1: #!/usr/bin/perl
2: # The LearningOnline Network with CAPA
3: # lonc maintains the connections to remote computers
4: #
5: # $Id: loncnew,v 1.96 2011/01/25 11:02:35 foxr Exp $
6: #
7: # Copyright Michigan State University Board of Trustees
8: #
9: # This file is part of the LearningOnline Network with CAPA (LON-CAPA).
10: ## LON-CAPA is free software; you can redistribute it and/or modify
11: # it under the terms of the GNU General Public License as published by
12: # the Free Software Foundation; either version 2 of the License, or
13: # (at your option) any later version.
14: #
15: # LON-CAPA is distributed in the hope that it will be useful,
16: # but WITHOUT ANY WARRANTY; without even the implied warranty of
17: # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
18: # GNU General Public License for more details.
19: #
20: # You should have received a copy of the GNU General Public License
21: # along with LON-CAPA; if not, write to the Free Software
22: # Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
23: #
24: # /home/httpd/html/adm/gpl.txt
25: #
26: # http://www.lon-capa.org/
27: #
28: #
29: # new lonc handles n request out bver m connections to londs.
30: # This module is based on the Event class.
31: # Development iterations:
32: # - Setup basic event loop. (done)
33: # - Add timer dispatch. (done)
34: # - Add ability to accept lonc UNIX domain sockets. (done)
35: # - Add ability to create/negotiate lond connections (done).
36: # - Add general logic for dispatching requests and timeouts. (done).
37: # - Add support for the lonc/lond requests. (done).
38: # - Add logging/status monitoring. (done)
39: # - Add Signal handling - HUP restarts. USR1 status report. (done)
40: # - Add Configuration file I/O (done).
41: # - Add management/status request interface. (done)
42: # - Add deferred request capability. (done)
43: # - Detect transmission timeouts. (done)
44: #
45:
46: use strict;
47: use lib "/home/httpd/lib/perl/";
48: use Event qw(:DEFAULT );
49: use POSIX qw(:signal_h);
50: use POSIX;
51: use IO::Socket;
52: use IO::Socket::INET;
53: use IO::Socket::UNIX;
54: use IO::File;
55: use IO::Handle;
56: use Socket;
57: use Crypt::IDEA;
58: use LONCAPA::Queue;
59: use LONCAPA::Stack;
60: use LONCAPA::LondConnection;
61: use LONCAPA::LondTransaction;
62: use LONCAPA::Configuration;
63: use Fcntl qw(:flock);
64:
65:
66: # Read the httpd configuration file to get perl variables
67: # normally set in apache modules:
68:
69: my $perlvarref = LONCAPA::Configuration::read_conf('loncapa.conf');
70: my %perlvar = %{$perlvarref};
71:
72: #
73: # parent and shared variables.
74:
75: my %ChildPid; # by pid -> host.
76: my %ChildHost; # by host.
77: my %listening_to; # Socket->host table for who the parent
78: # is listening to.
79: my %parent_dispatchers; # host-> listener watcher events.
80:
81: my %parent_handlers; # Parent signal handlers...
82:
83: my $MaxConnectionCount = 10; # Will get from config later.
84: my $ClientConnection = 0; # Uniquifier for client events.
85:
86: my $DebugLevel = 0;
87: my $NextDebugLevel= 2; # So Sigint can toggle this.
88: my $IdleTimeout= 5*60; # Seconds to wait prior to pruning connections.
89:
90: my $LogTransactions = 0; # When True, all transactions/replies get logged.
91: my $executable = $0; # Get the full path to me.
92:
93: #
94: # The variables below are only used by the child processes.
95: #
96: my $RemoteHost; # Name of host child is talking to.
97: my $RemoteHostId; # default lonid of host child is talking to.
98: my @all_host_ids;
99: my $UnixSocketDir= $perlvar{'lonSockDir'};
100: my $IdleConnections = Stack->new(); # Set of idle connections
101: my %ActiveConnections; # Connections to the remote lond.
102: my %ActiveTransactions; # LondTransactions in flight.
103: my %ActiveClients; # Serial numbers of active clients by socket.
104: my $WorkQueue = Queue->new(); # Queue of pending transactions.
105: my $ConnectionCount = 0;
106: my $IdleSeconds = 0; # Number of seconds idle.
107: my $Status = ""; # Current status string.
108: my $RecentLogEntry = "";
109: my $ConnectionRetries=5; # Number of connection retries allowed.
110: my $ConnectionRetriesLeft=5; # Number of connection retries remaining.
111: my $LondVersion = "unknown"; # Version of lond we talk with.
112: my $KeyMode = ""; # e.g. ssl, local, insecure from last connect.
113: my $LondConnecting = 0; # True when a connection is being built.
114:
115:
116:
117: my $I_am_child = 0; # True if this is the child process.
118:
119: #
120: # The hash below gives the HTML format for log messages
121: # given a severity.
122: #
123: my %LogFormats;
124:
125: $LogFormats{"CRITICAL"} = "<font color='red'>CRITICAL: %s</font>";
126: $LogFormats{"SUCCESS"} = "<font color='green'>SUCCESS: %s</font>";
127: $LogFormats{"INFO"} = "<font color='yellow'>INFO: %s</font>";
128: $LogFormats{"WARNING"} = "<font color='blue'>WARNING: %s</font>";
129: $LogFormats{"DEFAULT"} = " %s ";
130:
131:
132: # UpdateStatus;
133: # Update the idle status display to show how many connections
134: # are left, retries and other stuff.
135: #
136: sub UpdateStatus {
137: if ($ConnectionRetriesLeft > 0) {
138: ShowStatus(GetServerHost()." Connection count: ".$ConnectionCount
139: ." Retries remaining: ".$ConnectionRetriesLeft
140: ." ($KeyMode)");
141: } else {
142: ShowStatus(GetServerHost()." >> DEAD <<");
143: }
144: }
145:
146:
147: =pod
148:
149: =head2 LogPerm
150:
151: Makes an entry into the permanent log file.
152:
153: =cut
154:
155: sub LogPerm {
156: my $message=shift;
157: my $execdir=$perlvar{'lonDaemons'};
158: my $now=time;
159: my $local=localtime($now);
160: my $fh=IO::File->new(">>$execdir/logs/lonnet.perm.log");
161: chomp($message);
162: print $fh "$now:$message:$local\n";
163: }
164:
165: =pod
166:
167: =head2 Log
168:
169: Logs a message to the log file.
170: Parameters:
171:
172: =item severity
173:
174: One of CRITICAL, WARNING, INFO, SUCCESS used to select the
175: format string used to format the message. if the severity is
176: not a defined severity the Default format string is used.
177:
178: =item message
179:
180: The base message. In addtion to the format string, the message
181: will be appended to a string containing the name of our remote
182: host and the time will be formatted into the message.
183:
184: =cut
185:
186: sub Log {
187:
188: my ($severity, $message) = @_;
189:
190: if(!$LogFormats{$severity}) {
191: $severity = "DEFAULT";
192: }
193:
194: my $format = $LogFormats{$severity};
195:
196: # Put the window dressing in in front of the message format:
197:
198: my $now = time;
199: my $local = localtime($now);
200: my $finalformat = "$local ($$) [$RemoteHost] [$Status] ";
201: $finalformat = $finalformat.$format."\n";
202:
203: # open the file and put the result.
204:
205: my $execdir = $perlvar{'lonDaemons'};
206: my $fh = IO::File->new(">>$execdir/logs/lonc.log");
207: my $msg = sprintf($finalformat, $message);
208: $RecentLogEntry = $msg;
209: print $fh $msg;
210:
211:
212: }
213:
214:
215: =pod
216:
217: =head2 GetPeerName
218:
219: Returns the name of the host that a socket object is connected to.
220:
221: =cut
222:
223: sub GetPeername {
224:
225:
226: my ($connection, $AdrFamily) = @_;
227:
228: my $peer = $connection->peername();
229: my $peerport;
230: my $peerip;
231: if($AdrFamily == AF_INET) {
232: ($peerport, $peerip) = sockaddr_in($peer);
233: my $peername = gethostbyaddr($peerip, $AdrFamily);
234: return $peername;
235: } elsif ($AdrFamily == AF_UNIX) {
236: my $peerfile;
237: ($peerfile) = sockaddr_un($peer);
238: return $peerfile;
239: }
240: }
241: =pod
242:
243: =head2 Debug
244:
245: Invoked to issue a debug message.
246:
247: =cut
248:
249: sub Debug {
250:
251: my ($level, $message) = @_;
252:
253: if ($level <= $DebugLevel) {
254: Log("INFO", "-Debug- $message host = $RemoteHost");
255: }
256: }
257:
258: sub SocketDump {
259:
260: my ($level, $socket) = @_;
261:
262: if($level <= $DebugLevel) {
263: $socket->Dump(-1); # Ensure it will get dumped.
264: }
265: }
266:
267: =pod
268:
269: =head2 ShowStatus
270:
271: Place some text as our pid status.
272: and as what we return in a SIGUSR1
273:
274: =cut
275:
276: sub ShowStatus {
277: my $state = shift;
278: my $now = time;
279: my $local = localtime($now);
280: $Status = $local.": ".$state;
281: $0='lonc: '.$state.' '.$local;
282: }
283:
284: =pod
285:
286: =head2 SocketTimeout
287:
288: Called when an action on the socket times out. The socket is
289: destroyed and any active transaction is failed.
290:
291:
292: =cut
293:
294: sub SocketTimeout {
295: my $Socket = shift;
296: Log("WARNING", "A socket timeout was detected");
297: Debug(5, " SocketTimeout called: ");
298: $Socket->Dump(0);
299: if(exists($ActiveTransactions{$Socket})) {
300: FailTransaction($ActiveTransactions{$Socket});
301: }
302: KillSocket($Socket); # A transaction timeout also counts as
303: # a connection failure:
304: $ConnectionRetriesLeft--;
305: if($ConnectionRetriesLeft <= 0) {
306: Log("CRITICAL", "Host marked DEAD: ".GetServerHost());
307: $LondConnecting = 0;
308: }
309:
310: }
311:
312: #
313: # This function should be called by the child in all cases where it must
314: # exit. The child process must create a lock file for the AF_UNIX socket
315: # in order to prevent connection requests from lonnet in the time between
316: # process exit and the parent picking up the listen again.
317: #
318: # Parameters:
319: # exit_code - Exit status value, however see the next parameter.
320: # message - If this optional parameter is supplied, the exit
321: # is via a die with this message.
322: #
323: sub child_exit {
324: my ($exit_code, $message) = @_;
325:
326: # Regardless of how we exit, we may need to do the lock thing:
327:
328: #
329: # Create a lock file since there will be a time window
330: # between our exit and the parent's picking up the listen
331: # during which no listens will be done on the
332: # lonnet client socket.
333: #
334: my $lock_file = &GetLoncSocketPath().".lock";
335: open(LOCK,">$lock_file");
336: print LOCK "Contents not important";
337: close(LOCK);
338: unlink(&GetLoncSocketPath());
339:
340: if ($message) {
341: die($message);
342: } else {
343: exit($exit_code);
344: }
345: }
346: #----------------------------- Timer management ------------------------
347:
348: =pod
349:
350: =head2 Tick
351:
352: Invoked each timer tick.
353:
354: =cut
355:
356:
357: sub Tick {
358: my ($Event) = @_;
359: my $clock_watcher = $Event->w;
360:
361: my $client;
362: UpdateStatus();
363:
364: # Is it time to prune connection count:
365:
366:
367: if($IdleConnections->Count() &&
368: ($WorkQueue->Count() == 0)) { # Idle connections and nothing to do?
369: $IdleSeconds++;
370: if($IdleSeconds > $IdleTimeout) { # Prune a connection...
371: my $Socket = $IdleConnections->pop();
372: KillSocket($Socket);
373: $IdleSeconds = 0; # Otherwise all connections get trimmed to fast.
374: UpdateStatus();
375: if(($ConnectionCount == 0)) {
376: &child_exit(0);
377:
378: }
379: }
380: } else {
381: $IdleSeconds = 0; # Reset idle count if not idle.
382: }
383: #
384: # For each inflight transaction, tick down its timeout counter.
385: #
386:
387: foreach my $item (keys %ActiveConnections) {
388: my $State = $ActiveConnections{$item}->data->GetState();
389: if ($State ne 'Idle') {
390: Debug(5,"Ticking Socket $State $item");
391: $ActiveConnections{$item}->data->Tick();
392: }
393: }
394: # Do we have work in the queue, but no connections to service them?
395: # If so, try to make some new connections to get things going again.
396: #
397: # Note this code is dead now...
398: #
399: my $Requests = $WorkQueue->Count();
400: if (($ConnectionCount == 0) && ($Requests > 0) && (!$LondConnecting)) {
401: if ($ConnectionRetriesLeft > 0) {
402: Debug(5,"Work but no connections, Make a new one");
403: my $success;
404: $success = &MakeLondConnection;
405: if($success == 0) { # All connections failed:
406: Debug(5,"Work in queue failed to make any connectiouns\n");
407: EmptyQueue(); # Fail pending transactions with con_lost.
408: CloseAllLondConnections(); # Should all be closed but....
409: }
410: } else {
411: $LondConnecting = 0;
412: ShowStatus(GetServerHost()." >>> DEAD!!! <<<");
413: Debug(5,"Work in queue, but gave up on connections..flushing\n");
414: EmptyQueue(); # Connections can't be established.
415: CloseAllLondConnections(); # Should all already be closed but...
416: }
417:
418: }
419: if ($ConnectionCount == 0) {
420: $KeyMode = "";
421: $clock_watcher->cancel();
422: }
423: &UpdateStatus();
424: }
425:
426: =pod
427:
428: =head2 SetupTimer
429:
430: Sets up a 1 per sec recurring timer event. The event handler is used to:
431:
432: =item
433:
434: Trigger timeouts on communications along active sockets.
435:
436: =item
437:
438: Trigger disconnections of idle sockets.
439:
440: =cut
441:
442: sub SetupTimer {
443: Debug(6, "SetupTimer");
444: Event->timer(interval => 1, cb => \&Tick,
445: hard => 1);
446: }
447:
448: =pod
449:
450: =head2 ServerToIdle
451:
452: This function is called when a connection to the server is
453: ready for more work.
454:
455: If there is work in the Work queue the top element is dequeued
456: and the connection will start to work on it. If the work queue is
457: empty, the connection is pushed on the idle connection stack where
458: it will either get another work unit, or alternatively, if it sits there
459: long enough, it will be shut down and released.
460:
461: =cut
462:
463: sub ServerToIdle {
464: my $Socket = shift; # Get the socket.
465: $KeyMode = $Socket->{AuthenticationMode};
466: delete($ActiveTransactions{$Socket}); # Server has no transaction
467:
468: &Debug(5, "Server to idle");
469:
470: # If there's work to do, start the transaction:
471:
472: my $reqdata = $WorkQueue->dequeue(); # This is a LondTransaction
473: if ($reqdata ne undef) {
474: Debug(5, "Queue gave request data: ".$reqdata->getRequest());
475: &StartRequest($Socket, $reqdata);
476:
477: } else {
478:
479: # There's no work waiting, so push the server to idle list.
480: &Debug(5, "No new work requests, server connection going idle");
481: $IdleConnections->push($Socket);
482: }
483: }
484:
485: =pod
486:
487: =head2 ClientWritable
488:
489: Event callback for when a client socket is writable.
490:
491: This callback is established when a transaction reponse is
492: avaiable from lond. The response is forwarded to the unix socket
493: as it becomes writable in this sub.
494:
495: Parameters:
496:
497: =item Event
498:
499: The event that has been triggered. Event->w->data is
500: the data and Event->w->fd is the socket to write.
501:
502: =cut
503:
504: sub ClientWritable {
505: my $Event = shift;
506: my $Watcher = $Event->w;
507: if (!defined($Watcher)) {
508: &child_exit(-1,'No watcher for event in ClientWritable');
509: }
510: my $Data = $Watcher->data;
511: my $Socket = $Watcher->fd;
512:
513: # Try to send the data:
514:
515: &Debug(6, "ClientWritable writing".$Data);
516: &Debug(9, "Socket is: ".$Socket);
517:
518: if($Socket->connected) {
519: my $result = $Socket->send($Data, 0);
520:
521: # $result undefined: the write failed.
522: # otherwise $result is the number of bytes written.
523: # Remove that preceding string from the data.
524: # If the resulting data is empty, destroy the watcher
525: # and set up a read event handler to accept the next
526: # request.
527:
528: &Debug(9,"Send result is ".$result." Defined: ".defined($result));
529: if($result ne undef) {
530: &Debug(9, "send result was defined");
531: if($result == length($Data)) { # Entire string sent.
532: &Debug(9, "ClientWritable data all written");
533: $Watcher->cancel();
534: #
535: # Set up to read next request from socket:
536:
537: my $descr = sprintf("Connection to lonc client %d",
538: $ActiveClients{$Socket});
539: Event->io(cb => \&ClientRequest,
540: poll => 'r',
541: desc => $descr,
542: data => "",
543: fd => $Socket);
544:
545: } else { # Partial string sent.
546: $Watcher->data(substr($Data, $result));
547: if($result == 0) { # client hung up on us!!
548: # Log("INFO", "lonc pipe client hung up on us!");
549: $Watcher->cancel;
550: $Socket->shutdown(2);
551: $Socket->close();
552: }
553: }
554:
555: } else { # Error of some sort...
556:
557: # Some errnos are possible:
558: my $errno = $!;
559: if($errno == POSIX::EWOULDBLOCK ||
560: $errno == POSIX::EAGAIN ||
561: $errno == POSIX::EINTR) {
562: # No action taken...the socket will be writable firing the event again
563: # which will result in a retry of the write.
564: } else { # Unanticipated errno.
565: &Debug(5,"ClientWritable error or peer shutdown".$RemoteHost);
566: $Watcher->cancel; # Stop the watcher.
567: $Socket->shutdown(2); # Kill connection
568: $Socket->close(); # Close the socket.
569: }
570:
571: }
572: } else {
573: $Watcher->cancel(); # A delayed request...just cancel.
574: return;
575: }
576: }
577:
578: =pod
579:
580: =head2 CompleteTransaction
581:
582: Called when the reply data has been received for a lond
583: transaction. The reply data must now be sent to the
584: ultimate client on the other end of the Unix socket. This is
585: done by setting up a writable event for the socket with the
586: data the reply data.
587:
588: Parameters:
589:
590: =item Socket
591:
592: Socket on which the lond transaction occured. This is a
593: LondConnection. The data received is in the TransactionReply member.
594:
595: =item Transaction
596:
597: The transaction that is being completed.
598:
599: =cut
600:
601: sub CompleteTransaction {
602: &Debug(5,"Complete transaction");
603:
604: my ($Socket, $Transaction) = @_;
605:
606: if (!$Transaction->isDeferred()) { # Normal transaction
607: my $data = $Socket->GetReply(); # Data to send.
608: if($LogTransactions) {
609: Log("SUCCESS", "Reply from lond: '$data'");
610: }
611: StartClientReply($Transaction, $data);
612: } else { # Delete deferred transaction file.
613: Log("SUCCESS", "A delayed transaction was completed");
614: LogPerm("S:".$Transaction->getClient().":".$Transaction->getRequest());
615: unlink($Transaction->getFile());
616: }
617: }
618:
619: =pod
620:
621: =head1 StartClientReply
622:
623: Initiates a reply to a client where the reply data is a parameter.
624:
625: =head2 parameters:
626:
627: =item Transaction
628:
629: The transaction for which we are responding to the client.
630:
631: =item data
632:
633: The data to send to apached client.
634:
635: =cut
636:
637: sub StartClientReply {
638:
639: my ($Transaction, $data) = @_;
640:
641: my $Client = $Transaction->getClient();
642:
643: &Debug(8," Reply was: ".$data);
644: my $Serial = $ActiveClients{$Client};
645: my $desc = sprintf("Connection to lonc client %d",
646: $Serial);
647: Event->io(fd => $Client,
648: poll => "w",
649: desc => $desc,
650: cb => \&ClientWritable,
651: data => $data);
652: }
653:
654: =pod
655:
656: =head2 FailTransaction
657:
658: Finishes a transaction with failure because the associated lond socket
659: disconnected. There are two possibilities:
660: - The transaction is deferred: in which case we just quietly
661: delete the transaction since there is no client connection.
662: - The transaction is 'live' in which case we initiate the sending
663: of "con_lost" to the client.
664:
665: Deleting the transaction means killing it from the %ActiveTransactions hash.
666:
667: Parameters:
668:
669: =item client
670:
671: The LondTransaction we are failing.
672:
673:
674: =cut
675:
676: sub FailTransaction {
677: my $transaction = shift;
678:
679: # If the socket is dead, that's already logged.
680:
681: if ($ConnectionRetriesLeft > 0) {
682: Log("WARNING", "Failing transaction "
683: .$transaction->getLoggableRequest());
684: }
685: Debug(1, "Failing transaction: ".$transaction->getLoggableRequest());
686: if (!$transaction->isDeferred()) { # If the transaction is deferred we'll get to it.
687: my $client = $transaction->getClient();
688: Debug(1," Replying con_lost to ".$transaction->getRequest());
689: StartClientReply($transaction, "con_lost\n");
690: }
691:
692: }
693:
694: =pod
695:
696: =head1 EmptyQueue
697:
698: Fails all items in the work queue with con_lost.
699: Note that each item in the work queue is a transaction.
700:
701: =cut
702:
703: sub EmptyQueue {
704: $ConnectionRetriesLeft--; # Counts as connection failure too.
705: while($WorkQueue->Count()) {
706: my $request = $WorkQueue->dequeue(); # This is a transaction
707: FailTransaction($request);
708: }
709: }
710:
711: =pod
712:
713: =head2 CloseAllLondConnections
714:
715: Close all connections open on lond prior to exit e.g.
716:
717: =cut
718:
719: sub CloseAllLondConnections {
720: foreach my $Socket (keys %ActiveConnections) {
721: if(exists($ActiveTransactions{$Socket})) {
722: FailTransaction($ActiveTransactions{$Socket});
723: }
724: KillSocket($Socket);
725: }
726: }
727:
728: =pod
729:
730: =head2 KillSocket
731:
732: Destroys a socket. This function can be called either when a socket
733: has died of 'natural' causes or because a socket needs to be pruned due to
734: idleness. If the socket has died naturally, if there are no longer any
735: live connections a new connection is created (in case there are transactions
736: in the queue). If the socket has been pruned, it is never re-created.
737:
738: Parameters:
739:
740: =item Socket
741:
742: The socket to kill off.
743:
744: =item Restart
745:
746: nonzero if we are allowed to create a new connection.
747:
748: =cut
749:
750: sub KillSocket {
751: my $Socket = shift;
752:
753: Log("WARNING", "Shutting down a socket");
754: $Socket->Shutdown();
755:
756: # If the socket came from the active connection set,
757: # delete its transaction... note that FailTransaction should
758: # already have been called!!!
759: # otherwise it came from the idle set.
760: #
761:
762: if(exists($ActiveTransactions{$Socket})) {
763: delete ($ActiveTransactions{$Socket});
764: }
765: if(exists($ActiveConnections{$Socket})) {
766: $ActiveConnections{$Socket}->cancel;
767: delete($ActiveConnections{$Socket});
768: $ConnectionCount--;
769: if ($ConnectionCount < 0) { $ConnectionCount = 0; }
770: }
771: # If the connection count has gone to zero and there is work in the
772: # work queue, the work all gets failed with con_lost.
773: #
774: if($ConnectionCount == 0) {
775: EmptyQueue();
776: CloseAllLondConnections; # Should all already be closed but...
777: }
778: UpdateStatus();
779: }
780:
781: =pod
782:
783: =head2 LondReadable
784:
785: This function is called whenever a lond connection
786: is readable. The action is state dependent:
787:
788: =head3 State=Initialized
789:
790: We''re waiting for the challenge, this is a no-op until the
791: state changes.
792:
793: =head3 State=Challenged
794:
795: The challenge has arrived we need to transition to Writable.
796: The connection must echo the challenge back.
797:
798: =head3 State=ChallengeReplied
799:
800: The challenge has been replied to. The we are receiveing the
801: 'ok' from the partner.
802:
803: =head3 State=ReadingVersionString
804:
805: We have requested the lond version and are reading the
806: version back. Upon completion, we'll store the version away
807: for future use(?).
808:
809: =head3 State=HostSet
810:
811: We have selected the domain name of our peer (multhomed hosts)
812: and are getting the reply (presumably ok) back.
813:
814: =head3 State=RequestingKey
815:
816: The ok has been received and we need to send the request for
817: an encryption key. Transition to writable for that.
818:
819: =head3 State=ReceivingKey
820:
821: The the key has been requested, now we are reading the new key.
822:
823: =head3 State=Idle
824:
825: The encryption key has been negotiated or we have finished
826: reading data from the a transaction. If the callback data has
827: a client as well as the socket iformation, then we are
828: doing a transaction and the data received is relayed to the client
829: before the socket is put on the idle list.
830:
831: =head3 State=SendingRequest
832:
833: I do not think this state can be received here, but if it is,
834: the appropriate thing to do is to transition to writable, and send
835: the request.
836:
837: =head3 State=ReceivingReply
838:
839: We finished sending the request to the server and now transition
840: to readable to receive the reply.
841:
842: The parameter to this function are:
843:
844: The event. Implicit in this is the watcher and its data. The data
845: contains at least the lond connection object and, if a
846: transaction is in progress, the socket attached to the local client.
847:
848: =cut
849:
850: sub LondReadable {
851:
852: my $Event = shift;
853: my $Watcher = $Event->w;
854: my $Socket = $Watcher->data;
855: my $client = undef;
856:
857: &Debug(6,"LondReadable called state = ".$Socket->GetState());
858:
859:
860: my $State = $Socket->GetState(); # All action depends on the state.
861:
862: SocketDump(6, $Socket);
863: my $status = $Socket->Readable();
864:
865: &Debug(2, "Socket->Readable returned: $status");
866:
867: if($status != 0) {
868: # bad return from socket read. Currently this means that
869: # The socket has become disconnected. We fail the transaction.
870:
871: Log("WARNING",
872: "Lond connection lost.");
873: if(exists($ActiveTransactions{$Socket})) {
874: FailTransaction($ActiveTransactions{$Socket});
875: } else {
876: # Socket is connecting and failed... need to mark
877: # no longer connecting.
878:
879: $LondConnecting = 0;
880: }
881: $Watcher->cancel();
882: KillSocket($Socket);
883: $ConnectionRetriesLeft--; # Counts as connection failure
884: return;
885: }
886: SocketDump(6,$Socket);
887:
888: $State = $Socket->GetState(); # Update in case of transition.
889: &Debug(6, "After read, state is ".$State);
890:
891: if($State eq "Initialized") {
892:
893:
894: } elsif ($State eq "ChallengeReceived") {
895: # The challenge must be echoed back; The state machine
896: # in the connection takes care of setting that up. Just
897: # need to transition to writable:
898:
899: $Watcher->cb(\&LondWritable);
900: $Watcher->poll("w");
901:
902: } elsif ($State eq "ChallengeReplied") {
903:
904: } elsif ($State eq "RequestingVersion") {
905: # Need to ask for the version... that is writiability:
906:
907: $Watcher->cb(\&LondWritable);
908: $Watcher->poll("w");
909:
910: } elsif ($State eq "ReadingVersionString") {
911: # Read the rest of the version string...
912: } elsif ($State eq "SetHost") {
913: # Need to request the actual domain get set...
914:
915: $Watcher->cb(\&LondWritable);
916: $Watcher->poll("w");
917: } elsif ($State eq "HostSet") {
918: # Reading the 'ok' from the peer.
919:
920: } elsif ($State eq "RequestingKey") {
921: # The ok was received. Now we need to request the key
922: # That requires us to be writable:
923:
924: $Watcher->cb(\&LondWritable);
925: $Watcher->poll("w");
926:
927: } elsif ($State eq "ReceivingKey") {
928:
929: } elsif ($State eq "Idle") {
930:
931: # This is as good a spot as any to get the peer version
932: # string:
933:
934: if($LondVersion eq "unknown") {
935: $LondVersion = $Socket->PeerVersion();
936: Log("INFO", "Connected to lond version: $LondVersion");
937: }
938: # If necessary, complete a transaction and then go into the
939: # idle queue.
940: # Note that a trasition to idle indicates a live lond
941: # on the other end so reset the connection retries.
942: #
943: $ConnectionRetriesLeft = $ConnectionRetries; # success resets the count
944: $Watcher->cancel();
945: if(exists($ActiveTransactions{$Socket})) {
946: Debug(5,"Completing transaction!!");
947: CompleteTransaction($Socket,
948: $ActiveTransactions{$Socket});
949: } else {
950: my $count = $Socket->GetClientData();
951: Log("SUCCESS", "Connection ".$count." to "
952: .$RemoteHost." now ready for action");
953: }
954: ServerToIdle($Socket); # Next work unit or idle.
955:
956: #
957: $LondConnecting = 0; # Best spot I can think of for this.
958: #
959:
960: } elsif ($State eq "SendingRequest") {
961: # We need to be writable for this and probably don't belong
962: # here inthe first place.
963:
964: Debug(6, "SendingRequest state encountered in readable");
965: $Watcher->poll("w");
966: $Watcher->cb(\&LondWritable);
967:
968: } elsif ($State eq "ReceivingReply") {
969:
970:
971: } else {
972: # Invalid state.
973: Debug(4, "Invalid state in LondReadable");
974: }
975: }
976:
977: =pod
978:
979: =head2 LondWritable
980:
981: This function is called whenever a lond connection
982: becomes writable while there is a writeable monitoring
983: event. The action taken is very state dependent:
984:
985: =head3 State = Connected
986:
987: The connection is in the process of sending the 'init' hailing to the
988: lond on the remote end. The connection object''s Writable member is
989: called. On error, ConnectionError is called to destroy the connection
990: and remove it from the ActiveConnections hash
991:
992: =head3 Initialized
993:
994: 'init' has been sent, writability monitoring is removed and
995: readability monitoring is started with LondReadable as the callback.
996:
997: =head3 ChallengeReceived
998:
999: The connection has received the who are you challenge from the remote
1000: system, and is in the process of sending the challenge
1001: response. Writable is called.
1002:
1003: =head3 ChallengeReplied
1004:
1005: The connection has replied to the initial challenge The we switch to
1006: monitoring readability looking for the server to reply with 'ok'.
1007:
1008: =head3 RequestingKey
1009:
1010: The connection is in the process of requesting its encryption key.
1011: Writable is called.
1012:
1013: =head3 ReceivingKey
1014:
1015: The connection has sent the request for a key. Switch to readability
1016: monitoring to accept the key
1017:
1018: =head3 SendingRequest
1019:
1020: The connection is in the process of sending a request to the server.
1021: This request is part of a client transaction. All the states until
1022: now represent the client setup protocol. Writable is called.
1023:
1024: =head3 ReceivingReply
1025:
1026: The connection has sent a request. Now it must receive a reply.
1027: Readability monitoring is requested.
1028:
1029: This function is an event handler and therefore receives as
1030: a parameter the event that has fired. The data for the watcher
1031: of this event is a reference to a list of one or two elements,
1032: depending on state. The first (and possibly only) element is the
1033: socket. The second (present only if a request is in progress)
1034: is the socket on which to return a reply to the caller.
1035:
1036: =cut
1037:
1038: sub LondWritable {
1039: my $Event = shift;
1040: my $Watcher = $Event->w;
1041: my $Socket = $Watcher->data;
1042: my $State = $Socket->GetState();
1043:
1044: Debug(6,"LondWritable State = ".$State."\n");
1045:
1046:
1047: # Figure out what to do depending on the state of the socket:
1048:
1049:
1050:
1051:
1052: SocketDump(6,$Socket);
1053:
1054: # If the socket is writable, we must always write.
1055: # Only by writing will we undergo state transitions.
1056: # Old logic wrote in state specific code below, however
1057: # That forces us at least through another invocation of
1058: # this function after writability is possible again.
1059: # This logic also factors out common code for handling
1060: # write failures... in all cases, write failures
1061: # Kill the socket.
1062: # This logic makes the branches of the >big< if below
1063: # so that the writing states are actually NO-OPs.
1064:
1065: if ($Socket->Writable() != 0) {
1066: # The write resulted in an error.
1067: # We'll treat this as if the socket got disconnected:
1068: Log("WARNING", "Connection to ".$RemoteHost.
1069: " has been disconnected");
1070: if(exists($ActiveTransactions{$Socket})) {
1071: FailTransaction($ActiveTransactions{$Socket});
1072: } else {
1073: # In the process of conneting, so need to turn that off.
1074:
1075: $LondConnecting = 0;
1076: }
1077: $Watcher->cancel();
1078: KillSocket($Socket);
1079: return;
1080: }
1081:
1082:
1083:
1084: if ($State eq "Connected") {
1085:
1086: # "init" is being sent...
1087:
1088: } elsif ($State eq "Initialized") {
1089:
1090: # Now that init was sent, we switch
1091: # to watching for readability:
1092:
1093: $Watcher->cb(\&LondReadable);
1094: $Watcher->poll("r");
1095:
1096: } elsif ($State eq "ChallengeReceived") {
1097: # We received the challenge, now we
1098: # are echoing it back. This is a no-op,
1099: # we're waiting for the state to change
1100:
1101: } elsif ($State eq "ChallengeReplied") {
1102: # The echo was sent back, so we switch
1103: # to watching readability.
1104:
1105: $Watcher->cb(\&LondReadable);
1106: $Watcher->poll("r");
1107: } elsif ($State eq "RequestingVersion") {
1108: # Sending the peer a version request...
1109:
1110: } elsif ($State eq "ReadingVersionString") {
1111: # Transition to read since we have sent the
1112: # version command and now just need to read the
1113: # version string from the peer:
1114:
1115: $Watcher->cb(\&LondReadable);
1116: $Watcher->poll("r");
1117:
1118: } elsif ($State eq "SetHost") {
1119: # Setting the remote domain...
1120:
1121: } elsif ($State eq "HostSet") {
1122: # Back to readable to get the ok.
1123:
1124: $Watcher->cb(\&LondReadable);
1125: $Watcher->poll("r");
1126:
1127:
1128: } elsif ($State eq "RequestingKey") {
1129: # At this time we're requesting the key.
1130: # again, this is essentially a no-op.
1131:
1132: } elsif ($State eq "ReceivingKey") {
1133: # Now we need to wait for the key
1134: # to come back from the peer:
1135:
1136: $Watcher->cb(\&LondReadable);
1137: $Watcher->poll("r");
1138:
1139: } elsif ($State eq "SendingRequest") {
1140:
1141: # At this time we are sending a request to the
1142: # peer... write the next chunk:
1143:
1144:
1145: } elsif ($State eq "ReceivingReply") {
1146: # The send has completed. Wait for the
1147: # data to come in for a reply.
1148: Debug(8,"Writable sent request/receiving reply");
1149: $Watcher->cb(\&LondReadable);
1150: $Watcher->poll("r");
1151:
1152: } else {
1153: # Control only passes here on an error:
1154: # the socket state does not match any
1155: # of the known states... so an error
1156: # must be logged.
1157:
1158: &Debug(4, "Invalid socket state ".$State."\n");
1159: }
1160:
1161: }
1162:
1163: =pod
1164:
1165: =cut
1166:
1167:
1168: sub QueueDelayed {
1169: Debug(3,"QueueDelayed called");
1170:
1171: my $path = "$perlvar{'lonSockDir'}/delayed";
1172:
1173: Debug(4, "Delayed path: ".$path);
1174: opendir(DIRHANDLE, $path);
1175:
1176: my $host_id_re = '(?:'.join('|',map {quotemeta($_)} (@all_host_ids)).')';
1177: my @alldelayed = grep(/\.$host_id_re$/, readdir(DIRHANDLE));
1178: closedir(DIRHANDLE);
1179: foreach my $dfname (sort(@alldelayed)) {
1180: my $reqfile = "$path/$dfname";
1181: my ($host_id) = ($dfname =~ /\.([^.]*)$/);
1182: Debug(4, "queueing ".$reqfile." for $host_id");
1183: my $Handle = IO::File->new($reqfile);
1184: my $cmd = <$Handle>;
1185: chomp $cmd; # There may or may not be a newline...
1186: $cmd = $cmd."\n"; # now for sure there's exactly one newline.
1187: my $Transaction = LondTransaction->new("sethost:$host_id:$cmd");
1188: $Transaction->SetDeferred($reqfile);
1189: QueueTransaction($Transaction);
1190: }
1191:
1192: }
1193:
1194: =pod
1195:
1196: =head2 MakeLondConnection
1197:
1198: Create a new lond connection object, and start it towards its initial
1199: idleness. Once idle, it becomes elligible to receive transactions
1200: from the work queue. If the work queue is not empty when the
1201: connection is completed and becomes idle, it will dequeue an entry and
1202: start off on it.
1203:
1204: =cut
1205:
1206: sub MakeLondConnection {
1207: Debug(4,"MakeLondConnection to ".GetServerHost()." on port "
1208: .GetServerPort());
1209:
1210: my $Connection = LondConnection->new(&GetServerHost(),
1211: &GetServerPort(),
1212: &GetHostId());
1213:
1214: if($Connection eq undef) {
1215: Log("CRITICAL","Failed to make a connection with lond.");
1216: $ConnectionRetriesLeft--;
1217: return 0; # Failure.
1218: } else {
1219:
1220: $LondConnecting = 1; # Connection in progress.
1221: # The connection needs to have writability
1222: # monitored in order to send the init sequence
1223: # that starts the whole authentication/key
1224: # exchange underway.
1225: #
1226: my $Socket = $Connection->GetSocket();
1227: if($Socket eq undef) {
1228: &child_exit(-1, "did not get a socket from the connection");
1229: } else {
1230: &Debug(9,"MakeLondConnection got socket: ".$Socket);
1231: }
1232:
1233: $Connection->SetTimeoutCallback(\&SocketTimeout);
1234:
1235: my $event = Event->io(fd => $Socket,
1236: poll => 'w',
1237: cb => \&LondWritable,
1238: data => $Connection,
1239: desc => 'Connection to lond server');
1240: $ActiveConnections{$Connection} = $event;
1241: if ($ConnectionCount == 0) {
1242: &SetupTimer; # Need to handle timeouts with connections...
1243: }
1244: $ConnectionCount++;
1245: $Connection->SetClientData($ConnectionCount);
1246: Debug(4, "Connection count = ".$ConnectionCount);
1247: if($ConnectionCount == 1) { # First Connection:
1248: QueueDelayed;
1249: }
1250: Log("SUCESS", "Created connection ".$ConnectionCount
1251: ." to host ".GetServerHost());
1252: return 1; # Return success.
1253: }
1254:
1255: }
1256:
1257: =pod
1258:
1259: =head2 StartRequest
1260:
1261: Starts a lond request going on a specified lond connection.
1262: parameters are:
1263:
1264: =item $Lond
1265:
1266: Connection to the lond that will send the transaction and receive the
1267: reply.
1268:
1269: =item $Client
1270:
1271: Connection to the client that is making this request We got the
1272: request from this socket, and when the request has been relayed to
1273: lond and we get a reply back from lond it will get sent to this
1274: socket.
1275:
1276: =item $Request
1277:
1278: The text of the request to send.
1279:
1280: =cut
1281:
1282: sub StartRequest {
1283:
1284: my ($Lond, $Request) = @_;
1285:
1286: Debug(6, "StartRequest: ".$Request->getRequest());
1287:
1288: my $Socket = $Lond->GetSocket();
1289:
1290: $Request->Activate($Lond);
1291: $ActiveTransactions{$Lond} = $Request;
1292:
1293: $Lond->InitiateTransaction($Request->getRequest());
1294: my $event = Event->io(fd => $Socket,
1295: poll => "w",
1296: cb => \&LondWritable,
1297: data => $Lond,
1298: desc => "lond transaction connection");
1299: $ActiveConnections{$Lond} = $event;
1300: Debug(8," Start Request made watcher data with ".$event->data."\n");
1301: }
1302:
1303: =pod
1304:
1305: =head2 QueueTransaction
1306:
1307: If there is an idle lond connection, it is put to work doing this
1308: transaction. Otherwise, the transaction is placed in the work queue.
1309: If placed in the work queue and the maximum number of connections has
1310: not yet been created, a new connection will be started. Our goal is
1311: to eventually have a sufficient number of connections that the work
1312: queue will typically be empty. parameters are:
1313:
1314: =item Socket
1315:
1316: open on the lonc client.
1317:
1318: =item Request
1319:
1320: data to send to the lond.
1321:
1322: =cut
1323:
1324: sub QueueTransaction {
1325:
1326: my $requestData = shift; # This is a LondTransaction.
1327: my $cmd = $requestData->getRequest();
1328:
1329: Debug(6,"QueueTransaction: ".$cmd);
1330:
1331: my $LondSocket = $IdleConnections->pop();
1332: if(!defined $LondSocket) { # Need to queue request.
1333: Debug(5,"Must queue...");
1334: $WorkQueue->enqueue($requestData);
1335: Debug(5, "Queue Transaction startnew $ConnectionCount $LondConnecting");
1336: if(($ConnectionCount < $MaxConnectionCount) && (! $LondConnecting)) {
1337:
1338: if($ConnectionRetriesLeft > 0) {
1339: Debug(5,"Starting additional lond connection");
1340: if(&MakeLondConnection() == 0) {
1341: EmptyQueue(); # Fail transactions, can't make connection.
1342: CloseAllLondConnections; # Should all be closed but...
1343: }
1344: } else {
1345: ShowStatus(GetServerHost()." >>> DEAD !!!! <<<");
1346: $LondConnecting = 0;
1347: EmptyQueue(); # It's worse than that ... he's dead Jim.
1348: CloseAllLondConnections; # Should all be closed but..
1349: }
1350: }
1351: } else { # Can start the request:
1352: Debug(8,"Can start...");
1353: StartRequest($LondSocket, $requestData);
1354: }
1355: }
1356:
1357: #-------------------------- Lonc UNIX socket handling -------------------
1358: =pod
1359:
1360: =head2 ClientRequest
1361: Callback that is called when data can be read from the UNIX domain
1362: socket connecting us with an apache server process.
1363:
1364: =cut
1365:
1366: sub ClientRequest {
1367: Debug(6, "ClientRequest");
1368: my $event = shift;
1369: my $watcher = $event->w;
1370: my $socket = $watcher->fd;
1371: my $data = $watcher->data;
1372: my $thisread;
1373:
1374: Debug(9, " Watcher named: ".$watcher->desc);
1375:
1376: my $rv = $socket->recv($thisread, POSIX::BUFSIZ, 0);
1377: Debug(8, "rcv: data length = ".length($thisread)
1378: ." read =".$thisread);
1379: unless (defined $rv && length($thisread)) {
1380: # Likely eof on socket.
1381: Debug(5,"Client Socket closed on lonc for ".$RemoteHost);
1382: close($socket);
1383: $watcher->cancel();
1384: delete($ActiveClients{$socket});
1385: return;
1386: }
1387: Debug(8,"Data: ".$data." this read: ".$thisread);
1388: $data = $data.$thisread; # Append new data.
1389: $watcher->data($data);
1390: if($data =~ /\n$/) { # Request entirely read.
1391: if ($data eq "close_connection_exit\n") {
1392: Log("CRITICAL",
1393: "Request Close Connection ... exiting");
1394: CloseAllLondConnections();
1395: exit;
1396: } elsif ($data eq "reset_retries\n") {
1397: Log("INFO", "Resetting Connection Retries.");
1398: $ConnectionRetriesLeft = $ConnectionRetries;
1399: &UpdateStatus();
1400: my $Transaction = LondTransaction->new($data);
1401: $Transaction->SetClient($socket);
1402: StartClientReply($Transaction, "ok\n");
1403: $watcher->cancel();
1404: return;
1405: }
1406: Debug(8, "Complete transaction received: ".$data);
1407: if ($LogTransactions) {
1408: Log("SUCCESS", "Transaction: '$data'"); # Transaction has \n.
1409: }
1410: my $Transaction = LondTransaction->new($data);
1411: $Transaction->SetClient($socket);
1412: QueueTransaction($Transaction);
1413: $watcher->cancel(); # Done looking for input data.
1414: }
1415:
1416: }
1417:
1418: #
1419: # Accept a connection request for a client (lonc child) and
1420: # start up an event watcher to keep an eye on input from that
1421: # Event. This can be called both from NewClient and from
1422: # ChildProcess.
1423: # Parameters:
1424: # $socket - The listener socket.
1425: # Returns:
1426: # NONE
1427: # Side Effects:
1428: # An event is made to watch the accepted connection.
1429: # Active clients hash is updated to reflect the new connection.
1430: # The client connection count is incremented.
1431: #
1432: sub accept_client {
1433: my ($socket) = @_;
1434:
1435: Debug(8, "Entering accept for lonc UNIX socket\n");
1436: my $connection = $socket->accept(); # Accept the client connection.
1437: Debug(8,"Connection request accepted from "
1438: .GetPeername($connection, AF_UNIX));
1439:
1440:
1441: my $description = sprintf("Connection to lonc client %d",
1442: $ClientConnection);
1443: Debug(9, "Creating event named: ".$description);
1444: Event->io(cb => \&ClientRequest,
1445: poll => 'r',
1446: desc => $description,
1447: data => "",
1448: fd => $connection);
1449: $ActiveClients{$connection} = $ClientConnection;
1450: $ClientConnection++;
1451: }
1452:
1453: =pod
1454:
1455: =head2 NewClient
1456:
1457: Callback that is called when a connection is received on the unix
1458: socket for a new client of lonc. The callback is parameterized by the
1459: event.. which is a-priori assumed to be an io event, and therefore has
1460: an fd member that is the Listener socket. We Accept the connection
1461: and register a new event on the readability of that socket:
1462:
1463: =cut
1464:
1465: sub NewClient {
1466: Debug(6, "NewClient");
1467: my $event = shift; # Get the event parameters.
1468: my $watcher = $event->w;
1469: my $socket = $watcher->fd; # Get the event' socket.
1470:
1471: &accept_client($socket);
1472: }
1473:
1474: =pod
1475:
1476: =head2 GetLoncSocketPath
1477:
1478: Returns the name of the UNIX socket on which to listen for client
1479: connections.
1480:
1481: =head2 Parameters:
1482:
1483: host (optional) - Name of the host socket to return.. defaults to
1484: the return from GetServerHost().
1485:
1486: =cut
1487:
1488: sub GetLoncSocketPath {
1489:
1490: my $host = GetServerHost(); # Default host.
1491: if (@_) {
1492: ($host) = @_; # Override if supplied.
1493: }
1494: return $UnixSocketDir."/".$host;
1495: }
1496:
1497: =pod
1498:
1499: =head2 GetServerHost
1500:
1501: Returns the host whose lond we talk with.
1502:
1503: =cut
1504:
1505: sub GetServerHost {
1506: return $RemoteHost; # Setup by the fork.
1507: }
1508:
1509: =pod
1510:
1511: =head2 GetServerId
1512:
1513: Returns the hostid whose lond we talk with.
1514:
1515: =cut
1516:
1517: sub GetHostId {
1518: return $RemoteHostId; # Setup by the fork.
1519: }
1520:
1521: =pod
1522:
1523: =head2 GetServerPort
1524:
1525: Returns the lond port number.
1526:
1527: =cut
1528:
1529: sub GetServerPort {
1530: return $perlvar{londPort};
1531: }
1532:
1533: =pod
1534:
1535: =head2 SetupLoncListener
1536:
1537: Setup a lonc listener event. The event is called when the socket
1538: becomes readable.. that corresponds to the receipt of a new
1539: connection. The event handler established will accept the connection
1540: (creating a communcations channel), that int turn will establish
1541: another event handler to subess requests.
1542:
1543: =head2 Parameters:
1544:
1545: host (optional) Name of the host to set up a unix socket to.
1546:
1547: =cut
1548:
1549: sub SetupLoncListener {
1550: my ($host,$SocketName) = @_;
1551: if (!$host) { $host = &GetServerHost(); }
1552: if (!$SocketName) { $SocketName = &GetLoncSocketPath($host); }
1553:
1554:
1555: unlink($SocketName);
1556:
1557: my $socket;
1558: unless ($socket =IO::Socket::UNIX->new(Local => $SocketName,
1559: Listen => 250,
1560: Type => SOCK_STREAM)) {
1561: if($I_am_child) {
1562: &child_exit(-1, "Failed to create a lonc listener socket");
1563: } else {
1564: die "Failed to create a lonc listner socket";
1565: }
1566: }
1567: return $socket;
1568: }
1569:
1570: #
1571: # Toggle transaction logging.
1572: # Implicit inputs:
1573: # LogTransactions
1574: # Implicit Outputs:
1575: # LogTransactions
1576: sub ToggleTransactionLogging {
1577: print STDERR "Toggle transaction logging...\n";
1578: if(!$LogTransactions) {
1579: $LogTransactions = 1;
1580: } else {
1581: $LogTransactions = 0;
1582: }
1583:
1584:
1585: Log("SUCCESS", "Toggled transaction logging: $LogTransactions \n");
1586: }
1587:
1588: =pod
1589:
1590: =head2 ChildStatus
1591:
1592: Child USR1 signal handler to report the most recent status
1593: into the status file.
1594:
1595: We also use this to reset the retries count in order to allow the
1596: client to retry connections with a previously dead server.
1597:
1598: =cut
1599:
1600: sub ChildStatus {
1601: my $event = shift;
1602: my $watcher = $event->w;
1603:
1604: Debug(2, "Reporting child status because : ".$watcher->data);
1605: my $docdir = $perlvar{'lonDocRoot'};
1606:
1607: open(LOG,">>$docdir/lon-status/loncstatus.txt");
1608: flock(LOG,LOCK_EX);
1609: print LOG $$."\t".$RemoteHost."\t".$Status."\t".
1610: $RecentLogEntry."\n";
1611: #
1612: # Write out information about each of the connections:
1613: #
1614: if ($DebugLevel > 2) {
1615: print LOG "Active connection statuses: \n";
1616: my $i = 1;
1617: print STDERR "================================= Socket Status Dump:\n";
1618: foreach my $item (keys %ActiveConnections) {
1619: my $Socket = $ActiveConnections{$item}->data;
1620: my $state = $Socket->GetState();
1621: print LOG "Connection $i State: $state\n";
1622: print STDERR "---------------------- Connection $i \n";
1623: $Socket->Dump(-1); # Ensure it gets dumped..
1624: $i++;
1625: }
1626: }
1627: flock(LOG,LOCK_UN);
1628: close(LOG);
1629: $ConnectionRetriesLeft = $ConnectionRetries;
1630: UpdateStatus();
1631: }
1632:
1633: =pod
1634:
1635: =head2 SignalledToDeath
1636:
1637: Called in response to a signal that causes a chid process to die.
1638:
1639: =cut
1640:
1641:
1642: sub SignalledToDeath {
1643: my $event = shift;
1644: my $watcher= $event->w;
1645:
1646: Debug(2,"Signalled to death! via ".$watcher->data);
1647: my ($signal) = $watcher->data;
1648: chomp($signal);
1649: Log("CRITICAL", "Abnormal exit. Child $$ for $RemoteHost "
1650: ."died through "."\"$signal\"");
1651: #LogPerm("F:lonc: $$ on $RemoteHost signalled to death: "
1652: # ."\"$signal\"");
1653: exit 0;
1654:
1655: }
1656:
1657: =pod
1658:
1659: =head2 ToggleDebug
1660:
1661: This sub toggles trace debugging on and off.
1662:
1663: =cut
1664:
1665: sub ToggleDebug {
1666: my $Current = $DebugLevel;
1667: $DebugLevel = $NextDebugLevel;
1668: $NextDebugLevel = $Current;
1669:
1670: Log("SUCCESS", "New debugging level for $RemoteHost now $DebugLevel");
1671:
1672: }
1673:
1674: =pod
1675:
1676: =head2 ChildProcess
1677:
1678: This sub implements a child process for a single lonc daemon.
1679: Optional parameter:
1680: $socket - if provided, this is a socket already open for listen
1681: on the client socket. Otherwise, a new listen is set up.
1682:
1683: =cut
1684:
1685: sub ChildProcess {
1686: # We've inherited all the
1687: # events of our parent and those have to be cancelled or else
1688: # all holy bloody chaos will result.. trust me, I already made
1689: # >that< mistake.
1690:
1691: my $host = GetServerHost();
1692: foreach my $listener (keys %parent_dispatchers) {
1693: my $watcher = $parent_dispatchers{$listener};
1694: my $s = $watcher->fd;
1695: if ($listener ne $host) { # Close everyone but me.
1696: Debug(5, "Closing listen socket for $listener");
1697: $s->close();
1698: }
1699: Debug(5, "Killing watcher for $listener");
1700:
1701: $watcher->cancel();
1702: delete($parent_dispatchers{$listener});
1703:
1704: }
1705:
1706: # kill off the parent's signal handlers too!
1707: #
1708:
1709: for my $handler (keys %parent_handlers) {
1710: my $watcher = $parent_handlers{$handler};
1711: $watcher->cancel();
1712: delete($parent_handlers{$handler});
1713: }
1714:
1715: $I_am_child = 1; # Seems like in spite of it all I may still getting
1716: # parent event dispatches.. flag I'm a child.
1717:
1718:
1719: #
1720: # Signals must be handled by the Event framework...
1721: #
1722:
1723: Event->signal(signal => "QUIT",
1724: cb => \&SignalledToDeath,
1725: data => "QUIT");
1726: Event->signal(signal => "HUP",
1727: cb => \&ChildStatus,
1728: data => "HUP");
1729: Event->signal(signal => "USR1",
1730: cb => \&ChildStatus,
1731: data => "USR1");
1732: Event->signal(signal => "USR2",
1733: cb => \&ToggleTransactionLogging);
1734: Event->signal(signal => "INT",
1735: cb => \&ToggleDebug,
1736: data => "INT");
1737:
1738: # Block the pipe signal we'll get when the socket disconnects. We detect
1739: # socket disconnection via send/receive failures. On disconnect, the
1740: # socket becomes readable .. which will force the disconnect detection.
1741:
1742: my $set = POSIX::SigSet->new(SIGPIPE);
1743: sigprocmask(SIG_BLOCK, $set);
1744:
1745: # Figure out if we got passed a socket or need to open one to listen for
1746: # client requests.
1747:
1748: my ($socket) = @_;
1749: if (!$socket) {
1750:
1751: $socket = SetupLoncListener();
1752: }
1753: # Establish an event to listen for client connection requests.
1754:
1755:
1756: Event->io(cb => \&NewClient,
1757: poll => 'r',
1758: desc => 'Lonc Listener Unix Socket',
1759: fd => $socket);
1760:
1761: $Event::DebugLevel = $DebugLevel;
1762:
1763: Debug(9, "Making initial lond connection for ".$RemoteHost);
1764:
1765: # Setup the initial server connection:
1766:
1767: # &MakeLondConnection(); // let first work request do it.
1768:
1769: # need to accept the connection since the event may not fire.
1770:
1771: &accept_client($socket);
1772:
1773: Debug(9,"Entering event loop");
1774: my $ret = Event::loop(); # Start the main event loop.
1775:
1776:
1777: &child_exit (-1,"Main event loop exited!!!");
1778: }
1779:
1780: # Create a new child for host passed in:
1781:
1782: sub CreateChild {
1783: my ($host, $hostid) = @_;
1784:
1785: my $sigset = POSIX::SigSet->new(SIGINT);
1786: sigprocmask(SIG_BLOCK, $sigset);
1787: $RemoteHost = $host;
1788: ShowStatus('Parent keeping the flock'); # Update time in status message.
1789: Log("CRITICAL", "Forking server for ".$host);
1790: my $pid = fork;
1791: if($pid) { # Parent
1792: $RemoteHost = "Parent";
1793: $ChildPid{$pid} = $host;
1794: sigprocmask(SIG_UNBLOCK, $sigset);
1795: undef(@all_host_ids);
1796: } else { # child.
1797: $RemoteHostId = $hostid;
1798: ShowStatus("Connected to ".$RemoteHost);
1799: $SIG{INT} = 'DEFAULT';
1800: sigprocmask(SIG_UNBLOCK, $sigset);
1801: &ChildProcess(); # Does not return.
1802: }
1803: }
1804:
1805: # parent_client_connection:
1806: # Event handler that processes client connections for the parent process.
1807: # This sub is called when the parent is listening on a socket and
1808: # a connection request arrives. We must:
1809: # Start a child process to accept the connection request.
1810: # Kill our listen on the socket.
1811: # Parameter:
1812: # event - The event object that was created to monitor this socket.
1813: # event->w->fd is the socket.
1814: # Returns:
1815: # NONE
1816: #
1817: sub parent_client_connection {
1818: if ($I_am_child) {
1819: # Should not get here, but seem to anyway:
1820: &Debug(5," Child caught parent client connection event!!");
1821: my ($event) = @_;
1822: my $watcher = $event->w;
1823: $watcher->cancel(); # Try to kill it off again!!
1824: } else {
1825: &Debug(9, "parent_client_connection");
1826: my ($event) = @_;
1827: my $watcher = $event->w;
1828: my $socket = $watcher->fd;
1829: my $connection = $socket->accept(); # Accept the client connection.
1830: Event->io(cb => \&get_remote_hostname,
1831: poll => 'r',
1832: data => "",
1833: fd => $connection);
1834: }
1835: }
1836:
1837: sub get_remote_hostname {
1838: my ($event) = @_;
1839: my $watcher = $event->w;
1840: my $socket = $watcher->fd;
1841:
1842: my $thisread;
1843: my $rv = $socket->recv($thisread, POSIX::BUFSIZ, 0);
1844: Debug(8, "rcv: data length = ".length($thisread)." read =".$thisread);
1845: if (!defined($rv) || length($thisread) == 0) {
1846: # Likely eof on socket.
1847: Debug(5,"Client Socket closed on lonc for p_c_c");
1848: close($socket);
1849: $watcher->cancel();
1850: return;
1851: }
1852:
1853: my $data = $watcher->data().$thisread;
1854: $watcher->data($data);
1855: if($data =~ /\n$/) { # Request entirely read.
1856: chomp($data);
1857: } else {
1858: return;
1859: }
1860:
1861: &Debug(5,"Creating child for $data (parent_client_connection)");
1862: (my $hostname,my $lonid,@all_host_ids) = split(':',$data);
1863: $ChildHost{$hostname}++;
1864: if ($ChildHost{$hostname} == 1) {
1865: &CreateChild($hostname,$lonid);
1866: } else {
1867: &Log('WARNING',"Request for a second child on $hostname");
1868: }
1869: # Clean up the listen since now the child takes over until it exits.
1870: $watcher->cancel(); # Nolonger listening to this event
1871: $socket->send("done\n");
1872: $socket->close();
1873: }
1874:
1875: # parent_listen:
1876: # Opens a socket and starts a listen for the parent process on a client UNIX
1877: # domain socket.
1878: #
1879: # This involves:
1880: # Creating a socket for listen.
1881: # Removing any socket lock file
1882: # Adding an event handler for this socket becoming readable
1883: # To the parent's event dispatcher.
1884: # Parameters:
1885: # loncapa_host - LonCAPA cluster name of the host represented by the client
1886: # socket.
1887: # Returns:
1888: # NONE
1889: #
1890: sub parent_listen {
1891: my ($loncapa_host) = @_;
1892: Debug(5, "parent_listen: $loncapa_host");
1893:
1894: my ($socket,$file);
1895: if (!$loncapa_host) {
1896: $loncapa_host = 'common_parent';
1897: $file = $perlvar{'lonSockCreate'};
1898: } else {
1899: $file = &GetLoncSocketPath($loncapa_host);
1900: }
1901: $socket = &SetupLoncListener($loncapa_host,$file);
1902:
1903: $listening_to{$socket} = $loncapa_host;
1904: if (!$socket) {
1905: die "Unable to create a listen socket for $loncapa_host";
1906: }
1907:
1908: my $lock_file = $file.".lock";
1909: unlink($lock_file); # No problem if it doesn't exist yet [startup e.g.]
1910:
1911: my $watcher =
1912: Event->io(cb => \&parent_client_connection,
1913: poll => 'r',
1914: desc => "Parent listener unix socket ($loncapa_host)",
1915: data => "",
1916: fd => $socket);
1917: $parent_dispatchers{$loncapa_host} = $watcher;
1918:
1919: }
1920:
1921: sub parent_clean_up {
1922: my ($loncapa_host) = @_;
1923: Debug(1, "parent_clean_up: $loncapa_host");
1924:
1925: my $socket_file = &GetLoncSocketPath($loncapa_host);
1926: unlink($socket_file); # No problem if it doesn't exist yet [startup e.g.]
1927: my $lock_file = $socket_file.".lock";
1928: unlink($lock_file); # No problem if it doesn't exist yet [startup e.g.]
1929: }
1930:
1931:
1932:
1933: # This sub initiates a listen on the common unix domain lonc client socket.
1934: # loncnew starts up with no children, and only spawns off children when a
1935: # connection request occurs on the common client unix socket. The spawned
1936: # child continues to run until it has been idle a while at which point it
1937: # eventually exits and once more the parent picks up the listen.
1938: #
1939: # Parameters:
1940: # NONE
1941: # Implicit Inputs:
1942: # The configuration file that has been read in by LondConnection.
1943: # Returns:
1944: # NONE
1945: #
1946: sub listen_on_common_socket {
1947: Debug(5, "listen_on_common_socket");
1948: &parent_listen();
1949: }
1950:
1951: # server_died is called whenever a child process exits.
1952: # Since this is dispatched via a signal, we must process all
1953: # dead children until there are no more left. The action
1954: # is to:
1955: # - Remove the child from the bookeeping hashes
1956: # - Re-establish a listen on the unix domain socket associated
1957: # with that host.
1958: # Parameters:
1959: # The event, but we don't actually care about it.
1960: sub server_died {
1961: &Debug(9, "server_died called...");
1962:
1963: while(1) { # Loop until waitpid nowait fails.
1964: my $pid = waitpid(-1, WNOHANG);
1965: if($pid <= 0) {
1966: return; # Nothing left to wait for.
1967: }
1968: # need the host to restart:
1969:
1970: my $host = $ChildPid{$pid};
1971: if($host) { # It's for real...
1972: &Debug(9, "Caught sigchild for $host");
1973: delete($ChildPid{$pid});
1974: delete($ChildHost{$host});
1975: &parent_clean_up($host);
1976:
1977: } else {
1978: &Debug(5, "Caught sigchild for pid not in hosts hash: $pid");
1979: }
1980: }
1981:
1982: }
1983:
1984: #
1985: # Parent process logic pass 1:
1986: # For each entry in the hosts table, we will
1987: # fork off an instance of ChildProcess to service the transactions
1988: # to that host. Each pid will be entered in a global hash
1989: # with the value of the key, the host.
1990: # The parent will then enter a loop to wait for process exits.
1991: # Each exit gets logged and the child gets restarted.
1992: #
1993:
1994: #
1995: # Fork and start in new session so hang-up isn't going to
1996: # happen without intent.
1997: #
1998:
1999:
2000:
2001:
2002:
2003:
2004: ShowStatus("Forming new session");
2005: my $childpid = fork;
2006: if ($childpid != 0) {
2007: sleep 4; # Give child a chacne to break to
2008: exit 0; # a new sesion.
2009: }
2010: #
2011: # Write my pid into the pid file so I can be located
2012: #
2013:
2014: ShowStatus("Parent writing pid file:");
2015: my $execdir = $perlvar{'lonDaemons'};
2016: open (PIDSAVE, ">$execdir/logs/lonc.pid");
2017: print PIDSAVE "$$\n";
2018: close(PIDSAVE);
2019:
2020:
2021:
2022: if (POSIX::setsid() < 0) {
2023: print "Could not create new session\n";
2024: exit -1;
2025: }
2026:
2027: ShowStatus("Forking node servers");
2028:
2029: Log("CRITICAL", "--------------- Starting children ---------------");
2030:
2031: LondConnection::ReadConfig; # Read standard config files.
2032:
2033: $RemoteHost = "[parent]";
2034: &listen_on_common_socket();
2035:
2036: $RemoteHost = "Parent Server";
2037:
2038: # Maintain the population:
2039:
2040: ShowStatus("Parent keeping the flock");
2041:
2042:
2043: # We need to setup a SIGChild event to handle the exit (natural or otherwise)
2044: # of the children.
2045:
2046: Event->signal(cb => \&server_died,
2047: desc => "Child exit handler",
2048: signal => "CHLD");
2049:
2050:
2051: # Set up all the other signals we set up.
2052:
2053: $parent_handlers{INT} = Event->signal(cb => \&Terminate,
2054: desc => "Parent INT handler",
2055: signal => "INT");
2056: $parent_handlers{TERM} = Event->signal(cb => \&Terminate,
2057: desc => "Parent TERM handler",
2058: signal => "TERM");
2059: $parent_handlers{HUP} = Event->signal(cb => \&KillThemAll,
2060: desc => "Parent HUP handler.",
2061: signal => "HUP");
2062: $parent_handlers{USR1} = Event->signal(cb => \&CheckKids,
2063: desc => "Parent USR1 handler",
2064: signal => "USR1");
2065: $parent_handlers{USR2} = Event->signal(cb => \&UpdateKids,
2066: desc => "Parent USR2 handler.",
2067: signal => "USR2");
2068:
2069: # Start procdesing events.
2070:
2071: $Event::DebugLevel = $DebugLevel;
2072: Debug(9, "Parent entering event loop");
2073: my $ret = Event::loop();
2074: die "Main Event loop exited: $ret";
2075:
2076: =pod
2077:
2078: =head1 CheckKids
2079:
2080: Since kids do not die as easily in this implementation
2081: as the previous one, there is no need to restart the
2082: dead ones (all dead kids get restarted when they die!!)
2083: The only thing this function does is to pass USR1 to the
2084: kids so that they report their status.
2085:
2086: =cut
2087:
2088: sub CheckKids {
2089: Debug(2, "Checking status of children");
2090: my $docdir = $perlvar{'lonDocRoot'};
2091: my $fh = IO::File->new(">$docdir/lon-status/loncstatus.txt");
2092: my $now=time;
2093: my $local=localtime($now);
2094: print $fh "LONC status $local - parent $$ \n\n";
2095: foreach my $host (keys %parent_dispatchers) {
2096: print $fh "LONC Parent process listening for $host\n";
2097: }
2098: foreach my $pid (keys %ChildPid) {
2099: Debug(2, "Sending USR1 -> $pid");
2100: kill 'USR1' => $pid; # Tell Child to report status.
2101: }
2102:
2103: }
2104:
2105: =pod
2106:
2107: =head1 UpdateKids
2108:
2109: parent's SIGUSR2 handler. This handler:
2110:
2111: =item
2112:
2113: Rereads the hosts file.
2114:
2115: =item
2116:
2117: Kills off (via sigint) children for hosts that have disappeared.
2118:
2119: =item
2120:
2121: QUITs children for hosts that already exist (this just forces a status display
2122: and resets the connection retry count for that host.
2123:
2124: =item
2125:
2126: Starts new children for hosts that have been added to the hosts.tab file since
2127: the start of the master program and maintains them.
2128:
2129: =cut
2130:
2131: sub UpdateKids {
2132:
2133: Log("INFO", "Updating connections via SIGUSR2");
2134:
2135: # I'm not sure what I was thinking in the first implementation.
2136: # someone will have to work hard to convince me the effect is any
2137: # different than Restart, especially now that we don't start up
2138: # per host servers automatically, may as well just restart.
2139: # The down side is transactions that are in flight will get timed out
2140: # (lost unless they are critical).
2141:
2142: &KillThemAll();
2143: }
2144:
2145:
2146: =pod
2147:
2148: =head1 Restart
2149:
2150: Signal handler for HUP... all children are killed and
2151: we self restart. This is an el-cheapo way to re read
2152: the config file.
2153:
2154: =cut
2155:
2156: sub Restart {
2157: &KillThemAll; # First kill all the children.
2158: Log("CRITICAL", "Restarting");
2159: my $execdir = $perlvar{'lonDaemons'};
2160: unlink("$execdir/logs/lonc.pid");
2161: exec("$executable");
2162: }
2163:
2164: =pod
2165:
2166: =head1 KillThemAll
2167:
2168: Signal handler that kills all children by sending them a
2169: SIGHUP. Responds to sigint and sigterm.
2170:
2171: =cut
2172:
2173: sub KillThemAll {
2174: Debug(2, "Kill them all!!");
2175:
2176: #local($SIG{CHLD}) = 'IGNORE';
2177: # Our children >will< die.
2178: # but we need to catch their death and cleanup after them in case this is
2179: # a restart set of kills
2180: my @allpids = keys(%ChildPid);
2181: foreach my $pid (@allpids) {
2182: my $serving = $ChildPid{$pid};
2183: ShowStatus("Nicely Killing lonc for $serving pid = $pid");
2184: Log("CRITICAL", "Nicely Killing lonc for $serving pid = $pid");
2185: kill 'QUIT' => $pid;
2186: }
2187: ShowStatus("Finished killing child processes off.");
2188: }
2189:
2190:
2191: #
2192: # Kill all children via KILL. Just in case the
2193: # first shot didn't get them.
2194:
2195: sub really_kill_them_all_dammit
2196: {
2197: Debug(2, "Kill them all Dammit");
2198: local($SIG{CHLD} = 'IGNORE'); # In case some purist reenabled them.
2199: foreach my $pid (keys %ChildPid) {
2200: my $serving = $ChildPid{$pid};
2201: &ShowStatus("Nastily killing lonc for $serving pid = $pid");
2202: Log("CRITICAL", "Nastily killing lonc for $serving pid = $pid");
2203: kill 'KILL' => $pid;
2204: delete($ChildPid{$pid});
2205: my $execdir = $perlvar{'lonDaemons'};
2206: unlink("$execdir/logs/lonc.pid");
2207: }
2208: }
2209:
2210: =pod
2211:
2212: =head1 Terminate
2213:
2214: Terminate the system.
2215:
2216: =cut
2217:
2218: sub Terminate {
2219: &Log("CRITICAL", "Asked to kill children.. first be nice...");
2220: &KillThemAll;
2221: #
2222: # By now they really should all be dead.. but just in case
2223: # send them all SIGKILL's after a bit of waiting:
2224:
2225: sleep(4);
2226: &Log("CRITICAL", "Now kill children nasty");
2227: &really_kill_them_all_dammit;
2228: Log("CRITICAL","Master process exiting");
2229: exit 0;
2230:
2231: }
2232:
2233: sub my_hostname {
2234: use Sys::Hostname;
2235: my $name = &hostname();
2236: &Debug(9,"Name is $name");
2237: return $name;
2238: }
2239:
2240: =pod
2241:
2242: =head1 Theory
2243:
2244: The event class is used to build this as a single process with an
2245: event driven model. The following events are handled:
2246:
2247: =item UNIX Socket connection Received
2248:
2249: =item Request data arrives on UNIX data transfer socket.
2250:
2251: =item lond connection becomes writable.
2252:
2253: =item timer fires at 1 second intervals.
2254:
2255: All sockets are run in non-blocking mode. Timeouts managed by the timer
2256: handler prevents hung connections.
2257:
2258: Key data structures:
2259:
2260: =item RequestQueue
2261:
2262: A queue of requests received from UNIX sockets that are
2263: waiting for a chance to be forwarded on a lond connection socket.
2264:
2265: =item ActiveConnections
2266:
2267: A hash of lond connections that have transactions in process that are
2268: available to be timed out.
2269:
2270: =item ActiveTransactions
2271:
2272: A hash indexed by lond connections that contain the client reply
2273: socket for each connection that has an active transaction on it.
2274:
2275: =item IdleConnections
2276:
2277: A hash of lond connections that have no work to do. These connections
2278: can be closed if they are idle for a long enough time.
2279:
2280: =cut
2281:
2282: =pod
2283:
2284: =head1 Log messages
2285:
2286: The following is a list of log messages that can appear in the
2287: lonc.log file. Each log file has a severity and a message.
2288:
2289: =over 2
2290:
2291: =item Warning A socket timeout was detected
2292:
2293: If there are pending transactions in the socket's queue,
2294: they are failed (saved if critical). If the connection
2295: retry count gets exceeded by this, the
2296: remote host is marked as dead.
2297: Called when timeouts occured during the connection and
2298: connection dialog with a remote host.
2299:
2300: =item Critical Host makred DEAD <hostname>
2301:
2302: The numer of retry counts for contacting a host was
2303: exceeded. The host is marked dead an no
2304: further attempts will be made by that child.
2305:
2306: =item Info lonc pipe client hung up on us
2307:
2308: Write to the client pipe indicated no data transferred
2309: Socket to remote host is shut down. Reply to the client
2310: is discarded. Note: This is commented out in &ClientWriteable
2311:
2312: =item Success Reply from lond: <data>
2313:
2314: Can be enabled for debugging by setting LogTransactions to nonzero.
2315: Indicates a successful transaction with lond, <data> is the data received
2316: from the remote lond.
2317:
2318: =item Success A delayed transaction was completed
2319:
2320: A transaction that must be reliable was executed and completed
2321: as lonc restarted. This is followed by a mesage of the form
2322:
2323: S: client-name : request
2324:
2325: =item WARNING Failing transaction <cmd>:<subcmd>
2326:
2327: Transaction failed on a socket, but the failure retry count for the remote
2328: node has not yet been exhausted (the node is not yet marked dead).
2329: cmd is the command, subcmd is the subcommand. This results from a con_lost
2330: when communicating with lond.
2331:
2332: =item WARNING Shutting down a socket
2333:
2334: Called when a socket is being closed to lond. This is emitted both when
2335: idle pruning is being done and when the socket has been disconnected by the remote.
2336:
2337: =item WARNING Lond connection lost.
2338:
2339: Called when a read from lond's socket failed indicating lond has closed the
2340: connection or died. This should be followed by one or more
2341:
2342: "WARNING Failing transaction..." msgs for each in-flight or queued transaction.
2343:
2344: =item INFO Connected to lond version: <version>
2345:
2346: When connection negotiation is complete, the lond version is requested and logged here.
2347:
2348: =item SUCCESS Connection n to host now ready for action
2349:
2350: Emitted when connection has been completed with lond. n is then number of
2351: concurrent connections and host, the host to which the connection has just
2352: been established.
2353:
2354: =item WARNING Connection to host has been disconnected
2355:
2356: Write to a lond resulted in failure status. Connection to lond is dropped.
2357:
2358: =item SUCCESS Created connection n to host host
2359:
2360: Initial connection request to host..(before negotiation).
2361:
2362: =item CRITICAL Request Close Connection ... exiting
2363:
2364: Client has sent "close_connection_exit" The loncnew server is exiting.
2365:
2366: =item INFO Resetting Connection Retries
2367:
2368: Client has sent "reset_retries" The lond connection retries are reset to zero for the
2369: corresponding lond.
2370:
2371: =item SUCCESS Transaction <data>
2372:
2373: Only emitted if the global variable $LogTransactions was set to true.
2374: A client has requested a lond transaction <data> is the contents of the request.
2375:
2376: =item SUCCESS Toggled transaction logging <LogTransactions>
2377:
2378: The state of the $LogTransactions global has been toggled, and its current value
2379: (after being toggled) is displayed. When non zero additional logging of transactions
2380: is enabled for debugging purposes. Transaction logging is toggled on receipt of a USR2
2381: signal.
2382:
2383: =item CRITICAL Abnormal exit. Child <pid> for <host> died thorugh signal.
2384:
2385: QUIT signal received. lonc child process is exiting.
2386:
2387: =item SUCCESS New debugging level for <RemoteHost> now <DebugLevel>
2388:
2389: Debugging toggled for the host loncnew is talking with.
2390: Currently debugging is a level based scheme with higher number
2391: conveying more information. The daemon starts out at
2392: DebugLevel 0 and can toggle back and forth between that and
2393: DebugLevel 2 These are controlled by
2394: the global variables $DebugLevel and $NextDebugLevel
2395: The debug level can go up to 9.
2396: SIGINT toggles the debug level. The higher the debug level the
2397: more debugging information is spewed. See the Debug
2398: sub in loncnew.
2399:
2400: =item CRITICAL Forking server for host
2401:
2402: A child is being created to service requests for the specified host.
2403:
2404:
2405: =item WARNING Request for a second child on hostname
2406:
2407: Somehow loncnew was asked to start a second child on a host that already had a child
2408: servicing it. This request is not honored, but themessage is emitted. This could happen
2409: due to a race condition. When a client attempts to contact loncnew for a new host, a child
2410: is forked off to handle the requests for that server. The parent then backs off the Unix
2411: domain socket leaving it for the child to service all requests. If in the time between
2412: creating the child, and backing off, a new connection request comes in to the unix domain
2413: socket, this could trigger (unlikely but remotely possible),.
2414:
2415: =item CRITICAL ------ Starting Children ----
2416:
2417: This message should probably be changed to "Entering event loop" as the loncnew only starts
2418: children as needed. This message is emitted as new events are established and
2419: the event processing loop is entered.
2420:
2421: =item INFO Updating connections via SIGUSR2
2422:
2423: SIGUSR2 received. The original code would kill all clients, re-read the host file,
2424: then restart children for each host. Now that childrean aree started on demand, this
2425: just kills all child processes and lets requests start them as needed again.
2426:
2427:
2428: =item CRITICAL Restarting
2429:
2430: SigHUP received. all the children are killed and the script exec's itself to start again.
2431:
2432: =item CRITICAL Nicely killing lonc for host pid = <pid>
2433:
2434: Attempting to kill the child that is serving the specified host (pid given) cleanly via
2435: SIGQUIT The child should handle that, clean up nicely and exit.
2436:
2437: =item CRITICAL Nastily killing lonc for host pid = <pid>
2438:
2439: The child specified did not die when requested via SIGQUIT. Therefore it is killed
2440: via SIGKILL.
2441:
2442: =item CRITICAL Asked to kill children.. first be nice..
2443:
2444: In the parent's INT handler. INT kills the child processes. This inidicate loncnew
2445: is about to attempt to kill all known children via SIGQUIT. This message should be followed
2446: by one "Nicely killing" message for each extant child.
2447:
2448: =item CRITICAL Now kill children nasty
2449:
2450: In the parent's INT handler. remaining children are about to be killed via
2451: SIGKILL. Should be followed by a Nastily killing... for each lonc child that
2452: refused to die.
2453:
2454: =item CRITICAL Master process exiting
2455:
2456: In the parent's INT handler. just prior to the exit 0 call.
2457:
2458: =back
2459:
2460: =cut
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>
![[BACK]](/icons/back.gif){kind=icon}
Return to loncnew CVS log ![[TXT]](/icons/text.gif){kind=icon}
![[DIR]](/icons/dir.gif){kind=icon}
Up to [LON-CAPA] / loncom