Annotation of loncom/lonc, revision 1.55

1.1       albertel    1: #!/usr/bin/perl
                      2: 
                      3: # The LearningOnline Network
                      4: # lonc - LON TCP-Client Domain-Socket-Server
                      5: # provides persistent TCP connections to the other servers in the network
                      6: # through multiplexed domain sockets
                      7: #
1.55    ! albertel    8: # $Id: lonc,v 1.54 2003/08/29 18:25:01 albertel Exp $
1.22      www         9: #
                     10: # Copyright Michigan State University Board of Trustees
                     11: #
                     12: # This file is part of the LearningOnline Network with CAPA (LON-CAPA).
                     13: #
                     14: # LON-CAPA is free software; you can redistribute it and/or modify
                     15: # it under the terms of the GNU General Public License as published by
                     16: # the Free Software Foundation; either version 2 of the License, or
                     17: # (at your option) any later version.
                     18: #
                     19: # LON-CAPA is distributed in the hope that it will be useful,
                     20: # but WITHOUT ANY WARRANTY; without even the implied warranty of
                     21: # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
                     22: # GNU General Public License for more details.
                     23: #
                     24: # You should have received a copy of the GNU General Public License
                     25: # along with LON-CAPA; if not, write to the Free Software
                     26: # Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
                     27: #
                     28: # /home/httpd/html/adm/gpl.txt
                     29: #
                     30: # http://www.lon-capa.org/
                     31: #
1.1       albertel   32: # PID in subdir logs/lonc.pid
                     33: # kill kills
                     34: # HUP restarts
                     35: # USR1 tries to open connections again
                     36: 
1.2       www        37: # 6/4/99,6/5,6/7,6/8,6/9,6/10,6/11,6/12,7/14,7/19,
1.5       www        38: # 10/8,10/9,10/15,11/18,12/22,
1.10      www        39: # 2/8,7/25 Gerd Kortemeyer
                     40: # 12/05 Gerd Kortemeyer
1.23      harris41   41: # YEAR=2001
1.21      www        42: # 03/14/01,03/15,06/12,11/26,11/27,11/28 Gerd Kortemeyer
1.26      www        43: # YEAR=2002
1.29      www        44: # 2/19/02,02/22/02,02/25/02 Gerd Kortemeyer
1.33      foxr       45: # 3/07/02 Ron Fox 
1.1       albertel   46: # based on nonforker from Perl Cookbook
                     47: # - server who multiplexes without forking
1.40      harris41   48: 
                     49: use lib '/home/httpd/lib/perl/';
                     50: use LONCAPA::Configuration;
1.1       albertel   51: 
                     52: use POSIX;
                     53: use IO::Socket;
                     54: use IO::Select;
                     55: use IO::File;
                     56: use Socket;
                     57: use Fcntl;
                     58: use Tie::RefHash;
                     59: use Crypt::IDEA;
1.32      foxr       60: #use Net::Ping;
1.26      www        61: use LWP::UserAgent();
1.1       albertel   62: 
1.30      www        63: $status='';
                     64: $lastlog='';
                     65: $conserver='SHELL';
1.32      foxr       66: $DEBUG = 0;			# Set to 1 for annoyingly complete logs.
1.49      albertel   67: $VERSION='$Revison$'; #' stupid emacs
                     68: $remoteVERSION;
1.8       harris41   69: # -------------------------------- Set signal handlers to record abnormal exits
                     70: 
1.29      www        71: &status("Init exception handlers");
1.26      www        72: $SIG{QUIT}=\&catchexception;
1.8       harris41   73: $SIG{__DIE__}=\&catchexception;
                     74: 
1.41      matthew    75: # ---------------------------------- Read loncapa_apache.conf and loncapa.conf
1.42      harris41   76: &status("Read loncapa.conf and loncapa_apache.conf");
                     77: my $perlvarref=LONCAPA::Configuration::read_conf('loncapa.conf');
1.40      harris41   78: my %perlvar=%{$perlvarref};
                     79: undef $perlvarref;
1.7       www        80: 
1.13      harris41   81: # ----------------------------- Make sure this process is running from user=www
1.29      www        82: &status("Check user ID");
1.13      harris41   83: my $wwwid=getpwnam('www');
                     84: if ($wwwid!=$<) {
                     85:    $emailto="$perlvar{'lonAdmEMail'},$perlvar{'lonSysEMail'}";
                     86:    $subj="LON: $perlvar{'lonHostID'} User ID mismatch";
1.14      www        87:    system("echo 'User ID mismatch.  lonc must be run as user www.' |\
1.13      harris41   88:  mailto $emailto -s '$subj' > /dev/null");
                     89:    exit 1;
                     90: }
                     91: 
1.7       www        92: # --------------------------------------------- Check if other instance running
                     93: 
                     94: my $pidfile="$perlvar{'lonDaemons'}/logs/lonc.pid";
                     95: 
                     96: if (-e $pidfile) {
                     97:    my $lfh=IO::File->new("$pidfile");
                     98:    my $pide=<$lfh>;
                     99:    chomp($pide);
1.11      harris41  100:    if (kill 0 => $pide) { die "already running"; }
1.7       www       101: }
1.1       albertel  102: 
                    103: # ------------------------------------------------------------- Read hosts file
                    104: 
1.11      harris41  105: open (CONFIG,"$perlvar{'lonTabDir'}/hosts.tab") || die "Can't read host file";
1.1       albertel  106: 
                    107: while ($configline=<CONFIG>) {
                    108:     my ($id,$domain,$role,$name,$ip)=split(/:/,$configline);
                    109:     chomp($ip);
1.28      www       110:     if ($ip) {
                    111:      $hostip{$id}=$ip;
                    112:      $hostname{$id}=$name;
                    113:     }
1.1       albertel  114: }
1.27      www       115: 
1.1       albertel  116: close(CONFIG);
                    117: 
                    118: # -------------------------------------------------------- Routines for forking
                    119: 
                    120: %children               = ();       # keys are current child process IDs,
                    121:                                     # values are hosts
                    122: %childpid               = ();       # the other way around
                    123: 
                    124: %childatt               = ();       # number of attempts to start server
                    125:                                     # for ID
                    126: 
1.54      albertel  127: $childmaxattempts=15;
1.3       www       128: 
1.1       albertel  129: # ---------------------------------------------------- Fork once and dissociate
1.29      www       130: &status("Fork and dissociate");
1.1       albertel  131: $fpid=fork;
                    132: exit if $fpid;
1.11      harris41  133: die "Couldn't fork: $!" unless defined ($fpid);
1.1       albertel  134: 
1.11      harris41  135: POSIX::setsid() or die "Can't start new session: $!";
1.1       albertel  136: 
1.30      www       137: $conserver='PARENT';
                    138: 
1.1       albertel  139: # ------------------------------------------------------- Write our PID on disk
1.29      www       140: &status("Write PID");
1.1       albertel  141: $execdir=$perlvar{'lonDaemons'};
                    142: open (PIDSAVE,">$execdir/logs/lonc.pid");
                    143: print PIDSAVE "$$\n";
                    144: close(PIDSAVE);
1.5       www       145: &logthis("<font color=red>CRITICAL: ---------- Starting ----------</font>");
1.1       albertel  146: 
                    147: # ----------------------------- Ignore signals generated during initial startup
                    148: $SIG{HUP}=$SIG{USR1}='IGNORE';
                    149: # ------------------------------------------------------- Now we are on our own
                    150:     
                    151: # Fork off our children, one for every server
                    152: 
1.18      www       153: &status("Forking ...");
                    154: 
1.1       albertel  155: foreach $thisserver (keys %hostip) {
1.32      foxr      156:     #if (&online($hostname{$thisserver})) {
1.26      www       157:        make_new_child($thisserver);
1.32      foxr      158:     #}
1.1       albertel  159: }
                    160: 
                    161: &logthis("Done starting initial servers");
                    162: # ----------------------------------------------------- Install signal handlers
                    163: 
1.32      foxr      164: 
1.1       albertel  165: $SIG{INT}  = $SIG{TERM} = \&HUNTSMAN;
                    166: $SIG{HUP}  = \&HUPSMAN;
                    167: $SIG{USR1} = \&USRMAN;
                    168: 
                    169: # And maintain the population.
                    170: while (1) {
1.32      foxr      171:     my $deadpid = wait;		# Wait for the next child to die.
1.39      foxr      172:                                 # See who died and start new one
                    173:                                 # or a signal (e.g. USR1 for restart).
                    174:                                 # if a signal, the wait will fail
                    175:                                 # This is ordinarily detected by
                    176:                                 # checking for the existence of the
                    177:                                 # pid index inthe children hash since
                    178:                                 # the return value from a failed wait is -1
                    179:                                 # which is an impossible PID.
1.18      www       180:     &status("Woke up");
1.30      www       181:     my $skipping='';
1.32      foxr      182: 
                    183:     if(exists($children{$deadpid})) {
                    184: 
                    185: 	$thisserver = $children{$deadpid}; # Look name of dead guy's peer.
                    186: 
                    187: 	delete($children{$deadpid}); # Get rid of dead hash entry.
                    188: 
                    189: 	if($childatt{$thisserver} < $childmaxattempts) {
                    190: 	    $childatt{$thisserver}++;
                    191: 	    &logthis(
                    192: 	       "<font color=yellow>INFO: Trying to reconnect for $thisserver "
                    193:             ."($childatt{$thisserver} of $childmaxattempts attempts)</font>"); 
                    194: 	    make_new_child($thisserver);
                    195: 	
                    196: 	}
                    197: 	else {
                    198: 	    $skipping .= $thisserver.' ';
                    199: 	}
                    200: 	if($skipping) {
                    201: 	    &logthis("<font color=blue>WARNING: Skipped $skipping</font>");
                    202:   
                    203: 	}
1.30      www       204:     }
1.32      foxr      205: 
1.1       albertel  206: }
                    207: 
                    208: 
1.32      foxr      209: 
1.1       albertel  210: sub make_new_child {
                    211:    
1.30      www       212:     $newserver=shift;
1.1       albertel  213:     my $pid;
                    214:     my $sigset;
1.30      www       215:     &logthis("Attempting to start child for server $newserver");
1.1       albertel  216:     # block signal for fork
                    217:     $sigset = POSIX::SigSet->new(SIGINT);
                    218:     sigprocmask(SIG_BLOCK, $sigset)
1.11      harris41  219:         or die "Can't block SIGINT for fork: $!\n";
1.1       albertel  220:     
1.11      harris41  221:     die "fork: $!" unless defined ($pid = fork);
1.1       albertel  222:     
                    223:     if ($pid) {
                    224:         # Parent records the child's birth and returns.
                    225:         sigprocmask(SIG_UNBLOCK, $sigset)
1.11      harris41  226:             or die "Can't unblock SIGINT for fork: $!\n";
1.30      www       227:         $children{$pid} = $newserver;
1.32      foxr      228:         $childpid{$newserver} = $pid;
1.1       albertel  229:         return;
                    230:     } else {
1.30      www       231:         $conserver=$newserver;
1.1       albertel  232:         # Child can *not* return from this subroutine.
                    233:         $SIG{INT} = 'DEFAULT';      # make SIGINT kill us as it did before
1.18      www       234:         $SIG{USR1}= \&logstatus;
                    235:    
1.1       albertel  236:         # unblock signals
                    237:         sigprocmask(SIG_UNBLOCK, $sigset)
1.11      harris41  238:             or die "Can't unblock SIGINT for fork: $!\n";
1.1       albertel  239: 
                    240: # ----------------------------- This is the modified main program of non-forker
                    241: 
                    242: $port = "$perlvar{'lonSockDir'}/$conserver";
                    243: 
                    244: unlink($port);
1.18      www       245: 
1.29      www       246: # -------------------------------------------------------------- Open other end
1.1       albertel  247: 
1.29      www       248: &openremote($conserver);
1.32      foxr      249: 	&logthis("<font color=green> Connection to $conserver open </font>");
1.3       www       250: # ----------------------------------------- We're online, send delayed messages
1.18      www       251:     &status("Checking for delayed messages");
1.32      foxr      252: 
1.4       www       253:     my @allbuffered;
1.3       www       254:     my $path="$perlvar{'lonSockDir'}/delayed";
1.4       www       255:     opendir(DIRHANDLE,$path);
                    256:     @allbuffered=grep /\.$conserver$/, readdir DIRHANDLE;
                    257:     closedir(DIRHANDLE);
1.3       www       258:     my $dfname;
1.44      www       259:     foreach (sort @allbuffered) {
1.30      www       260:         &status("Sending delayed: $_");
1.4       www       261:         $dfname="$path/$_";
1.32      foxr      262:         if($DEBUG) { &logthis('Sending '.$dfname); }
1.3       www       263:         my $wcmd;
                    264:         {
                    265:          my $dfh=IO::File->new($dfname);
1.4       www       266:          $cmd=<$dfh>;
1.3       www       267:         }
                    268:         chomp($cmd);
                    269:         my $bcmd=$cmd;
                    270:         if ($cmd =~ /^encrypt\:/) {
                    271: 	    my $rcmd=$cmd;
                    272:             $rcmd =~ s/^encrypt\://;
                    273:             chomp($rcmd);
                    274:             my $cmdlength=length($rcmd);
                    275:             $rcmd.="         ";
                    276:             my $encrequest='';
                    277:             for (my $encidx=0;$encidx<=$cmdlength;$encidx+=8) {
                    278:                 $encrequest.=
                    279:                     unpack("H16",$cipher->encrypt(substr($rcmd,$encidx,8)));
                    280:             }
                    281:             $cmd="enc:$cmdlength:$encrequest\n";
                    282:         }
1.33      foxr      283: 	$answer = londtransaction($remotesock, $cmd, 60);
1.3       www       284: 	chomp($answer);
1.20      www       285: 
                    286:         if (($answer ne '') && ($@!~/timeout/)) {
1.3       www       287: 	    unlink("$dfname");
1.30      www       288:             &logthis("Delayed $cmd: >$answer<");
1.3       www       289:             &logperm("S:$conserver:$bcmd");
                    290:         }        
1.23      harris41  291:     }
1.32      foxr      292: 	if($DEBUG) { &logthis("<font color=green> Delayed transactions sent"); }
1.1       albertel  293: 
                    294: # ------------------------------------------------------- Listen to UNIX socket
1.30      www       295: &status("Opening socket");
1.1       albertel  296: unless (
                    297:   $server = IO::Socket::UNIX->new(Local  => $port,
                    298:                                   Type   => SOCK_STREAM,
                    299:                                   Listen => 10 )
1.5       www       300:    ) { 
                    301:        my $st=120+int(rand(240));
                    302:        &logthis(
                    303:          "<font color=blue>WARNING: ".
1.33      foxr      304:          "Can't make server socket ($st secs):  .. exiting</font>");
1.5       www       305:        sleep($st);
1.1       albertel  306:        exit; 
                    307:      };
1.32      foxr      308:    
1.1       albertel  309: # -----------------------------------------------------------------------------
                    310: 
1.5       www       311: &logthis("<font color=green>$conserver online</font>");
                    312: 
                    313: # -----------------------------------------------------------------------------
1.1       albertel  314: # begin with empty buffers
                    315: %inbuffer  = ();
                    316: %outbuffer = ();
                    317: %ready     = ();
1.35      foxr      318: %servers   = ();	# To be compatible with make filevector.  indexed by
1.37      foxr      319: 			# File ids, values are sockets.
1.35      foxr      320: 			# note that the accept socket is omitted.
1.1       albertel  321: 
                    322: tie %ready, 'Tie::RefHash';
                    323: 
1.37      foxr      324: # nonblock($server);
                    325: # $select = IO::Select->new($server);
1.1       albertel  326: 
                    327: # Main loop: check reads/accepts, check writes, check ready to process
1.37      foxr      328: 
1.46      albertel  329: status("Main loop $conserver");
1.1       albertel  330: while (1) {
                    331:     my $client;
                    332:     my $rv;
                    333:     my $data;
                    334: 
1.35      foxr      335:     my $infdset;		# bit vec of fd's to select on input.
                    336: 
                    337:     my $outfdset;		# Bit vec of fd's to select on output.
                    338: 
                    339: 
                    340:     $infdset = MakeFileVector(\%servers);
                    341:     $outfdset= MakeFileVector(\%outbuffer);
1.37      foxr      342:     vec($infdset, $server->fileno, 1) = 1;
                    343:     if($DEBUG) {
                    344: 	&logthis("Adding ".$server->fileno.
                    345: 		 " to input select vector (listner)".
                    346: 		 unpack("b*",$infdset)."\n");
1.1       albertel  347:     }
1.37      foxr      348:     DoSelect(\$infdset, \$outfdset); # Wait for input.
                    349:     if($DEBUG) {
                    350: 	&logthis("Doselect completed!");
                    351: 	&logthis("ins = ".unpack("b*",$infdset)."\n");
                    352: 	&logthis("outs= ".unpack("b*",$outfdset)."\n");
                    353: 		 
1.1       albertel  354:     }
1.15      www       355: 
1.37      foxr      356:     # Checkfor new connections:
                    357:     if (vec($infdset, $server->fileno, 1)) {
                    358: 	if($DEBUG) {
                    359: 	    &logthis("New connection established");
                    360: 	}
                    361: 	# accept a new connection
                    362: 	&status("Accept new connection: $conserver");
                    363: 	$client = $server->accept();
1.55    ! albertel  364: 	if (!$client) {
        !           365: 	    &logthis("Got stupid nonexisent client on ".$server->fileno." $conserver \n");
        !           366: 	} else {
        !           367: 	    if($DEBUG) {
        !           368: 		&logthis("New client fd = ".$client->fileno."\n");
        !           369: 	    }
        !           370: 	    $servers{$client->fileno} = $client;
        !           371: 	    nonblock($client);
        !           372: 	    $client->sockopt(SO_KEEPALIVE, 1); # Enable monitoring of
        !           373: 	                                       # connection liveness.
1.37      foxr      374: 	}
                    375:     }
                    376:     HandleInput($infdset, \%servers, \%inbuffer, \%outbuffer, \%ready);
                    377:     HandleOutput($outfdset, \%servers, \%outbuffer, \%inbuffer,
                    378: 		 \%ready);
                    379: # -------------------------------------------------------- Wow, connection lost
1.15      www       380: 
1.37      foxr      381: }
                    382:    
1.1       albertel  383:     }
                    384: }
1.25      albertel  385: 
1.1       albertel  386: # ------------------------------------------------------- End of make_new_child
                    387: 
1.35      foxr      388: 
                    389: #
                    390: #  Make a vector of file descriptors to wait for in a select.
                    391: #  parameters:
                    392: #     \%fdhash  -reference to a hash which has IO::Socket's as indices.  
                    393: #                We only care about the indices, not the values.
                    394: #  A select vector is created from all indices of the hash.
                    395: 
                    396: sub MakeFileVector
                    397: {
                    398:     my $fdhash = shift;
                    399:     my $selvar = "";
                    400: 
1.37      foxr      401:     foreach $socket (keys %$fdhash) {
                    402: 	if($DEBUG) {
                    403: 	    &logthis("Adding  ".$socket.
                    404: 		     "to select vector. (client)\n");
                    405: 	}
                    406: 	vec($selvar, $socket, 1) = 1;
1.35      foxr      407:     }
                    408:     return $selvar;
                    409: }
                    410: 
                    411: 
                    412: #
                    413: #  HandleOutput:
                    414: #    Processes output on a buffered set of file descriptors which are
                    415: #    ready to be read.
                    416: #  Parameters:
1.37      foxr      417: #    $selvector - Vector of file descriptors which are writable.
1.35      foxr      418: #    \%sockets  - Vector of socket references indexed by socket.
                    419: #    \%buffers  - Reference to a hash containing output buffers.
                    420: #                 Hashes are indexed by sockets.  The file descriptors of some
                    421: #                 of those sockets will be present in $selvector.
                    422: #                 For each one of those, we will attempt to write the output
                    423: #                 buffer to the socket.  Note that we will assume that
                    424: #                 the sockets are being run in non blocking mode.
                    425: #   \%inbufs    - Reference to hash containing input buffers.
                    426: #   \%readys    - Reference to hash containing flags for items with complete
                    427: #                 requests.
                    428: #
                    429: sub HandleOutput
                    430: {
                    431:     my $selvector = shift;
                    432:     my $sockets   = shift;
                    433:     my $buffers   = shift;
                    434:     my $inbufs    = shift;
                    435:     my $readys    = shift;
1.37      foxr      436:     my $sock;
1.35      foxr      437: 
1.37      foxr      438:     if($DEBUG) {
                    439: 	&logthis("HandleOutput entered\n");
                    440:     }
                    441: 
                    442:     foreach $sock (keys %$sockets) {
1.35      foxr      443: 	my $socket = $sockets->{$sock};
1.37      foxr      444: 	if(vec($selvector, $sock, 1)) { # $socket is writable.
                    445: 	    if($DEBUG) {
                    446: 		&logthis("Sending $buffers->{$sock} \n");
                    447: 	    }
                    448: 	    my $rv = $socket->send($buffers->{$sock}, 0);
1.35      foxr      449: 	    $errno = $!;
                    450: 	    unless ($buffers->{$sock} eq "con_lost\n") {
                    451: 		unless (defined $rv) { # Write failed... could be EINTR
                    452: 		    unless ($errno == POSIX::EINTR) {
                    453: 			&logthis("Write failed on writable socket");
                    454: 		    }		# EINTR is not an error .. just retry.
                    455: 		    next;
                    456: 		}
                    457: 		if( ($rv == length $buffers->{$sock})    ||
                    458: 		    ($errno == POSIX::EWOULDBLOCK)       ||
                    459: 		    ($errno == POSIX::EAGAIN)            || # same as above.
                    460: 		    ($errno == POSIX::EINTR)             || # signal during IO
                    461: 		    ($errno == 0)) {
                    462: 		    substr($buffers->{$sock}, 0, $rv)=""; # delete written part
                    463: 		    delete $buffers->{$sock} unless length $buffers->{$sock};
                    464: 		} else {
                    465: 		    # For some reason the write failed with an error code
                    466: 		    # we didn't look for.  Shutdown the socket.
                    467: 		    &logthis("Unable to write data with ".$errno.": ".
                    468: 			     "Dropping data: ".length($buffers->{$sock}).
                    469: 			     ", $rv");
                    470: 		    #
                    471: 		    # kill off the buffers in the hash:
                    472: 
                    473: 		    delete $buffers->{$sock};
                    474: 		    delete $inbufs->{$sock};
                    475: 		    delete $readys->{$sock};
                    476: 
1.37      foxr      477: 		    close($socket); # Close the client socket.
1.35      foxr      478: 		    next;
                    479: 		}
                    480: 	    } else {		# Kludgy way to mark lond connection lost.
                    481: 		&logthis(
                    482: 		 "<font color=red>CRITICAL lond connection lost</font>");
                    483: 		status("Connection lost");
                    484: 		$remotesock->shutdown(2);
                    485: 		&logthis("Attempting to open a new connection");
1.37      foxr      486: 		&openremote($conserver);
1.35      foxr      487: 	    }
                    488: 		   
                    489: 	}
                    490:     }
                    491: 
                    492: }
                    493: #
                    494: #   HandleInput - Deals with input on client sockets.
                    495: #                 Each socket has an associated input buffer.
                    496: #                 For each readable socket, the currently available
                    497: #                 data is appended to this buffer.
                    498: #                 If necessary, the buffer is created.
                    499: #                 On various failures, we may shutdown the client.
                    500: #  Parameters:
                    501: #     $selvec   - Vector of readable sockets.
                    502: #     \%sockets - Refers to the  Hash of sockets indexed by sockets.  
                    503: #                 Each of these may or may not have it's fd bit set 
                    504: #                 in the $selvec.
                    505: #     \%ibufs   - Refers to the hash of input buffers indexed by socket.
                    506: #     \%obufs   - Hash of output buffers indexed by socket. 
                    507: #     \%ready   - Hash of ready flags indicating the existence of a completed
                    508: #                 Request.
                    509: sub HandleInput 
                    510: {
                    511: 
                    512:     # Marshall the parameters.   Note that the hashes are actually
                    513:     # references not values.
                    514: 
                    515:     my $selvec  = shift;
                    516:     my $sockets = shift;
                    517:     my $ibufs   = shift;
                    518:     my $obufs   = shift;
                    519:     my $ready   = shift;
1.37      foxr      520:     my $sock;
1.35      foxr      521: 
1.38      foxr      522:     if($DEBUG) {
                    523: 	&logthis("Entered HandleInput\n");
                    524:     }
1.37      foxr      525:     foreach $sock (keys %$sockets) {
1.35      foxr      526: 	my $socket = $sockets->{$sock};
1.37      foxr      527: 	if(vec($selvec, $sock, 1)) { # Socket which is readable.
1.35      foxr      528: 
                    529: 	    #  Attempt to read the data and do error management.
                    530: 	    my $data = '';
1.37      foxr      531: 	    my $rv = $socket->recv($data, POSIX::BUFSIZ, 0);
                    532: 	    if($DEBUG) {
                    533: 		&logthis("Received $data from socket");
                    534: 	    }
1.35      foxr      535: 	    unless (defined($rv) && length $data) {
                    536: 
                    537: 		# Read an end of file.. this is a disconnect from the peer.
                    538: 
                    539: 		delete $sockets->{$sock};
                    540: 		delete $ibufs->{$sock};
                    541: 		delete $obufs->{$sock};
                    542: 		delete $ready->{$sock};
                    543: 
                    544: 		status("Idle");
1.37      foxr      545: 		close $socket;
1.35      foxr      546: 		next;
                    547: 	    }
                    548: 	    #  Append the read data to the input buffer. If the buffer
                    549: 	    # now contains a \n the request is complete and we can 
                    550: 	    # mark this in the $ready hash (one request for each \n.)
                    551: 
                    552: 	    $ibufs->{$sock} .= $data;
                    553: 	    while($ibufs->{$sock} =~ s/(.*\n)//) {
                    554: 		push(@{$ready->{$sock}}, $1);
                    555: 	    }
                    556: 	    
                    557: 	}
                    558:     }
                    559:     #  Now handle any requests which are ready:
                    560: 
                    561:     foreach $client (keys %ready) {
                    562: 	handle($client);
1.36      foxr      563:     }
                    564: }
                    565: 
                    566: # DoSelect:  does a select with no timeout.  On signal (errno == EINTR), 
                    567: #            the select is retried until there are items in the returned
                    568: #            vectors.  
                    569: #
                    570: # Parameters:
                    571: #   \$readvec   - Reference to a vector of file descriptors to 
                    572: #                 check for readability.
                    573: #   \$writevec  - Reference to a vector of file descriptors to check for
                    574: #                 writability.
                    575: #  On exit, the referents are modified with vectors indicating which 
                    576: #  file handles are readable/writable.
                    577: #
                    578: sub DoSelect {
                    579:     my $readvec = shift;
                    580:     my $writevec= shift;
                    581:     my $outs;
                    582:     my $ins;
                    583: 
                    584:     while (1) {
1.37      foxr      585: 	my $nfds = select( $ins = $$readvec, $outs = $$writevec, undef, undef);
                    586: 	if($nfds) {
                    587: 	    if($DEBUG) {
                    588: 		&logthis("select exited with ".$nfds." fds\n");
                    589: 		&logthis("ins = ".unpack("b*",$ins).
                    590: 			 " readvec = ".unpack("b*",$$readvec)."\n");
                    591: 		&logthis("outs = ".unpack("b*",$outs).
                    592: 			 " writevec = ".unpack("b*",$$writevec)."\n");
                    593: 	    }
1.36      foxr      594: 	    $$readvec  = $ins;
                    595: 	    $$writevec = $outs;
                    596: 	    return;
                    597: 	} else {
1.37      foxr      598: 	    if($DEBUG) {
                    599: 		&logthis("Select exited with no bits set in mask\n");
                    600: 	    }
1.36      foxr      601: 	    die "Select failed" unless $! == EINTR;
                    602: 	}
1.35      foxr      603:     }
                    604: }
                    605: 
1.1       albertel  606: # handle($socket) deals with all pending requests for $client
1.35      foxr      607: #
1.1       albertel  608: sub handle {
                    609:     # requests are in $ready{$client}
                    610:     # send output to $outbuffer{$client}
                    611:     my $client = shift;
                    612:     my $request;
                    613:     foreach $request (@{$ready{$client}}) {
                    614: # ============================================================= Process request
                    615:         # $request is the text of the request
                    616:         # put text of reply into $outbuffer{$client}
1.29      www       617: # ------------------------------------------------------------ Is this the end?
1.33      foxr      618: 	chomp($request);
1.32      foxr      619: 	if($DEBUG) {
                    620:      &logthis("<font color=green> Request $request processing starts</font>");
                    621:         }
1.29      www       622:         if ($request eq "close_connection_exit\n") {
1.30      www       623: 	    &status("Request close connection");
1.29      www       624:            &logthis(
1.32      foxr      625:      "<font color=red>CRITICAL: Request Close Connection ... exiting</font>");
1.29      www       626:            $remotesock->shutdown(2);
                    627:            $server->close();
                    628:            exit;
                    629:         }
1.1       albertel  630: # -----------------------------------------------------------------------------
                    631:         if ($request =~ /^encrypt\:/) {
                    632: 	    my $cmd=$request;
                    633:             $cmd =~ s/^encrypt\://;
                    634:             chomp($cmd);
                    635:             my $cmdlength=length($cmd);
                    636:             $cmd.="         ";
                    637:             my $encrequest='';
                    638:             for (my $encidx=0;$encidx<=$cmdlength;$encidx+=8) {
                    639:                 $encrequest.=
                    640:                     unpack("H16",$cipher->encrypt(substr($cmd,$encidx,8)));
                    641:             }
1.33      foxr      642:             $request="enc:$cmdlength:$encrequest";
1.1       albertel  643:         }
1.19      www       644: # --------------------------------------------------------------- Main exchange
1.54      albertel  645: 	$answer = londtransaction($remotesock, $request, 60);
1.33      foxr      646: 
                    647: 	if($DEBUG) { 
                    648: 	    &logthis("<font color=green> Request data exchange complete");
                    649: 	}
                    650: 	if ($@=~/timeout/) { 
                    651: 	    $answer='';
                    652: 	    &logthis(
                    653: 		     "<font color=red>CRITICAL: Timeout: $request</font>");
                    654: 	}  
1.19      www       655: 
                    656: 
1.1       albertel  657:         if ($answer) {
                    658: 	   if ($answer =~ /^enc/) {
                    659:                my ($cmd,$cmdlength,$encinput)=split(/:/,$answer);
                    660:                chomp($encinput);
                    661: 	       $answer='';
                    662:                for (my $encidx=0;$encidx<length($encinput);$encidx+=16) {
                    663:                   $answer.=$cipher->decrypt(
                    664:                    pack("H16",substr($encinput,$encidx,16))
                    665:                   );
                    666: 	       }
                    667: 	      $answer=substr($answer,0,$cmdlength);
                    668: 	      $answer.="\n";
                    669: 	   }
1.33      foxr      670: 	   if($DEBUG) {
                    671: 	       &logthis("sending $answer to client\n");
                    672: 	   }
1.1       albertel  673:            $outbuffer{$client} .= $answer;
                    674:         } else {
                    675:            $outbuffer{$client} .= "con_lost\n";
                    676:         }
                    677: 
1.30      www       678:      &status("Completed: $request");
1.32      foxr      679: 	if($DEBUG) {
                    680: 	    &logthis("<font color=green> Request processing complete</font>");
                    681: 	}
1.1       albertel  682: # ===================================================== Done processing request
                    683:     }
                    684:     delete $ready{$client};
                    685: # -------------------------------------------------------------- End non-forker
1.32      foxr      686:     if($DEBUG) {
                    687: 	&logthis("<font color=green> requests for child handled</font>");
                    688:     }
1.1       albertel  689: }
                    690: # ---------------------------------------------------------- End make_new_child
                    691: 
                    692: # nonblock($socket) puts socket into nonblocking mode
                    693: sub nonblock {
                    694:     my $socket = shift;
                    695:     my $flags;
                    696: 
                    697:     
                    698:     $flags = fcntl($socket, F_GETFL, 0)
1.11      harris41  699:             or die "Can't get flags for socket: $!\n";
1.1       albertel  700:     fcntl($socket, F_SETFL, $flags | O_NONBLOCK)
1.11      harris41  701:             or die "Can't make socket nonblocking: $!\n";
1.29      www       702: }
                    703: 
                    704: 
                    705: sub openremote {
                    706: # ---------------------------------------------------- Client to network server
                    707: 
                    708:     my $conserver=shift;
                    709: 
1.49      albertel  710:     &status("Opening TCP $conserver");
1.32      foxr      711:     my $st=120+int(rand(240)); # Sleep before opening:
1.29      www       712: 
1.49      albertel  713:     unless (
                    714: 	    $remotesock = IO::Socket::INET->new(PeerAddr => $hostip{$conserver},
                    715: 						PeerPort => $perlvar{'londPort'},
                    716: 						Proto    => "tcp",
                    717: 						Type     => SOCK_STREAM)
                    718: 	   ) {
                    719: 
                    720: 	&logthis(
                    721: 		 "<font color=blue>WARNING: Couldn't connect to $conserver ($st secs): </font>");
                    722: 	sleep($st);
                    723: 	exit;
                    724:     };
1.29      www       725: # ----------------------------------------------------------------- Init dialog
                    726: 
1.49      albertel  727:     &logthis("<font color=green>INFO Connected to $conserver, initing</font>");
                    728:     &status("Init dialogue: $conserver");
1.29      www       729: 
1.49      albertel  730:     $answer = londtransaction($remotesock, "init", 60);
1.33      foxr      731:     chomp($answer);
                    732:     $answer = londtransaction($remotesock, $answer, 60);
                    733:     chomp($answer);
1.29      www       734: 
1.49      albertel  735:     if ($@=~/timeout/) {
                    736: 	&logthis("Timed out during init.. exiting");
                    737: 	exit;
                    738:     }
                    739: 
                    740:     if ($answer ne 'ok') {
                    741: 	&logthis("Init reply: >$answer<");
                    742: 	my $st=120+int(rand(240));
                    743: 	&logthis("<font color=blue>WARNING: Init failed ($st secs)</font>");
                    744: 	sleep($st);
                    745: 	exit;
                    746:     }
                    747: 
                    748:     $answer = londtransaction($remotesock,"sethost:$conserver",60);
                    749:     chomp($answer);
                    750:     if ( $answer ne 'ok') {
                    751: 	&logthis('<font color="blue">WARNING: unable to specify remote host'.
                    752: 		 $answer.'</font>');
                    753:     }
                    754: 
                    755:     $answer = londtransaction($remotesock,"version:$VERSION",60);
                    756:     chomp($answer);
                    757:     if ($answer =~ /^version:/) {
                    758: 	$remoteVERSION=(split(/:/,$answer))[1];
                    759:     } else {
                    760: 	&logthis('<font color="blue">WARNING: request remote version failed :'.
                    761: 		 $answer.': my version is :'.$VERSION.':</font>');
                    762:     }
1.29      www       763: 
1.49      albertel  764:     sleep 5;
                    765:     &status("Ponging $conserver");
1.53      albertel  766:     $answer= londtransaction($remotesock,"pong",60);
1.49      albertel  767:     chomp($answer);
                    768:     if ($answer!~/^$conserver/) {
                    769: 	&logthis("Pong reply: >$answer<");
                    770:     }
1.29      www       771: # ----------------------------------------------------------- Initialize cipher
                    772: 
1.49      albertel  773:     &status("Initialize cipher");
1.53      albertel  774:     my $buildkey=londtransaction($remotesock,"ekey",60);
1.49      albertel  775:     my $key=$conserver.$perlvar{'lonHostID'};
                    776:     $key=~tr/a-z/A-Z/;
                    777:     $key=~tr/G-P/0-9/;
                    778:     $key=~tr/Q-Z/0-9/;
                    779:     $key=$key.$buildkey.$key.$buildkey.$key.$buildkey;
                    780:     $key=substr($key,0,32);
                    781:     my $cipherkey=pack("H32",$key);
                    782:     if ($cipher=new IDEA $cipherkey) {
                    783: 	&logthis("Secure connection initialized");
                    784:     } else {
                    785: 	my $st=120+int(rand(240));
                    786: 	&logthis("<font color=blue>WARNING: ".
                    787: 		 "Could not establish secure connection ($st secs)!</font>");
                    788: 	sleep($st);
                    789: 	exit;
                    790:     }
1.32      foxr      791:     &logthis("<font color=green> Remote open success </font>");
1.8       harris41  792: }
1.30      www       793: 
                    794: 
                    795: 
                    796: # grabs exception and records it to log before exiting
                    797: sub catchexception {
                    798:     my ($signal)=@_;
                    799:     $SIG{QUIT}='DEFAULT';
                    800:     $SIG{__DIE__}='DEFAULT';
                    801:     chomp($signal);
                    802:     &logthis("<font color=red>CRITICAL: "
                    803:      ."ABNORMAL EXIT. Child $$ for server [$wasserver] died through "
1.33      foxr      804:      ."\"$signal\" with parameter </font>");
                    805:     die("Signal abend");
1.30      www       806: }
                    807: 
                    808: # -------------------------------------- Routines to see if other box available
                    809: 
1.32      foxr      810: #sub online {
                    811: #    my $host=shift;
                    812: #    &status("Pinging ".$host);
                    813: #    my $p=Net::Ping->new("tcp",20);
                    814: #    my $online=$p->ping("$host");
                    815: #    $p->close();
                    816: #    undef ($p);
                    817: #    return $online;
                    818: #}
1.30      www       819: 
                    820: sub connected {
                    821:     my ($local,$remote)=@_;
                    822:     &status("Checking connection $local to $remote");
                    823:     $local=~s/\W//g;
                    824:     $remote=~s/\W//g;
                    825: 
                    826:     unless ($hostname{$local}) { return 'local_unknown'; }
                    827:     unless ($hostname{$remote}) { return 'remote_unknown'; }
                    828: 
1.32      foxr      829:     #unless (&online($hostname{$local})) { return 'local_offline'; }
1.30      www       830: 
                    831:     my $ua=new LWP::UserAgent;
                    832:     
                    833:     my $request=new HTTP::Request('GET',
                    834:       "http://".$hostname{$local}.'/cgi-bin/ping.pl?'.$remote);
                    835: 
                    836:     my $response=$ua->request($request);
                    837: 
                    838:     unless ($response->is_success) { return 'local_error'; }
                    839: 
                    840:     my $reply=$response->content;
                    841:     $reply=(split("\n",$reply))[0];
                    842:     $reply=~s/\W//g;
                    843:     if ($reply ne $remote) { return $reply; }
                    844:     return 'ok';
                    845: }
                    846: 
                    847: 
                    848: 
                    849: sub hangup {
                    850:     foreach (keys %children) {
                    851:         $wasserver=$children{$_};
                    852:         &status("Closing $wasserver");
                    853:         &logthis('Closing '.$wasserver.': '.&subreply('exit',$wasserver));
                    854:         &status("Kill PID $_ for $wasserver");
                    855: 	kill ('INT',$_);
                    856:     }
                    857: }
                    858: 
                    859: sub HUNTSMAN {                      # signal handler for SIGINT
                    860:     local($SIG{CHLD}) = 'IGNORE';   # we're going to kill our children
                    861:     &hangup();
                    862:     my $execdir=$perlvar{'lonDaemons'};
                    863:     unlink("$execdir/logs/lonc.pid");
                    864:     &logthis("<font color=red>CRITICAL: Shutting down</font>");
                    865:     exit;                           # clean up with dignity
                    866: }
                    867: 
                    868: sub HUPSMAN {                      # signal handler for SIGHUP
                    869:     local($SIG{CHLD}) = 'IGNORE';  # we're going to kill our children
                    870:     &hangup();
                    871:     &logthis("<font color=red>CRITICAL: Restarting</font>");
1.50      foxr      872:     my $execdir=$perlvar{'lonDaemons'};
1.30      www       873:     unlink("$execdir/logs/lonc.pid");
                    874:     exec("$execdir/lonc");         # here we go again
                    875: }
                    876: 
                    877: sub checkchildren {
                    878:     &initnewstatus();
                    879:     &logstatus();
                    880:     &logthis('Going to check on the children');
                    881:     foreach (sort keys %children) {
                    882: 	sleep 1;
                    883:         unless (kill 'USR1' => $_) {
                    884: 	    &logthis ('<font color=red>CRITICAL: Child '.$_.' is dead</font>');
                    885:             &logstatus($$.' is dead');
                    886:         } 
                    887:     }
                    888: }
                    889: 
                    890: sub USRMAN {
                    891:     &logthis("USR1: Trying to establish connections again");
1.39      foxr      892:     #
                    893:     #  It is really important not to just clear the childatt hash or we will
                    894:     #  lose all memory of the children.  What we really want to do is this:
                    895:     #  For each index where childatt is >= $childmaxattempts
                    896:     #  Zero the associated counter and do a make_child for the host.
                    897:     #  Regardles, the childatt entry is zeroed:
                    898:     my $host;
                    899:     foreach $host (keys %childatt) {
                    900: 	if ($childatt{$host} >= $childmaxattempts) {
                    901: 	    $childatt{$host} = 0;
                    902: 	    &logthis("<font color=green>INFO: Restarting child for server: "
                    903: 		     .$host."</font>\n");
                    904: 	    make_new_child($host);
                    905: 	}
                    906: 	else {
                    907: 	    $childatt{$host} = 0;
                    908: 	}
                    909:     }
                    910:     &checkchildren();		# See if any children are still dead...
1.30      www       911: }
                    912: 
                    913: # -------------------------------------------------- Non-critical communication
                    914: sub subreply { 
                    915:  my ($cmd,$server)=@_;
                    916:  my $answer='';
                    917:  if ($server ne $perlvar{'lonHostID'}) { 
                    918:     my $peerfile="$perlvar{'lonSockDir'}/$server";
                    919:     my $sclient=IO::Socket::UNIX->new(Peer    =>"$peerfile",
                    920:                                       Type    => SOCK_STREAM,
                    921:                                       Timeout => 10)
                    922:        or return "con_lost";
                    923: 
                    924: 
1.33      foxr      925:     $answer = londtransaction($sclient, $cmd, 10);
                    926: 
1.30      www       927:     if ((!$answer) || ($@=~/timeout/)) { $answer="con_lost"; }
                    928:     $SIG{ALRM}='DEFAULT';
                    929:     $SIG{__DIE__}=\&catchexception;
                    930:  } else { $answer='self_reply'; }
                    931:  return $answer;
                    932: }
                    933: 
                    934: # --------------------------------------------------------------------- Logging
                    935: 
                    936: sub logthis {
                    937:     my $message=shift;
                    938:     my $execdir=$perlvar{'lonDaemons'};
                    939:     my $fh=IO::File->new(">>$execdir/logs/lonc.log");
                    940:     my $now=time;
                    941:     my $local=localtime($now);
                    942:     $lastlog=$local.': '.$message;
                    943:     print $fh "$local ($$) [$conserver] [$status]: $message\n";
                    944: }
                    945: 
1.33      foxr      946: #--------------------------------------  londtransaction:
                    947: #  
                    948: #  Performs a transaction with lond with timeout support.
                    949: #    result = londtransaction(socket,request,timeout)
                    950: #
                    951: sub londtransaction {
                    952:     my ($socket, $request, $tmo) = @_;
                    953: 
                    954:     if($DEBUG) {
                    955: 	&logthis("londtransaction request: $request");
                    956:     }
                    957: 
                    958:     # Set the signal handlers: ALRM for timeout and disble the others.
                    959: 
                    960:     $SIG{ALRM} = sub { die "timeout" };
                    961:     $SIG{__DIE__} = 'DEFAULT';
                    962:     
                    963:     # Disable all but alarm so that only that can interupt the
                    964:     # send /receive.
                    965:     #
                    966:     my $sigset = POSIX::SigSet->new(QUIT, USR1, HUP, INT, TERM);
                    967:     my $priorsigs = POSIX::SigSet->new;
                    968:     unless (defined sigprocmask(SIG_BLOCK, $sigset, $priorsigs)) {
                    969: 	&logthis("<font color=red> CRITICAL -- londtransaction ".
                    970: 		"failed to block signals </font>");
                    971: 	die "could not block signals in londtransaction";
                    972:     }
                    973:     $answer = '';
                    974:     #
                    975:     #  Send request to lond.
                    976:     #
                    977:     eval { 
                    978: 	alarm($tmo);
                    979: 	print $socket "$request\n";
                    980: 	alarm(0);
                    981:     };
                    982:     #  If request didn't timeout, try for the response.
                    983:     #
                    984: 
                    985:     if ($@!~/timeout/) {
                    986: 	eval {
                    987: 	    alarm($tmo);
                    988: 	    $answer = <$socket>;
                    989: 	    if($DEBUG) {
                    990: 		&logthis("Received $answer in londtransaction");
                    991: 	    }
                    992: 	    alarm(0);
                    993: 	};
                    994:     } else {
1.55    ! albertel  995: 	&logthis("lonc - $conserver - suiciding on send Timeout");
        !           996: 	die("lonc - $conserver - suiciding on send Timeout");
1.33      foxr      997:     }
1.47      albertel  998:     if ($@ =~ /timeout/) {
1.55    ! albertel  999: 	&logthis("lonc - $conserver - suiciding on read Timeout");
        !          1000: 	die("lonc - $conserver - suiciding on read Timeout");
1.33      foxr     1001:     }
                   1002:     #
                   1003:     # Restore the initial sigmask set.
                   1004:     #
                   1005:     unless (defined sigprocmask(SIG_UNBLOCK, $priorsigs)) {
                   1006: 	&logthis("<font color=red> CRITICAL -- londtransaction ".
                   1007: 		"failed to re-enable signal processing. </font>");
                   1008: 	die "londtransaction failed to re-enable signals";
                   1009:     }
                   1010:     #
                   1011:     # go back to the prior handler set.
                   1012:     #
                   1013:     $SIG{ALRM} = 'DEFAULT';
                   1014:     $SIG{__DIE__} = \&cathcexception;
                   1015: 
                   1016:     #    chomp $answer;
                   1017:     if ($DEBUG) {
                   1018: 	&logthis("Returning $answer in londtransaction");
                   1019:     }
                   1020:     return $answer;
                   1021: 
                   1022: }
1.30      www      1023: 
                   1024: sub logperm {
                   1025:     my $message=shift;
                   1026:     my $execdir=$perlvar{'lonDaemons'};
                   1027:     my $now=time;
                   1028:     my $local=localtime($now);
                   1029:     my $fh=IO::File->new(">>$execdir/logs/lonnet.perm.log");
                   1030:     print $fh "$now:$message:$local\n";
                   1031: }
                   1032: # ------------------------------------------------------------------ Log status
                   1033: 
                   1034: sub logstatus {
                   1035:     my $docdir=$perlvar{'lonDocRoot'};
                   1036:     my $fh=IO::File->new(">>$docdir/lon-status/loncstatus.txt");
                   1037:     print $fh $$."\t".$conserver."\t".$status."\t".$lastlog."\n";
                   1038: }
                   1039: 
                   1040: sub initnewstatus {
                   1041:     my $docdir=$perlvar{'lonDocRoot'};
                   1042:     my $fh=IO::File->new(">$docdir/lon-status/loncstatus.txt");
                   1043:     my $now=time;
                   1044:     my $local=localtime($now);
                   1045:     print $fh "LONC status $local - parent $$\n\n";
                   1046: }
                   1047: 
                   1048: # -------------------------------------------------------------- Status setting
                   1049: 
                   1050: sub status {
                   1051:     my $what=shift;
                   1052:     my $now=time;
                   1053:     my $local=localtime($now);
                   1054:     $status=$local.': '.$what;
1.43      www      1055:     $0='lonc: '.$what.' '.$local;
1.30      www      1056: }
                   1057: 
                   1058: 
1.1       albertel 1059: 
1.23      harris41 1060: # ----------------------------------- POD (plain old documentation, CPAN style)
                   1061: 
                   1062: =head1 NAME
                   1063: 
                   1064: lonc - LON TCP-MySQL-Server Daemon for handling database requests.
                   1065: 
                   1066: =head1 SYNOPSIS
                   1067: 
1.31      harris41 1068: Usage: B<lonc>
                   1069: 
1.23      harris41 1070: Should only be run as user=www.  This is a command-line script which
1.31      harris41 1071: is invoked by B<loncron>.  There is no expectation that a typical user
                   1072: will manually start B<lonc> from the command-line.  (In other words,
                   1073: DO NOT START B<lonc> YOURSELF.)
1.23      harris41 1074: 
1.51      bowersj2 1075: =head1 OVERVIEW
                   1076: 
                   1077: =head2 Physical Overview
                   1078: 
                   1079: =begin latex 
                   1080: 
                   1081: \begin{figure} 
                   1082:   \begin{center}
                   1083:     \includegraphics[width=0.65\paperwidth,keepaspectratio]{LONCAPA_Network_Diagram}
                   1084:   \end{center}
                   1085:   \caption{\label{Overview_Of_Network}Overview of Network}
                   1086: \end{figure}
                   1087: 
                   1088: =end latex
                   1089: 
                   1090: Physically, the Network consists of relatively inexpensive
                   1091: upper-PC-class server machines which are linked through the commodity
                   1092: internet in a load-balancing, dynamically content-replicating and
                   1093: failover-secure way.
                   1094: 
                   1095: All machines in the Network are connected with each other through
                   1096: two-way persistent TCP/IP connections. Clients (B<B>, B<F>, B<G> and
                   1097: B<H> in Fig. Overview of Network) connect to the servers via standard
                   1098: HTTP. There are two classes of servers, B<Library Servers> (B<A> and
                   1099: B<E> in Fig. Overview of Network) and B<Access Servers> (B<C>, B<D>,
                   1100: B<I> and B<J> in Fig. Overview of Network).
                   1101: 
                   1102: B<Library Servers> X<library server> X<server, library> are used to
                   1103: store all personal records of a set of users, and are responsible for
                   1104: their initial authentication when a session is opened on any server in
                   1105: the Network. For Authors, Library Servers also hosts their
                   1106: construction area and the authoritative copy of the current and
                   1107: previous versions of every resource that was published by that
                   1108: author. Library servers can be used as backups to host sessions when
                   1109: all access servers in the Network are overloaded. Otherwise, for
                   1110: learners, access servers are used to host the sessions. Library
                   1111: servers need to have strong I/O capabilities.
                   1112: 
                   1113: B<Access Servers> X<access server> X<server, access> provide LON-CAPA
                   1114: service to users, using the library servers as their data source. The
                   1115: network is designed so that the number of concurrent sessions can be
                   1116: increased over a wide range by simply adding additional access servers
                   1117: before having to add additional library servers. Preliminary tests
                   1118: showed that a library server could handle up to 10 access servers
                   1119: fully parallel. Access servers can generally be cheaper hardware then
                   1120: library servers require.
                   1121: 
                   1122: The Network is divided into B<domains> X<domain>, which are logical
                   1123: boundaries between participating institutions. These domains can be
                   1124: used to limit the flow of personal user information across the
                   1125: network, set access privileges and enforce royalty schemes. LON-CAPA
                   1126: domains bear no relationship to any other domain, including domains
                   1127: used by the DNS system; LON-CAPA domains may be freely configured in
                   1128: any manner that suits your use pattern.
                   1129: 
                   1130: =head2 Example Transactions
                   1131: 
                   1132: Fig. Overview of Network also depicts examples for several kinds of
                   1133: transactions conducted across the Network.
                   1134: 
                   1135: An instructor at client B<B> modifies and publishes a resource on her
                   1136: Home Server B<A>. Server B<A> has a record of all server machines
                   1137: currently subscribed to this resource, and replicates it to servers
                   1138: B<D> and B<I>. However, server B<D> is currently offline, so the
                   1139: update notification gets buffered on B<A> until B<D> comes online
                   1140: again. Servers B<C> and B<J> are currently not subscribed to this
                   1141: resource.
                   1142: 
                   1143: Learners B<F> and B<G> have open sessions on server B<I>, and the new
                   1144: resource is immediately available to them.
                   1145: 
                   1146: Learner B<H> tries to connect to server B<I> for a new session,
                   1147: however, the machine is not reachable, so he connects to another
                   1148: Access Server B<J> instead. This server currently does not have all
                   1149: necessary resources locally present to host learner B<H>, but
                   1150: subscribes to them and replicates them as they are accessed by B<H>.
                   1151: 
                   1152: Learner B<H> solves a problem on server B<J>. Library Server B<E> is
                   1153: B<H>'s Home Server, so this information gets forwarded to B<E>, where
                   1154: the records of H are updated.
                   1155: 
1.52      bowersj2 1156: =head2 lond, lonc, and lonnet
1.51      bowersj2 1157: 
                   1158: =begin latex
                   1159: 
                   1160: \begin{figure}
1.52      bowersj2 1161: \includegraphics[width=0.65\paperwidth,keepaspectratio]{LONCAPA_Network_Diagram2}
1.51      bowersj2 1162:   \caption{\label{Overview_Of_Network_Communication}Overview of
                   1163: Network Communication} \end{figure}
                   1164: 
                   1165: =end latex
                   1166: 
                   1167: Fig. Overview of Network Communication elaborates on the details of
                   1168: this network infrastructure. It depicts three servers (B<A>, B<B> and
                   1169: B<C>) and a client who has a session on server B<C>.
                   1170: 
                   1171: As B<C> accesses different resources in the system, different
                   1172: handlers, which are incorporated as modules into the child processes
                   1173: of the web server software, process these requests.
                   1174: 
                   1175: Our current implementation uses C<mod_perl> inside of the Apache web
                   1176: server software. As an example, server B<C> currently has four active
                   1177: web server software child processes. The chain of handlers dealing
                   1178: with a certain resource is determined by both the server content
                   1179: resource area (see below) and the MIME type, which in turn is
                   1180: determined by the URL extension. For most URL structures, both an
                   1181: authentication handler and a content handler are registered.
                   1182: 
                   1183: Handlers use a common library C<lonnet> X<lonnet> to interact with
                   1184: both locally present temporary session data and data across the server
                   1185: network. For example, lonnet provides routines for finding the home
                   1186: server of a user, finding the server with the lowest loadavg, sending
                   1187: simple command-reply sequences, and sending critical messages such as
                   1188: a homework completion, etc. For a non-critical message, the routines
                   1189: reply with a simple "connection lost" if the message could not be
                   1190: delivered. For critical messages, lonnet tries to re-establish
                   1191: connections, re-send the command, etc. If no valid reply could be
                   1192: received, it answers "connection deferred" and stores the message in
                   1193: buffer space to be sent at a later point in time. Also, failed
                   1194: critical messages are logged.
                   1195: 
                   1196: The interface between C<lonnet> and the Network is established by a
                   1197: multiplexed UNIX domain socket, denoted B<DS> in Fig. Overview of
                   1198: Network Communication. The rationale behind this rather involved
                   1199: architecture is that httpd processes (Apache children) dynamically
                   1200: come and go on the timescale of minutes, based on workload and number
                   1201: of processed requests. Over the lifetime of an httpd child, however,
                   1202: it has to establish several hundred connections to several different
                   1203: servers in the Network.
                   1204: 
                   1205: On the other hand, establishing a TCP/IP connection is resource
                   1206: consuming for both ends of the line, and to optimize this connectivity
                   1207: between different servers, connections in the Network are designed to
                   1208: be persistent on the timescale of months, until either end is
                   1209: rebooted. This mechanism will be elaborated on below.
                   1210: 
                   1211: =begin latex
                   1212: 
                   1213: \begin{figure}
                   1214: \begin{lyxcode}
                   1215: msul1:msu:library:zaphod.lite.msu.edu:35.8.63.51
                   1216: 
                   1217: msua1:msu:access:agrajag.lite.msu.edu:35.8.63.68
                   1218: 
                   1219: msul2:msu:library:frootmig.lite.msu.edu:35.8.63.69
                   1220: 
                   1221: msua2:msu:access:bistromath.lite.msu.edu:35.8.63.67
                   1222: 
                   1223: hubl14:hub:library:hubs128-pc-14.cl.msu.edu:35.8.116.34
                   1224: 
                   1225: hubl15:hub:library:hubs128-pc-15.cl.msu.edu:35.8.116.35
                   1226: 
                   1227: hubl16:hub:library:hubs128-pc-16.cl.msu.edu:35.8.116.36
                   1228: 
                   1229: huba20:hub:access:hubs128-pc-20.cl.msu.edu:35.8.116.40
                   1230: 
                   1231: huba21:hub:access:hubs128-pc-21.cl.msu.edu:35.8.116.41
                   1232: 
                   1233: huba22:hub:access:hubs128-pc-22.cl.msu.edu:35.8.116.42
                   1234: 
                   1235: huba23:hub:access:hubs128-pc-23.cl.msu.edu:35.8.116.43
                   1236: 
                   1237: hubl25:other:library:hubs128-pc-25.cl.msu.edu:35.8.116.45
                   1238: 
                   1239: huba27:other:access:hubs128-pc-27.cl.msu.edu:35.8.116.47
                   1240: \end{lyxcode}
                   1241: 
                   1242: \caption{\label{Example_Of_hosts.tab}Example of Hosts Lookup table\texttt{/home/httpd/lonTabs/hosts.tab}} 
                   1243: \end{figure}
                   1244: 
                   1245: =end latex
                   1246: 
                   1247: Establishing a connection to a UNIX domain socket is far less resource
                   1248: consuming than the establishing of a TCP/IP connection. C<lonc>
                   1249: X<lonc> is a proxy daemon that forks off a child for every server in
                   1250: the Network. Which servers are members of the Network is determined by
                   1251: a lookup table, such as the one in Fig. Examples of Hosts. In order,
                   1252: the entries denote an internal name for the server, the domain of the
                   1253: server, the type of the server, the host name and the IP address.
                   1254: 
                   1255: The C<lonc> parent process maintains the population and listens for
                   1256: signals to restart or shutdown, as well as I<USR1>. Every child
                   1257: establishes a multiplexed UNIX domain socket for its server and opens
                   1258: a TCP/IP connection to the lond daemon (discussed below) on the remote
                   1259: machine, which it keeps alive. If the connection is interrupted, the
                   1260: child dies, whereupon the parent makes several attempts to fork
                   1261: another child for that server.
                   1262: 
                   1263: When starting a new child (a new connection), first an init-sequence
                   1264: is carried out, which includes receiving the information from the
                   1265: remote C<lond> which is needed to establish the 128-bit encryption key
                   1266: - the key is different for every connection. Next, any buffered
                   1267: (delayed) messages for the server are sent.
                   1268: 
                   1269: In normal operation, the child listens to the UNIX socket, forwards
                   1270: requests to the TCP connection, gets the reply from C<lond>, and sends
                   1271: it back to the UNIX socket. Also, C<lonc> takes care to the encryption
                   1272: and decryption of messages.
                   1273: 
                   1274: C<lond> X<lond> is the remote end of the TCP/IP connection and acts as
                   1275: a remote command processor. It receives commands, executes them, and
                   1276: sends replies. In normal operation, a C<lonc> child is constantly
                   1277: connected to a dedicated C<lond> child on the remote server, and the
                   1278: same is true vice versa (two persistent connections per server
                   1279: combination).
                   1280: 
                   1281: lond listens to a TCP/IP port (denoted B<P> in Fig. Overview of
                   1282: Network Communication) and forks off enough child processes to have
                   1283: one for each other server in the network plus two spare children. The
                   1284: parent process maintains the population and listens for signals to
                   1285: restart or shutdown. Client servers are authenticated by IP.
                   1286: 
                   1287: When a new client server comes online, C<lond> sends a signal I<USR1>
                   1288: to lonc, whereupon C<lonc> tries again to reestablish all lost
                   1289: connections, even if it had given up on them before - a new client
                   1290: connecting could mean that that machine came online again after an
                   1291: interruption.
                   1292: 
                   1293: The gray boxes in Fig. Overview of Network Communication denote the
                   1294: entities involved in an example transaction of the Network. The Client
                   1295: is logged into server B<C>, while server B<B> is her Home
                   1296: Server. Server B<C> can be an access server or a library server, while
                   1297: server B<B> is a library server. She submits a solution to a homework
                   1298: problem, which is processed by the appropriate handler for the MIME
                   1299: type "problem". Through C<lonnet>, the handler writes information
                   1300: about this transaction to the local session data. To make a permanent
                   1301: log entry, C<lonnet> establishes a connection to the UNIX domain
                   1302: socket for server B<B>. C<lonc> receives this command, encrypts it,
                   1303: and sends it through the persistent TCP/IP connection to the TCP/IP
                   1304: port of the remote C<lond>. C<lond> decrypts the command, executes it
                   1305: by writing to the permanent user data files of the client, and sends
                   1306: back a reply regarding the success of the operation. If the operation
                   1307: was unsuccessful, or the connection would have broken down, C<lonc>
                   1308: would write the command into a FIFO buffer stack to be sent again
                   1309: later. C<lonc> now sends a reply regarding the overall success of the
                   1310: operation to C<lonnet> via the UNIX domain port, which is eventually
                   1311: received back by the handler.
                   1312: 
                   1313: =head2 Dynamic Resource Replication
                   1314: 
                   1315: Since resources are assembled into higher order resources simply by
                   1316: reference, in principle it would be sufficient to retrieve them from
                   1317: the respective Home Servers of the authors. However, there are several
                   1318: problems with this simple approach: since the resource assembly
                   1319: mechanism is designed to facilitate content assembly from a large
                   1320: number of widely distributed sources, individual sessions would depend
                   1321: on a large number of machines and network connections to be available,
                   1322: thus be rather fragile. Also, frequently accessed resources could
                   1323: potentially drive individual machines in the network into overload
                   1324: situations.
                   1325: 
                   1326: Finally, since most resources depend on content handlers on the Access
                   1327: Servers to be served to a client within the session context, the raw
                   1328: source would first have to be transferred across the Network from the
                   1329: respective Library Server to the Access Server, processed there, and
                   1330: then transferred on to the client.
                   1331: 
                   1332: =begin latex
                   1333: 
                   1334: \begin{figure}
                   1335: \includegraphics[width=0.75\paperwidth,keepaspectratio]{Dynamic_Replication_Request}
                   1336:   \caption{\label{Dynamic_Replication}Dynamic Replication} 
                   1337: \end{figure}
                   1338: 
                   1339: =end latex
                   1340: 
                   1341: To enable resource assembly in a reliable and scalable way, a dynamic
                   1342: resource replication scheme was developed. Fig. "Dynamic Replication"
                   1343: shows the details of this mechanism.
                   1344: 
                   1345: Anytime a resource out of the resource space is requested, a handler
                   1346: routine is called which in turn calls the replication routine. As a
                   1347: first step, this routines determines whether or not the resource is
                   1348: currently in replication transfer (Step B<D1a>). During replication
                   1349: transfer, the incoming data is stored in a temporary file, and Step
                   1350: B<D1a> checks for the presence of that file. If transfer of a resource
                   1351: is actively going on, the controlling handler receives an error
                   1352: message, waits for a few seconds, and then calls the replication
                   1353: routine again. If the resource is still in transfer, the client will
                   1354: receive the message "Service currently not available".
                   1355: 
                   1356: In the next step (Step B<D1b>), the replication routine checks if the
                   1357: URL is locally present. If it is, the replication routine returns OK
                   1358: to the controlling handler, which in turn passes the request on to the
                   1359: next handler in the chain.
                   1360: 
                   1361: If the resource is not locally present, the Home Server of the
                   1362: resource author (as extracted from the URL) is determined (Step
                   1363: B<D2>). This is done by contacting all library servers in the author?s
                   1364: domain (as determined from the lookup table, see Fig. 1.1.2B). In Step
                   1365: B<D2b> a query is sent to the remote server whether or not it is the
                   1366: Home Server of the author (in our current implementation, an
                   1367: additional cache is used to store already identified Home Servers (not
                   1368: shown in the figure)). In Step B<D2c>, the remote server answers the
                   1369: query with True or False. If the Home Server was found, the routine
                   1370: continues, otherwise it contacts the next server (Step D2a). If no
                   1371: server could be found, a "File not Found" error message is issued. In
                   1372: our current implementation, in this step the Home Server is also
                   1373: written into a cache for faster access if resources by the same author
                   1374: are needed again (not shown in the figure).
                   1375: 
                   1376: =begin latex
                   1377: 
                   1378: \begin{figure}
                   1379: \includegraphics[width=0.75\paperwidth,keepaspectratio]{Dynamic_Replication_Change}
                   1380:   \caption{\label{Dynamic_Replication_Change}Dynamic Replication: Change} \end{figure}
                   1381: 
                   1382: =end latex
                   1383: 
                   1384: In Step B<D3a>, the routine sends a subscribe command for the URL to
                   1385: the Home Server of the author. The Home Server first determines if the
                   1386: resource is present, and if the access privileges allow it to be
                   1387: copied to the requesting server (B<D3b>). If this is true, the
                   1388: requesting server is added to the list of subscribed servers for that
                   1389: resource (Step B<D3c>). The Home Server will reply with either OK or
                   1390: an error message, which is determined in Step D4. If the remote
                   1391: resource was not present, the error message "File not Found" will be
                   1392: passed on to the client, if the access was not allowed, the error
                   1393: message "Access Denied" is passed on. If the operation succeeded, the
                   1394: requesting server sends an HTTP request for the resource out of the
                   1395: C</raw> server content resource area of the Home Server.
                   1396: 
                   1397: The Home Server will then check if the requesting server is part of
                   1398: the network, and if it is subscribed to the resource (Step B<D5b>). If
                   1399: it is, it will send the resource via HTTP to the requesting server
                   1400: without any content handlers processing it (Step B<D5c>). The
                   1401: requesting server will store the incoming data in a temporary data
                   1402: file (Step B<D5a>) - this is the file that Step B<D1a> checks for. If
                   1403: the transfer could not complete, and appropriate error message is sent
                   1404: to the client (Step B<D6>). Otherwise, the transferred temporary file
                   1405: is renamed as the actual resource, and the replication routine returns
                   1406: OK to the controlling handler (Step B<D7>).
                   1407: 
                   1408: Fig. "Dynamic Replication: Change" depicts the process of modifying a
                   1409: resource. When an author publishes a new version of a resource, the
                   1410: Home Server will contact every server currently subscribed to the
                   1411: resource (Step B<U1>), as determined from the list of subscribed
                   1412: servers for the resource generated in Step B<D3c>. The subscribing
                   1413: servers will receive and acknowledge the update message (Step
                   1414: B<U1c>). The update mechanism finishes when the last subscribed server
                   1415: has been contacted (messages to unreachable servers are buffered).
                   1416: 
                   1417: Each subscribing server will check if the resource in question had
                   1418: been accessed recently, that is, within a configurable amount of time
                   1419: (Step B<U2>).
                   1420: 
                   1421: If the resource had not been accessed recently, the local copy of the
                   1422: resource is deleted (Step B<U3a>) and an unsubscribe command is sent
                   1423: to the Home Server (Step B<U3b>). The Home Server will check if the
                   1424: server had indeed originally subscribed to the resource (Step B<U3c>)
                   1425: and then delete the server from the list of subscribed servers for the
                   1426: resource (Step B<U3d>).
                   1427: 
                   1428: If the resource had been accessed recently, the modified resource will
                   1429: be copied over using the same mechanism as in Step B<D5a> through
                   1430: B<D7>, which represents steps Steps B<U4a> through B<U6> in the
                   1431: replication figure.
                   1432: 
1.52      bowersj2 1433: =head2 Load Balancing 
1.51      bowersj2 1434: 
1.52      bowersj2 1435: X<load balancing>C<lond> provides a function to query the server's current loadavg. As
1.51      bowersj2 1436: a configuration parameter, one can determine the value of loadavg,
                   1437: which is to be considered 100%, for example, 2.00.
                   1438: 
                   1439: Access servers can have a list of spare access servers,
                   1440: C</home/httpd/lonTabs/spares.tab>, to offload sessions depending on
                   1441: own workload. This check happens is done by the login handler. It
                   1442: re-directs the login information and session to the least busy spare
                   1443: server if itself is overloaded. An additional round-robin IP scheme
                   1444: possible. See Fig. "Load Balancing Sample" for an example of a
                   1445: load-balancing scheme.
                   1446: 
                   1447: =begin latex
                   1448: 
                   1449: \begin{figure}
                   1450: \includegraphics[width=0.75\paperwidth,keepaspectratio]{Load_Balancing_Example}
                   1451:   \caption{\label{Load_Balancing_Example}Load Balancing Example} \end{figure}
                   1452: 
                   1453: =end latex
                   1454: 
1.23      harris41 1455: =head1 DESCRIPTION
                   1456: 
                   1457: Provides persistent TCP connections to the other servers in the network
                   1458: through multiplexed domain sockets
                   1459: 
1.31      harris41 1460: B<lonc> forks off children processes that correspond to the other servers
                   1461: in the network.  Management of these processes can be done at the
                   1462: parent process level or the child process level.
                   1463: 
1.51      bowersj2 1464: After forking off the children, B<lonc> the B<parent> executes a main
                   1465: loop which simply waits for processes to exit.  As a process exits, a
                   1466: new process managing a link to the same peer as the exiting process is
                   1467: created.
1.33      foxr     1468: 
1.31      harris41 1469: B<logs/lonc.log> is the location of log messages.
                   1470: 
                   1471: The process management is now explained in terms of linux shell commands,
                   1472: subroutines internal to this code, and signal assignments:
                   1473: 
                   1474: =over 4
                   1475: 
                   1476: =item *
                   1477: 
                   1478: PID is stored in B<logs/lonc.pid>
                   1479: 
                   1480: This is the process id number of the parent B<lonc> process.
                   1481: 
                   1482: =item *
                   1483: 
                   1484: SIGTERM and SIGINT
                   1485: 
                   1486: Parent signal assignment:
                   1487:  $SIG{INT}  = $SIG{TERM} = \&HUNTSMAN;
                   1488: 
                   1489: Child signal assignment:
                   1490:  $SIG{INT}  = 'DEFAULT'; (and SIGTERM is DEFAULT also)
                   1491: (The child dies and a SIGALRM is sent to parent, awaking parent from slumber
                   1492:  to restart a new child.)
                   1493: 
                   1494: Command-line invocations:
                   1495:  B<kill> B<-s> SIGTERM I<PID>
                   1496:  B<kill> B<-s> SIGINT I<PID>
                   1497: 
                   1498: Subroutine B<HUNTSMAN>:
                   1499:  This is only invoked for the B<lonc> parent I<PID>.
                   1500: This kills all the children, and then the parent.
                   1501: The B<lonc.pid> file is cleared.
                   1502: 
                   1503: =item *
                   1504: 
                   1505: SIGHUP
                   1506: 
                   1507: Current bug:
                   1508:  This signal can only be processed the first time
                   1509: on the parent process.  Subsequent SIGHUP signals
                   1510: have no effect.
                   1511: 
                   1512: Parent signal assignment:
                   1513:  $SIG{HUP}  = \&HUPSMAN;
                   1514: 
                   1515: Child signal assignment:
                   1516:  none (nothing happens)
                   1517: 
                   1518: Command-line invocations:
                   1519:  B<kill> B<-s> SIGHUP I<PID>
                   1520: 
                   1521: Subroutine B<HUPSMAN>:
                   1522:  This is only invoked for the B<lonc> parent I<PID>,
                   1523: This kills all the children, and then the parent.
                   1524: The B<lonc.pid> file is cleared.
                   1525: 
                   1526: =item *
                   1527: 
                   1528: SIGUSR1
                   1529: 
                   1530: Parent signal assignment:
                   1531:  $SIG{USR1} = \&USRMAN;
                   1532: 
                   1533: Child signal assignment:
                   1534:  $SIG{USR1}= \&logstatus;
                   1535: 
                   1536: Command-line invocations:
                   1537:  B<kill> B<-s> SIGUSR1 I<PID>
                   1538: 
                   1539: Subroutine B<USRMAN>:
                   1540:  When invoked for the B<lonc> parent I<PID>,
                   1541: SIGUSR1 is sent to all the children, and the status of
                   1542: each connection is logged.
                   1543: 
1.23      harris41 1544: 
1.31      harris41 1545: =back
1.23      harris41 1546: 
                   1547: =cut

FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>