfh>) { + my ($time,$sdf,$rest)=split(/:/,$line,3); + next unless (($sdf eq 'F') || ($sdf eq 'S') || ($sdf eq 'D')); + next if (($checkfrom) && ($time <= $checkfrom)); + my ($dserv,$dcmd); + if ($sdf eq 'S') { + my ($serva,$cmda,$servb,$cmdb) = split(/:/,$rest); + if ($cmda eq 'sethost') { + chomp($cmdb); + $dcmd = $cmdb; + } else { + $dcmd = $cmda; + } + if (($serva =~ /^LondTransaction/) || ($serva eq '')) { + unless (($servb eq '') || ($servb =~ m{^/})) { + $dserv = $servb; + } + } else { + $dserv = $serva; + } + } else { + ($dserv,$dcmd) = split(/:/,$rest); + } + if ($sdf eq 'F') { + my $local=localtime($time); + &log($fh,"Failed: $time, $dserv, $dcmd
"); + $warnings++; + } + next if ((($dserv eq '') || ($dcmd eq '')) && ($sdf ne 'F')); + if ($sdf eq 'S') { + if ($dcmd eq 'update') { + if ($hostname{$dserv}) { + if ($exclusions{$serverhomes{$hostname{$dserv}}}) { + $ignored --; + } else { + $unsend --; + } + } + if (exists($bymachine{$dserv})) { + $bymachine{$dserv} --; + } else { + $bymachine{$dserv} = -1; + } + } else { + if ($hostname{$dserv}) { + $unsend --; + } + } + } elsif ($sdf eq 'D') { + if ($dcmd eq 'update') { + if ($hostname{$dserv}) { + if ($exclusions{$serverhomes{$hostname{$dserv}}}) { + $ignored ++; + } else { + $unsend ++; + } + } + if (exists($bymachine{$dserv})) { + $bymachine{$dserv} ++; + } else { + $bymachine{$dserv} = 1; + } + } else { + if ($hostname{$dserv}) { + $unsend ++; + } + } + } + } + undef $dfh; + my $nodest = 0; + my $retired = 0; + my %active; + if (keys(%bymachine)) { + unless ($checkexcluded) { + %serverhomes = &read_serverhomeIDs(); + } + foreach my $key (keys(%bymachine)) { + if ($bymachine{$key} > 0) { + if ($hostname{$key}) { + $active{$serverhomes{$hostname{$key}}} += $bymachine{$key}; + } else { + $retired ++; + $nodest += $bymachine{$key}; + } + } + } + } + if (keys(%active)) { + &log($fh,"

Unsend messages by node, active (undegraded) nodes in cluster

\n"); + foreach my $key (sort(keys(%active))) { + &log($fh,&encode_entities("$key => $active{$key}",'<>&"')."\n"); + } + } + &log($fh,"

Total unsend messages: $unsend for ".scalar(keys(%active))." active (undegraded) nodes in cluster.

\n"); + if (keys(%exclusions) > 0) { + &log($fh,"

Total incomplete updates $ignored for ".scalar(keys(%exclusions))." degraded nodes in cluster.

\n"); + } + if ($retired) { + &log($fh,"

Total unsent $nodest for $retired nodes no longer in cluster.

\n"); + } + if ($unsend > 0) { + $warnings=$warnings+$weights{'U'}*$unsend; + } } if ($unsend) { $simplestatus{'unsend'}=$unsend; } @@ -737,9 +908,13 @@ sub check_delayed_msg { } sub finish_logging { - my ($fh)=@_; + my ($fh,$weightsref)=@_; + my %weights; + if (ref($weightsref) eq 'HASH') { + %weights = %{$weightsref}; + } &log($fh,"\n"); - $totalcount=$notices+4*$warnings+100*$errors; + $totalcount=($weights{'N'}*$notices)+($weights{'W'}*$warnings)+($weights{'E'}*$errors); &errout($fh); &log($fh,"

Total Error Count: $totalcount

"); my $now=time; @@ -941,12 +1116,324 @@ sub write_hostips { return; } +sub clean_nosslverify { + my ($fh) = @_; + my %unlinked; + if (-d "$perlvar{'lonSockDir'}/nosslverify") { + if (opendir(my $dh,"$perlvar{'lonSockDir'}/nosslverify")) { + while (my $fname=readdir($dh)) { + next if ($fname =~ /^\.+$/); + if (unlink("/home/httpd/sockets/nosslverify/$fname")) { + &log($fh,"Unlinking $fname
"); + $unlinked{$fname} = 1; + } + } + closedir($dh); + } + } + &log($fh,"

Removed ".scalar(keys(%unlinked))." nosslverify clients

"); + return %unlinked; +} +sub clean_lonc_childpids { + my $childpiddir = "$perlvar{'lonDocRoot'}/lon-status/loncchld"; + if (-d $childpiddir) { + if (opendir(my $dh,$childpiddir)) { + while (my $fname=readdir($dh)) { + next if ($fname =~ /^\.+$/); + unlink("$childpiddir/$fname"); + } + closedir($dh); + } + } +} + +sub write_connection_config { + my ($domconf,%connectssl,%changes); + $domconf = &get_domain_config(); + if (ref($domconf) eq 'HASH') { + if (ref($domconf->{'ssl'}) eq 'HASH') { + foreach my $connect ('connto','connfrom') { + if (ref($domconf->{'ssl'}->{$connect}) eq 'HASH') { + my ($sslreq,$sslnoreq,$currsetting); + my %contypes; + foreach my $type ('dom','intdom','other') { + $connectssl{$connect.'_'.$type} = $domconf->{'ssl'}->{$connect}->{$type}; + } + } + } + } + if (keys(%connectssl)) { + my %currconf; + if (open(my $fh,'<',"$perlvar{'lonTabDir'}/connectionrules.tab")) { + while (my $line = <$fh>) { + chomp($line); + my ($name,$value) = split(/=/,$line); + if ($value =~ /^(?:no|yes|req)$/) { + if ($name =~ /^conn(to|from)_(dom|intdom|other)$/) { + $currconf{$name} = $value; + } + } + } + close($fh); + } + if (open(my $fh,'>',"$perlvar{'lonTabDir'}/connectionrules.tab")) { + my $count = 0; + foreach my $key (sort(keys(%connectssl))) { + print $fh "$key=$connectssl{$key}\n"; + if (exists($currconf{$key})) { + unless ($currconf{$key} eq $connectssl{$key}) { + $changes{$key} = 1; + } + } else { + $changes{$key} = 1; + } + $count ++; + } + close($fh); + print "Completed writing SSL options for lonc/lond for $count items.\n"; + } + } else { + print "Writing of SSL options skipped - no connection rules in domain configuration.\n"; + } + } else { + print "Retrieval of SSL options for lonc/lond skipped - no configuration data available for domain.\n"; + } + return %changes; +} + +sub get_domain_config { + my ($dom,$primlibserv,$isprimary,$url,%confhash); + $dom = $perlvar{'lonDefDomain'}; + $primlibserv = &Apache::lonnet::domain($dom,'primary'); + if ($primlibserv eq $perlvar{'lonHostID'}) { + $isprimary = 1; + } elsif ($primlibserv ne '') { + my $protocol = $Apache::lonnet::protocol{$primlibserv}; + my $hostname = &Apache::lonnet::hostname($primlibserv); + unless ($protocol eq 'https') { + $protocol = 'http'; + } + $url = $protocol.'://'.$hostname.'/cgi-bin/listdomconfig.pl?primary='.$primlibserv.'&format=raw'; + } + if ($isprimary) { + my $lonusersdir = $perlvar{'lonUsersDir'}; + my $fname = $lonusersdir.'/'.$dom.'/configuration.db'; + if (-e $fname) { + my $dbref=&LONCAPA::locking_hash_tie($fname,&GDBM_READER()); + if (ref($dbref) eq 'HASH') { + foreach my $key (sort(keys(%{$dbref}))) { + my $value = $dbref->{$key}; + if ($value =~ s/^__FROZEN__//) { + $value = thaw(&LONCAPA::unescape($value)); + } else { + $value = &LONCAPA::unescape($value); + } + $confhash{$key} = $value; + } + &LONCAPA::locking_hash_untie($dbref); + } + } + } else { + my $request=new HTTP::Request('GET',$url); + my $response=&LONCAPA::LWPReq::makerequest($primlibserv,$request,'',\%perlvar,5); + unless ($response->is_error()) { + my $content = $response->content; + if ($content) { + my @pairs=split(/\&/,$content); + foreach my $item (@pairs) { + my ($key,$value)=split(/=/,$item,2); + my $what = &LONCAPA::unescape($key); + if ($value =~ s/^__FROZEN__//) { + $value = thaw(&LONCAPA::unescape($value)); + } else { + $value = &LONCAPA::unescape($value); + } + $confhash{$what}=$value; + } + } + } + } + return \%confhash; +} + +sub write_hosttypes { + my %intdom = &Apache::lonnet::all_host_intdom(); + my %hostdom = &Apache::lonnet::all_host_domain(); + my $dom = $hostdom{$perlvar{'lonHostID'}}; + my $internetdom = $intdom{$perlvar{'lonHostID'}}; + my %changes; + if (($dom ne '') && ($internetdom ne '')) { + if (keys(%hostdom)) { + my %currhosttypes; + if (open(my $fh,'<',"$perlvar{'lonTabDir'}/hosttypes.tab")) { + while (my $line = <$fh>) { + chomp($line); + my ($name,$value) = split(/:/,$line); + if (($name ne '') && ($value =~ /^(dom|intdom|other)$/)) { + $currhosttypes{$name} = $value; + } + } + close($fh); + } + if (open(my $fh,'>',"$perlvar{'lonTabDir'}/hosttypes.tab")) { + my $count = 0; + foreach my $lonid (sort(keys(%hostdom))) { + my $type = 'other'; + if ($hostdom{$lonid} eq $dom) { + $type = 'dom'; + } elsif ($intdom{$lonid} eq $internetdom) { + $type = 'intdom'; + } + print $fh "$lonid:$type\n"; + if (exists($currhosttypes{$lonid})) { + if ($type ne $currhosttypes{$lonid}) { + $changes{$lonid} = 1; + } + } else { + $changes{$lonid} = 1; + } + $count ++; + } + close($fh); + print "Completed writing host type data for $count hosts.\n"; + } + } else { + print "Writing of host types skipped - no hosts found.\n"; + } + } else { + print "Writing of host types skipped - could not determine this host's LON-CAPA domain or 'internet' domain.\n"; + } + return %changes; +} + +sub update_revocation_list { + my ($result,$changed) = &Apache::lonnet::fetch_crl_pemfile(); + if ($result eq 'ok') { + print "Certificate Revocation List (from CA) updated.\n"; + } else { + print "Certificate Revocation List from (CA) not updated.\n"; + } + return $changed; +} + +sub reset_nosslverify_pids { + my ($fh,%sslrem) = @_; + &checkon_daemon($fh,'lond',40000,'USR2'); + my $loncpidfile="$perlvar{'lonDaemons'}/logs/lonc.pid"; + my $loncppid; + if ((-e $loncpidfile) && (open(my $pfh,'<',$loncpidfile))) { + $loncppid=<$pfh>; + chomp($loncppid); + close($pfh); + if ($loncppid =~ /^\d+$/) { + my %pids_by_host; + my $docdir = $perlvar{'lonDocRoot'}; + if (-d "$docdir/lon-status/loncchld") { + if (opendir(my $dh,"$docdir/lon-status/loncchld")) { + while (my $file = readdir($dh)) { + next if ($file =~ /^\./); + if (open(my $fh,'<',"$docdir/lon-status/loncchld/$file")) { + my $record = <$fh>; + chomp($record); + close($fh); + my ($remotehost,$authmode) = split(/:/,$record); + $pids_by_host{$remotehost}{$authmode}{$file} = 1; + } + } + closedir($dh); + if (keys(%pids_by_host)) { + foreach my $host (keys(%pids_by_host)) { + if ($sslrem{$host}) { + if (ref($pids_by_host{$host}) eq 'HASH') { + if (ref($pids_by_host{$host}{'insecure'}) eq 'HASH') { + if (keys(%{$pids_by_host{$host}{'insecure'}})) { + foreach my $pid (keys(%{$pids_by_host{$host}{'insecure'}})) { + if (open(PIPE,"ps -o ppid= -p $pid |")) { + my $ppid = ; + chomp($ppid); + close(PIPE); + $ppid =~ s/(^\s+|\s+$)//g; + if (($ppid == $loncppid) && (kill 0 => $pid)) { + kill QUIT => $pid; + } + } + } + } + } + } + } + } + } + } + } + } + } + return; +} + +sub get_permcount_settings { + my ($domconf) = @_; + my ($defaults,$names) = &Apache::loncommon::lon_status_items(); + my (%weights,$threshold,$sysmail,$reportstatus,%exclusions); + foreach my $type ('E','W','N','U') { + $weights{$type} = $defaults->{$type}; + } + $threshold = $defaults->{'threshold'}; + $sysmail = $defaults->{'sysmail'}; + $reportstatus = 1; + if (ref($domconf) eq 'HASH') { + if (ref($domconf->{'contacts'}) eq 'HASH') { + if ($domconf->{'contacts'}{'reportstatus'} == 0) { + $reportstatus = 0; + } + if (ref($domconf->{'contacts'}{'lonstatus'}) eq 'HASH') { + if (ref($domconf->{'contacts'}{'lonstatus'}{weights}) eq 'HASH') { + foreach my $type ('E','W','N','U') { + if (exists($domconf->{'contacts'}{'lonstatus'}{weights}{$type})) { + $weights{$type} = $domconf->{'contacts'}{'lonstatus'}{weights}{$type}; + } + } + } + if (ref($domconf->{'contacts'}{'lonstatus'}{'excluded'}) eq 'ARRAY') { + my @excluded = @{$domconf->{'contacts'}{'lonstatus'}{'excluded'}}; + if (@excluded) { + map { $exclusions{$_} = 1; } @excluded; + } + } + if (exists($domconf->{'contacts'}{'lonstatus'}{'threshold'})) { + $threshold = $domconf->{'contacts'}{'lonstatus'}{'threshold'}; + } + if (exists($domconf->{'contacts'}{'lonstatus'}{'sysmail'})) { + $sysmail = $domconf->{'contacts'}{'lonstatus'}{'sysmail'}; + } + } + } + } + return ($threshold,$sysmail,$reportstatus,\%weights,\%exclusions); +} + +sub read_serverhomeIDs { + my %server; + if (-e "$perlvar{'lonTabDir'}/serverhomeIDs.tab") { + if (open(my $fh,'<',"$perlvar{'lonTabDir'}/serverhomeIDs.tab")) { + while (<$fh>) { + my($host,$id) = split(/:/); + chomp($id); + $server{$host} = $id; + } + close($fh); + } + } + return %server; +} + sub send_mail { + my ($sysmail,$reportstatus) = @_; my $defdom = $perlvar{'lonDefDomain'}; my $origmail = $perlvar{'lonAdmEMail'}; my $emailto = &Apache::loncommon::build_recipient_list(undef, 'lonstatusmail',$defdom,$origmail); - if ($totalcount>2500) { + if (($totalcount>$sysmail) && ($reportstatus)) { $emailto.=",$perlvar{'lonSysEMail'}"; } my $from; @@ -1123,6 +1610,10 @@ sub main () { &checkon_daemon($fh,'lonr',40000); } if ($justreload) { + &clean_nosslverify($fh); + &write_connection_config(); + &write_hosttypes(); + &update_revocation_list(); &checkon_daemon($fh,'lond',40000,'USR2'); &checkon_daemon($fh,'lonc',40000,'USR2'); } @@ -1130,14 +1621,26 @@ sub main () { &test_connections($fh); } if (!$justcheckdaemons && !$justcheckconnections && !$justreload && !$justiptables) { - &check_delayed_msg($fh); - &finish_logging($fh); - &log_simplestatus(); + my $domconf = &get_domain_config(); + my ($threshold,$sysmail,$reportstatus,$weightsref,$exclusionsref) = + &get_permcount_settings($domconf); + &check_delayed_msg($fh,$weightsref,$exclusionsref); &write_loncaparevs(); &write_serverhomeIDs(); &write_checksums(); &write_hostips(); - if ($totalcount>200 && !$noemail) { &send_mail(); } + my %sslrem = &clean_nosslverify($fh); + my %conchgs = &write_connection_config(); + my %hosttypechgs = &write_hosttypes(); + my $hadcrlchg = &update_revocation_list(); + if ((keys(%conchgs) > 0) || (keys(%hosttypechgs) > 0) || + $hadcrlchg || (keys(%sslrem) > 0)) { + &checkon_daemon($fh,'lond',40000,'USR2'); + &reset_nosslverify_pids($fh,%sslrem); + } + &finish_logging($fh,$weightsref); + &log_simplestatus(); + if ($totalcount>$threshold && !$noemail) { &send_mail($sysmail,$reportstatus); } } }