File:
[LON-CAPA] /
loncom /
loncnew
Revision
1.34:
download - view:
text,
annotated -
select for diffs
Thu Dec 11 23:18:37 2003 UTC (20 years, 9 months ago) by
albertel
Branches:
MAIN
CVS tags:
HEAD
- connections could get frozen in the connectiong state if the remote end accepted the connection but never got back to us on any of our commands BUG#2488
1: #!/usr/bin/perl
2: # The LearningOnline Network with CAPA
3: # lonc maintains the connections to remote computers
4: #
5: # $Id: loncnew,v 1.34 2003/12/11 23:18:37 albertel 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.
39: # - Add Signal handling - HUP restarts. USR1 status report.
40: # - Add Configuration file I/O (done).
41: # - Add management/status request interface.
42: # - Add deferred request capability. (done)
43: # - Detect transmission timeouts.
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 LONCAPA::HashIterator;
64:
65:
66: #
67: # Disable all signals we might receive from outside for now.
68: #
69:
70:
71: # Read the httpd configuration file to get perl variables
72: # normally set in apache modules:
73:
74: my $perlvarref = LONCAPA::Configuration::read_conf('loncapa.conf');
75: my %perlvar = %{$perlvarref};
76:
77: #
78: # parent and shared variables.
79:
80: my %ChildHash; # by pid -> host.
81: my %HostToPid; # By host -> pid.
82: my %HostHash; # by loncapaname -> IP.
83:
84:
85: my $MaxConnectionCount = 10; # Will get from config later.
86: my $ClientConnection = 0; # Uniquifier for client events.
87:
88: my $DebugLevel = 0;
89: my $NextDebugLevel= 2; # So Sigint can toggle this.
90: my $IdleTimeout= 3600; # Wait an hour before pruning connections.
91:
92: #
93: # The variables below are only used by the child processes.
94: #
95: my $RemoteHost; # Name of host child is talking to.
96: my $UnixSocketDir= $perlvar{'lonSockDir'};
97: my $IdleConnections = Stack->new(); # Set of idle connections
98: my %ActiveConnections; # Connections to the remote lond.
99: my %ActiveTransactions; # LondTransactions in flight.
100: my %ActiveClients; # Serial numbers of active clients by socket.
101: my $WorkQueue = Queue->new(); # Queue of pending transactions.
102: my $ConnectionCount = 0;
103: my $IdleSeconds = 0; # Number of seconds idle.
104: my $Status = ""; # Current status string.
105: my $RecentLogEntry = "";
106: my $ConnectionRetries=2; # Number of connection retries allowed.
107: my $ConnectionRetriesLeft=2; # Number of connection retries remaining.
108:
109: #
110: # The hash below gives the HTML format for log messages
111: # given a severity.
112: #
113: my %LogFormats;
114:
115: $LogFormats{"CRITICAL"} = "<font color=red>CRITICAL: %s</font>";
116: $LogFormats{"SUCCESS"} = "<font color=green>SUCCESS: %s</font>";
117: $LogFormats{"INFO"} = "<font color=yellow>INFO: %s</font>";
118: $LogFormats{"WARNING"} = "<font color=blue>WARNING: %s</font>";
119: $LogFormats{"DEFAULT"} = " %s ";
120:
121:
122:
123: =pod
124:
125: =head2 LogPerm
126:
127: Makes an entry into the permanent log file.
128:
129: =cut
130: sub LogPerm {
131: my $message=shift;
132: my $execdir=$perlvar{'lonDaemons'};
133: my $now=time;
134: my $local=localtime($now);
135: my $fh=IO::File->new(">>$execdir/logs/lonnet.perm.log");
136: print $fh "$now:$message:$local\n";
137: }
138:
139: =pod
140:
141: =head2 Log
142:
143: Logs a message to the log file.
144: Parameters:
145:
146: =item severity
147:
148: One of CRITICAL, WARNING, INFO, SUCCESS used to select the
149: format string used to format the message. if the severity is
150: not a defined severity the Default format string is used.
151:
152: =item message
153:
154: The base message. In addtion to the format string, the message
155: will be appended to a string containing the name of our remote
156: host and the time will be formatted into the message.
157:
158: =cut
159:
160: sub Log {
161: my $severity = shift;
162: my $message = shift;
163:
164: if(!$LogFormats{$severity}) {
165: $severity = "DEFAULT";
166: }
167:
168: my $format = $LogFormats{$severity};
169:
170: # Put the window dressing in in front of the message format:
171:
172: my $now = time;
173: my $local = localtime($now);
174: my $finalformat = "$local ($$) [$RemoteHost] [$Status] ";
175: my $finalformat = $finalformat.$format."\n";
176:
177: # open the file and put the result.
178:
179: my $execdir = $perlvar{'lonDaemons'};
180: my $fh = IO::File->new(">>$execdir/logs/lonc.log");
181: my $msg = sprintf($finalformat, $message);
182: $RecentLogEntry = $msg;
183: print $fh $msg;
184:
185:
186: }
187:
188:
189: =pod
190:
191: =head2 GetPeerName
192:
193: Returns the name of the host that a socket object is connected to.
194:
195: =cut
196:
197: sub GetPeername {
198: my $connection = shift;
199: my $AdrFamily = shift;
200: my $peer = $connection->peername();
201: my $peerport;
202: my $peerip;
203: if($AdrFamily == AF_INET) {
204: ($peerport, $peerip) = sockaddr_in($peer);
205: my $peername = gethostbyaddr($peerip, $AdrFamily);
206: return $peername;
207: } elsif ($AdrFamily == AF_UNIX) {
208: my $peerfile;
209: ($peerfile) = sockaddr_un($peer);
210: return $peerfile;
211: }
212: }
213: #----------------------------- Timer management ------------------------
214: =pod
215:
216: =head2 Debug
217:
218: Invoked to issue a debug message.
219:
220: =cut
221:
222: sub Debug {
223: my $level = shift;
224: my $message = shift;
225: if ($level <= $DebugLevel) {
226: Log("INFO", "-Debug- $message host = $RemoteHost");
227: }
228: }
229:
230: sub SocketDump {
231: my $level = shift;
232: my $socket= shift;
233: if($level <= $DebugLevel) {
234: $socket->Dump();
235: }
236: }
237:
238: =pod
239:
240: =head2 ShowStatus
241:
242: Place some text as our pid status.
243: and as what we return in a SIGUSR1
244:
245: =cut
246: sub ShowStatus {
247: my $state = shift;
248: my $now = time;
249: my $local = localtime($now);
250: $Status = $local.": ".$state;
251: $0='lonc: '.$state.' '.$local;
252: }
253:
254: =pod
255:
256: =head 2 SocketTimeout
257:
258: Called when an action on the socket times out. The socket is
259: destroyed and any active transaction is failed.
260:
261:
262: =cut
263: sub SocketTimeout {
264: my $Socket = shift;
265:
266: KillSocket($Socket); # A transaction timeout also counts as
267: # a connection failure:
268: $ConnectionRetriesLeft--;
269: }
270:
271: =pod
272:
273: =head2 Tick
274:
275: Invoked each timer tick.
276:
277: =cut
278:
279:
280: sub Tick {
281: my $client;
282: if($ConnectionRetriesLeft > 0) {
283: ShowStatus(GetServerHost()." Connection count: ".$ConnectionCount
284: ." Retries remaining: ".$ConnectionRetriesLeft);
285: } else {
286: ShowStatus(GetServerHost()." >> DEAD <<");
287: }
288: # Is it time to prune connection count:
289:
290:
291: if($IdleConnections->Count() &&
292: ($WorkQueue->Count() == 0)) { # Idle connections and nothing to do?
293: $IdleSeconds++;
294: if($IdleSeconds > $IdleTimeout) { # Prune a connection...
295: my $Socket = $IdleConnections->pop();
296: KillSocket($Socket);
297: }
298: } else {
299: $IdleSeconds = 0; # Reset idle count if not idle.
300: }
301: #
302: # For each inflight transaction, tick down its timeout counter.
303: #
304: foreach my $item (keys %ActiveTransactions) {
305: my $Socket = $ActiveTransactions{$item}->getServer();
306: $Socket->Tick();
307: }
308: foreach my $item (keys %ActiveConnections) {
309: my $State = $ActiveConnections{$item}->data->GetState();
310: if ($State ne 'Idle' && $State ne 'SendingRequest' &&
311: $State ne 'ReceivingReply') {
312: Debug(5,"Ticking Socket $State $item");
313: $ActiveConnections{$item}->data->Tick();
314: }
315: }
316: # Do we have work in the queue, but no connections to service them?
317: # If so, try to make some new connections to get things going again.
318: #
319:
320: my $Requests = $WorkQueue->Count();
321: if (($ConnectionCount == 0) && ($Requests > 0)) {
322: if ($ConnectionRetriesLeft > 0) {
323: my $Connections = ($Requests <= $MaxConnectionCount) ?
324: $Requests : $MaxConnectionCount;
325: Debug(5,"Work but no connections, start ".$Connections." of them");
326: my $successCount = 0;
327: for (my $i =0; $i < $Connections; $i++) {
328: $successCount += MakeLondConnection();
329: }
330: if($successCount == 0) { # All connections failed:
331: Debug(5,"Work in queue failed to make any connectiouns\n");
332: EmptyQueue(); # Fail pending transactions with con_lost.
333: }
334: } else {
335: ShowStatus(GetServerHost()." >>> DEAD!!! <<<");
336: Debug(5,"Work in queue, but gave up on connections..flushing\n");
337: EmptyQueue(); # Connections can't be established.
338: }
339:
340: }
341: }
342:
343: =pod
344:
345: =head2 SetupTimer
346:
347: Sets up a 1 per sec recurring timer event. The event handler is used to:
348:
349: =item
350:
351: Trigger timeouts on communications along active sockets.
352:
353: =item
354:
355: Trigger disconnections of idle sockets.
356:
357: =cut
358:
359: sub SetupTimer {
360: Debug(6, "SetupTimer");
361: Event->timer(interval => 1, cb => \&Tick );
362: }
363:
364: =pod
365:
366: =head2 ServerToIdle
367:
368: This function is called when a connection to the server is
369: ready for more work.
370:
371: If there is work in the Work queue the top element is dequeued
372: and the connection will start to work on it. If the work queue is
373: empty, the connection is pushed on the idle connection stack where
374: it will either get another work unit, or alternatively, if it sits there
375: long enough, it will be shut down and released.
376:
377: =cut
378:
379: sub ServerToIdle {
380: my $Socket = shift; # Get the socket.
381: delete($ActiveTransactions{$Socket}); # Server has no transaction
382:
383: &Debug(5, "Server to idle");
384:
385: # If there's work to do, start the transaction:
386:
387: my $reqdata = $WorkQueue->dequeue(); # This is a LondTransaction
388: if ($reqdata ne undef) {
389: Debug(5, "Queue gave request data: ".$reqdata->getRequest());
390: &StartRequest($Socket, $reqdata);
391:
392: } else {
393:
394: # There's no work waiting, so push the server to idle list.
395: &Debug(5, "No new work requests, server connection going idle");
396: $IdleConnections->push($Socket);
397: }
398: }
399:
400: =pod
401:
402: =head2 ClientWritable
403:
404: Event callback for when a client socket is writable.
405:
406: This callback is established when a transaction reponse is
407: avaiable from lond. The response is forwarded to the unix socket
408: as it becomes writable in this sub.
409:
410: Parameters:
411:
412: =item Event
413:
414: The event that has been triggered. Event->w->data is
415: the data and Event->w->fd is the socket to write.
416:
417: =cut
418:
419: sub ClientWritable {
420: my $Event = shift;
421: my $Watcher = $Event->w;
422: my $Data = $Watcher->data;
423: my $Socket = $Watcher->fd;
424:
425: # Try to send the data:
426:
427: &Debug(6, "ClientWritable writing".$Data);
428: &Debug(9, "Socket is: ".$Socket);
429:
430: if($Socket->connected) {
431: my $result = $Socket->send($Data, 0);
432:
433: # $result undefined: the write failed.
434: # otherwise $result is the number of bytes written.
435: # Remove that preceding string from the data.
436: # If the resulting data is empty, destroy the watcher
437: # and set up a read event handler to accept the next
438: # request.
439:
440: &Debug(9,"Send result is ".$result." Defined: ".defined($result));
441: if($result ne undef) {
442: &Debug(9, "send result was defined");
443: if($result == length($Data)) { # Entire string sent.
444: &Debug(9, "ClientWritable data all written");
445: $Watcher->cancel();
446: #
447: # Set up to read next request from socket:
448:
449: my $descr = sprintf("Connection to lonc client %d",
450: $ActiveClients{$Socket});
451: Event->io(cb => \&ClientRequest,
452: poll => 'r',
453: desc => $descr,
454: data => "",
455: fd => $Socket);
456:
457: } else { # Partial string sent.
458: $Watcher->data(substr($Data, $result));
459: if($result == 0) { # client hung up on us!!
460: Log("INFO", "lonc pipe client hung up on us!");
461: $Watcher->cancel;
462: $Socket->shutdown(2);
463: $Socket->close();
464: }
465: }
466:
467: } else { # Error of some sort...
468:
469: # Some errnos are possible:
470: my $errno = $!;
471: if($errno == POSIX::EWOULDBLOCK ||
472: $errno == POSIX::EAGAIN ||
473: $errno == POSIX::EINTR) {
474: # No action taken?
475: } else { # Unanticipated errno.
476: &Debug(5,"ClientWritable error or peer shutdown".$RemoteHost);
477: $Watcher->cancel; # Stop the watcher.
478: $Socket->shutdown(2); # Kill connection
479: $Socket->close(); # Close the socket.
480: }
481:
482: }
483: } else {
484: $Watcher->cancel(); # A delayed request...just cancel.
485: }
486: }
487:
488: =pod
489:
490: =head2 CompleteTransaction
491:
492: Called when the reply data has been received for a lond
493: transaction. The reply data must now be sent to the
494: ultimate client on the other end of the Unix socket. This is
495: done by setting up a writable event for the socket with the
496: data the reply data.
497:
498: Parameters:
499:
500: =item Socket
501:
502: Socket on which the lond transaction occured. This is a
503: LondConnection. The data received is in the TransactionReply member.
504:
505: =item Transaction
506:
507: The transaction that is being completed.
508:
509: =cut
510:
511: sub CompleteTransaction {
512: &Debug(5,"Complete transaction");
513: my $Socket = shift;
514: my $Transaction = shift;
515:
516: if (!$Transaction->isDeferred()) { # Normal transaction
517: my $data = $Socket->GetReply(); # Data to send.
518: StartClientReply($Transaction, $data);
519: } else { # Delete deferred transaction file.
520: Log("SUCCESS", "A delayed transaction was completed");
521: LogPerm("S:$Transaction->getClient() :".$Transaction->getRequest());
522: unlink $Transaction->getFile();
523: }
524: }
525: =pod
526: =head1 StartClientReply
527:
528: Initiates a reply to a client where the reply data is a parameter.
529:
530: =head2 parameters:
531:
532: =item Transaction
533:
534: The transaction for which we are responding to the client.
535:
536: =item data
537:
538: The data to send to apached client.
539:
540: =cut
541: sub StartClientReply {
542: my $Transaction = shift;
543: my $data = shift;
544:
545:
546: my $Client = $Transaction->getClient();
547:
548: &Debug(8," Reply was: ".$data);
549: my $Serial = $ActiveClients{$Client};
550: my $desc = sprintf("Connection to lonc client %d",
551: $Serial);
552: Event->io(fd => $Client,
553: poll => "w",
554: desc => $desc,
555: cb => \&ClientWritable,
556: data => $data);
557: }
558: =pod
559: =head2 FailTransaction
560:
561: Finishes a transaction with failure because the associated lond socket
562: disconnected. There are two possibilities:
563: - The transaction is deferred: in which case we just quietly
564: delete the transaction since there is no client connection.
565: - The transaction is 'live' in which case we initiate the sending
566: of "con_lost" to the client.
567:
568: Deleting the transaction means killing it from the
569: %ActiveTransactions hash.
570:
571: Parameters:
572:
573: =item client
574:
575: The LondTransaction we are failing.
576:
577: =cut
578:
579: sub FailTransaction {
580: my $transaction = shift;
581: Log("WARNING", "Failing transaction ".$transaction->getRequest());
582: Debug(1, "Failing transaction: ".$transaction->getRequest());
583: if (!$transaction->isDeferred()) { # If the transaction is deferred we'll get to it.
584: my $client = $transaction->getClient();
585: Debug(1," Replying con_lost to ".$transaction->getRequest());
586: StartClientReply($transaction, "con_lost\n");
587: }
588: if($ConnectionRetriesLeft <= 0) {
589: Log("CRITICAL", "Host marked dead: ".GetServerHost());
590: }
591:
592: }
593:
594: =pod
595: =head1 EmptyQueue
596:
597: Fails all items in the work queue with con_lost.
598: Note that each item in the work queue is a transaction.
599:
600: =cut
601: sub EmptyQueue {
602: $ConnectionRetriesLeft--; # Counts as connection failure too.
603: while($WorkQueue->Count()) {
604: my $request = $WorkQueue->dequeue(); # This is a transaction
605: FailTransaction($request);
606: }
607: }
608:
609: =pod
610:
611: =head2 CloseAllLondConnections
612:
613: Close all connections open on lond prior to exit e.g.
614:
615: =cut
616: sub CloseAllLondConnections {
617: foreach my $Socket (keys %ActiveConnections) {
618: KillSocket($Socket);
619: }
620: }
621: =cut
622:
623: =pod
624:
625: =head2 KillSocket
626:
627: Destroys a socket. This function can be called either when a socket
628: has died of 'natural' causes or because a socket needs to be pruned due to
629: idleness. If the socket has died naturally, if there are no longer any
630: live connections a new connection is created (in case there are transactions
631: in the queue). If the socket has been pruned, it is never re-created.
632:
633: Parameters:
634:
635: =item Socket
636:
637: The socket to kill off.
638:
639: =item Restart
640:
641: nonzero if we are allowed to create a new connection.
642:
643:
644: =cut
645: sub KillSocket {
646: my $Socket = shift;
647:
648: Log("WARNING", "Shutting down a socket");
649: $Socket->Shutdown();
650:
651: # If the socket came from the active connection set,
652: # delete its transaction... note that FailTransaction should
653: # already have been called!!!
654: # otherwise it came from the idle set.
655: #
656:
657: if(exists($ActiveTransactions{$Socket})) {
658: delete ($ActiveTransactions{$Socket});
659: }
660: if(exists($ActiveConnections{$Socket})) {
661: delete($ActiveConnections{$Socket});
662: }
663: $ConnectionCount--;
664:
665: # If the connection count has gone to zero and there is work in the
666: # work queue, the work all gets failed with con_lost.
667: #
668: if($ConnectionCount == 0) {
669: EmptyQueue();
670: }
671: }
672:
673: =pod
674:
675: =head2 LondReadable
676:
677: This function is called whenever a lond connection
678: is readable. The action is state dependent:
679:
680: =head3 State=Initialized
681:
682: We''re waiting for the challenge, this is a no-op until the
683: state changes.
684:
685: =head3 State=Challenged
686:
687: The challenge has arrived we need to transition to Writable.
688: The connection must echo the challenge back.
689:
690: =head3 State=ChallengeReplied
691:
692: The challenge has been replied to. The we are receiveing the
693: 'ok' from the partner.
694:
695: =head3 State=RequestingKey
696:
697: The ok has been received and we need to send the request for
698: an encryption key. Transition to writable for that.
699:
700: =head3 State=ReceivingKey
701:
702: The the key has been requested, now we are reading the new key.
703:
704: =head3 State=Idle
705:
706: The encryption key has been negotiated or we have finished
707: reading data from the a transaction. If the callback data has
708: a client as well as the socket iformation, then we are
709: doing a transaction and the data received is relayed to the client
710: before the socket is put on the idle list.
711:
712: =head3 State=SendingRequest
713:
714: I do not think this state can be received here, but if it is,
715: the appropriate thing to do is to transition to writable, and send
716: the request.
717:
718: =head3 State=ReceivingReply
719:
720: We finished sending the request to the server and now transition
721: to readable to receive the reply.
722:
723: The parameter to this function are:
724:
725: The event. Implicit in this is the watcher and its data. The data
726: contains at least the lond connection object and, if a
727: transaction is in progress, the socket attached to the local client.
728:
729: =cut
730:
731: sub LondReadable {
732:
733: my $Event = shift;
734: my $Watcher = $Event->w;
735: my $Socket = $Watcher->data;
736: my $client = undef;
737:
738: &Debug(6,"LondReadable called state = ".$Socket->GetState());
739:
740:
741: my $State = $Socket->GetState(); # All action depends on the state.
742:
743: SocketDump(6, $Socket);
744: my $status = $Socket->Readable();
745:
746: &Debug(2, "Socket->Readable returned: $status");
747:
748: if($status != 0) {
749: # bad return from socket read. Currently this means that
750: # The socket has become disconnected. We fail the transaction.
751:
752: Log("WARNING",
753: "Lond connection lost.");
754: if(exists($ActiveTransactions{$Socket})) {
755: FailTransaction($ActiveTransactions{$Socket});
756: }
757: $Watcher->cancel();
758: KillSocket($Socket);
759: $ConnectionRetriesLeft--; # Counts as connection failure
760: return;
761: }
762: SocketDump(6,$Socket);
763:
764: $State = $Socket->GetState(); # Update in case of transition.
765: &Debug(6, "After read, state is ".$State);
766:
767: if($State eq "Initialized") {
768:
769:
770: } elsif ($State eq "ChallengeReceived") {
771: # The challenge must be echoed back; The state machine
772: # in the connection takes care of setting that up. Just
773: # need to transition to writable:
774:
775: $Watcher->cb(\&LondWritable);
776: $Watcher->poll("w");
777:
778: } elsif ($State eq "ChallengeReplied") {
779:
780:
781: } elsif ($State eq "RequestingKey") {
782: # The ok was received. Now we need to request the key
783: # That requires us to be writable:
784:
785: $Watcher->cb(\&LondWritable);
786: $Watcher->poll("w");
787:
788: } elsif ($State eq "ReceivingKey") {
789:
790: } elsif ($State eq "Idle") {
791: # If necessary, complete a transaction and then go into the
792: # idle queue.
793: # Note that a trasition to idle indicates a live lond
794: # on the other end so reset the connection retries.
795: #
796: $ConnectionRetriesLeft = $ConnectionRetries; # success resets the count
797: $Watcher->cancel();
798: if(exists($ActiveTransactions{$Socket})) {
799: Debug(5,"Completing transaction!!");
800: CompleteTransaction($Socket,
801: $ActiveTransactions{$Socket});
802: } else {
803: Log("SUCCESS", "Connection ".$ConnectionCount." to "
804: .$RemoteHost." now ready for action");
805: }
806: ServerToIdle($Socket); # Next work unit or idle.
807:
808: } elsif ($State eq "SendingRequest") {
809: # We need to be writable for this and probably don't belong
810: # here inthe first place.
811:
812: Deubg(6, "SendingRequest state encountered in readable");
813: $Watcher->poll("w");
814: $Watcher->cb(\&LondWritable);
815:
816: } elsif ($State eq "ReceivingReply") {
817:
818:
819: } else {
820: # Invalid state.
821: Debug(4, "Invalid state in LondReadable");
822: }
823: }
824:
825: =pod
826:
827: =head2 LondWritable
828:
829: This function is called whenever a lond connection
830: becomes writable while there is a writeable monitoring
831: event. The action taken is very state dependent:
832:
833: =head3 State = Connected
834:
835: The connection is in the process of sending the 'init' hailing to the
836: lond on the remote end. The connection object''s Writable member is
837: called. On error, ConnectionError is called to destroy the connection
838: and remove it from the ActiveConnections hash
839:
840: =head3 Initialized
841:
842: 'init' has been sent, writability monitoring is removed and
843: readability monitoring is started with LondReadable as the callback.
844:
845: =head3 ChallengeReceived
846:
847: The connection has received the who are you challenge from the remote
848: system, and is in the process of sending the challenge
849: response. Writable is called.
850:
851: =head3 ChallengeReplied
852:
853: The connection has replied to the initial challenge The we switch to
854: monitoring readability looking for the server to reply with 'ok'.
855:
856: =head3 RequestingKey
857:
858: The connection is in the process of requesting its encryption key.
859: Writable is called.
860:
861: =head3 ReceivingKey
862:
863: The connection has sent the request for a key. Switch to readability
864: monitoring to accept the key
865:
866: =head3 SendingRequest
867:
868: The connection is in the process of sending a request to the server.
869: This request is part of a client transaction. All the states until
870: now represent the client setup protocol. Writable is called.
871:
872: =head3 ReceivingReply
873:
874: The connection has sent a request. Now it must receive a reply.
875: Readability monitoring is requested.
876:
877: This function is an event handler and therefore receives as
878: a parameter the event that has fired. The data for the watcher
879: of this event is a reference to a list of one or two elements,
880: depending on state. The first (and possibly only) element is the
881: socket. The second (present only if a request is in progress)
882: is the socket on which to return a reply to the caller.
883:
884: =cut
885:
886: sub LondWritable {
887: my $Event = shift;
888: my $Watcher = $Event->w;
889: my $Socket = $Watcher->data;
890: my $State = $Socket->GetState();
891:
892: Debug(6,"LondWritable State = ".$State."\n");
893:
894:
895: # Figure out what to do depending on the state of the socket:
896:
897:
898:
899:
900: SocketDump(6,$Socket);
901:
902: if ($State eq "Connected") {
903:
904: if ($Socket->Writable() != 0) {
905: # The write resulted in an error.
906: # We'll treat this as if the socket got disconnected:
907: Log("WARNING", "Connection to ".$RemoteHost.
908: " has been disconnected");
909: FailTransaction($ActiveTransactions{$Socket});
910: $Watcher->cancel();
911: KillSocket($Socket);
912: return;
913: }
914: # "init" is being sent...
915:
916:
917: } elsif ($State eq "Initialized") {
918:
919: # Now that init was sent, we switch
920: # to watching for readability:
921:
922: $Watcher->cb(\&LondReadable);
923: $Watcher->poll("r");
924:
925: } elsif ($State eq "ChallengeReceived") {
926: # We received the challenge, now we
927: # are echoing it back. This is a no-op,
928: # we're waiting for the state to change
929:
930: if($Socket->Writable() != 0) {
931:
932: $Watcher->cancel();
933: KillSocket($Socket);
934: return;
935: }
936:
937: } elsif ($State eq "ChallengeReplied") {
938: # The echo was sent back, so we switch
939: # to watching readability.
940:
941: $Watcher->cb(\&LondReadable);
942: $Watcher->poll("r");
943:
944: } elsif ($State eq "RequestingKey") {
945: # At this time we're requesting the key.
946: # again, this is essentially a no-op.
947: # we'll write the next chunk until the
948: # state changes.
949:
950: if($Socket->Writable() != 0) {
951: # Write resulted in an error.
952:
953: $Watcher->cancel();
954: KillSocket($Socket);
955: return;
956:
957: }
958: } elsif ($State eq "ReceivingKey") {
959: # Now we need to wait for the key
960: # to come back from the peer:
961:
962: $Watcher->cb(\&LondReadable);
963: $Watcher->poll("r");
964:
965: } elsif ($State eq "SendingRequest") {
966: # At this time we are sending a request to the
967: # peer... write the next chunk:
968:
969: if($Socket->Writable() != 0) {
970:
971: if(exists($ActiveTransactions{$Socket})) {
972: Debug(3, "Lond connection lost, failing transactions");
973: FailTransaction($ActiveTransactions{$Socket});
974: }
975: $Watcher->cancel();
976: KillSocket($Socket);
977: return;
978:
979: }
980:
981: } elsif ($State eq "ReceivingReply") {
982: # The send has completed. Wait for the
983: # data to come in for a reply.
984: Debug(8,"Writable sent request/receiving reply");
985: $Watcher->cb(\&LondReadable);
986: $Watcher->poll("r");
987:
988: } else {
989: # Control only passes here on an error:
990: # the socket state does not match any
991: # of the known states... so an error
992: # must be logged.
993:
994: &Debug(4, "Invalid socket state ".$State."\n");
995: }
996:
997: }
998: =pod
999:
1000: =cut
1001: sub QueueDelayed {
1002: Debug(3,"QueueDelayed called");
1003:
1004: my $path = "$perlvar{'lonSockDir'}/delayed";
1005:
1006: Debug(4, "Delayed path: ".$path);
1007: opendir(DIRHANDLE, $path);
1008:
1009: my @alldelayed = grep /\.$RemoteHost$/, readdir DIRHANDLE;
1010: closedir(DIRHANDLE);
1011: my $dfname;
1012: my $reqfile;
1013: foreach $dfname (sort @alldelayed) {
1014: $reqfile = "$path/$dfname";
1015: Debug(4, "queueing ".$reqfile);
1016: my $Handle = IO::File->new($reqfile);
1017: my $cmd = <$Handle>;
1018: chomp $cmd; # There may or may not be a newline...
1019: $cmd = $cmd."\n"; # now for sure there's exactly one newline.
1020: my $Transaction = LondTransaction->new($cmd);
1021: $Transaction->SetDeferred($reqfile);
1022: QueueTransaction($Transaction);
1023: }
1024:
1025: }
1026:
1027: =pod
1028:
1029: =head2 MakeLondConnection
1030:
1031: Create a new lond connection object, and start it towards its initial
1032: idleness. Once idle, it becomes elligible to receive transactions
1033: from the work queue. If the work queue is not empty when the
1034: connection is completed and becomes idle, it will dequeue an entry and
1035: start off on it.
1036:
1037: =cut
1038:
1039: sub MakeLondConnection {
1040: Debug(4,"MakeLondConnection to ".GetServerHost()." on port "
1041: .GetServerPort());
1042:
1043: my $Connection = LondConnection->new(&GetServerHost(),
1044: &GetServerPort());
1045:
1046: if($Connection eq undef) { # Needs to be more robust later.
1047: Log("CRITICAL","Failed to make a connection with lond.");
1048: $ConnectionRetriesLeft--;
1049: return 0; # Failure.
1050: } else {
1051:
1052: # The connection needs to have writability
1053: # monitored in order to send the init sequence
1054: # that starts the whole authentication/key
1055: # exchange underway.
1056: #
1057: my $Socket = $Connection->GetSocket();
1058: if($Socket eq undef) {
1059: die "did not get a socket from the connection";
1060: } else {
1061: &Debug(9,"MakeLondConnection got socket: ".$Socket);
1062: }
1063:
1064: $Connection->SetTimeoutCallback(\&SocketTimeout);
1065:
1066: my $event = Event->io(fd => $Socket,
1067: poll => 'w',
1068: cb => \&LondWritable,
1069: data => $Connection,
1070: desc => 'Connection to lond server');
1071: $ActiveConnections{$Connection} = $event;
1072:
1073: $ConnectionCount++;
1074: Debug(4, "Connection count = ".$ConnectionCount);
1075: if($ConnectionCount == 1) { # First Connection:
1076: QueueDelayed;
1077: }
1078: Log("SUCESS", "Created connection ".$ConnectionCount
1079: ." to host ".GetServerHost());
1080: return 1; # Return success.
1081: }
1082:
1083: }
1084:
1085: =pod
1086:
1087: =head2 StartRequest
1088:
1089: Starts a lond request going on a specified lond connection.
1090: parameters are:
1091:
1092: =item $Lond
1093:
1094: Connection to the lond that will send the transaction and receive the
1095: reply.
1096:
1097: =item $Client
1098:
1099: Connection to the client that is making this request We got the
1100: request from this socket, and when the request has been relayed to
1101: lond and we get a reply back from lond it will get sent to this
1102: socket.
1103:
1104: =item $Request
1105:
1106: The text of the request to send.
1107:
1108: =cut
1109:
1110: sub StartRequest {
1111: my $Lond = shift;
1112: my $Request = shift; # This is a LondTransaction.
1113:
1114: Debug(6, "StartRequest: ".$Request->getRequest());
1115:
1116: my $Socket = $Lond->GetSocket();
1117:
1118: $Request->Activate($Lond);
1119: $ActiveTransactions{$Lond} = $Request;
1120:
1121: $Lond->InitiateTransaction($Request->getRequest());
1122: my $event = Event->io(fd => $Socket,
1123: poll => "w",
1124: cb => \&LondWritable,
1125: data => $Lond,
1126: desc => "lond transaction connection");
1127: $ActiveConnections{$Lond} = $event;
1128: Debug(8," Start Request made watcher data with ".$event->data."\n");
1129: }
1130:
1131: =pod
1132:
1133: =head2 QueueTransaction
1134:
1135: If there is an idle lond connection, it is put to work doing this
1136: transaction. Otherwise, the transaction is placed in the work queue.
1137: If placed in the work queue and the maximum number of connections has
1138: not yet been created, a new connection will be started. Our goal is
1139: to eventually have a sufficient number of connections that the work
1140: queue will typically be empty. parameters are:
1141:
1142: =item Socket
1143:
1144: open on the lonc client.
1145:
1146: =item Request
1147:
1148: data to send to the lond.
1149:
1150: =cut
1151:
1152: sub QueueTransaction {
1153:
1154: my $requestData = shift; # This is a LondTransaction.
1155: my $cmd = $requestData->getRequest();
1156:
1157: Debug(6,"QueueTransaction: ".$cmd);
1158:
1159: my $LondSocket = $IdleConnections->pop();
1160: if(!defined $LondSocket) { # Need to queue request.
1161: Debug(5,"Must queue...");
1162: $WorkQueue->enqueue($requestData);
1163: if($ConnectionCount < $MaxConnectionCount) {
1164: if($ConnectionRetriesLeft > 0) {
1165: Debug(5,"Starting additional lond connection");
1166: if(MakeLondConnection() == 0) {
1167: EmptyQueue(); # Fail transactions, can't make connection.
1168: }
1169: } else {
1170: ShowStatus(GetServerHost()." >>> DEAD !!!! <<<");
1171: EmptyQueue(); # It's worse than that ... he's dead Jim.
1172: }
1173: }
1174: } else { # Can start the request:
1175: Debug(8,"Can start...");
1176: StartRequest($LondSocket, $requestData);
1177: }
1178: }
1179:
1180: #-------------------------- Lonc UNIX socket handling ---------------------
1181:
1182: =pod
1183:
1184: =head2 ClientRequest
1185: Callback that is called when data can be read from the UNIX domain
1186: socket connecting us with an apache server process.
1187:
1188: =cut
1189:
1190: sub ClientRequest {
1191: Debug(6, "ClientRequest");
1192: my $event = shift;
1193: my $watcher = $event->w;
1194: my $socket = $watcher->fd;
1195: my $data = $watcher->data;
1196: my $thisread;
1197:
1198: Debug(9, " Watcher named: ".$watcher->desc);
1199:
1200: my $rv = $socket->recv($thisread, POSIX::BUFSIZ, 0);
1201: Debug(8, "rcv: data length = ".length($thisread)
1202: ." read =".$thisread);
1203: unless (defined $rv && length($thisread)) {
1204: # Likely eof on socket.
1205: Debug(5,"Client Socket closed on lonc for ".$RemoteHost);
1206: close($socket);
1207: $watcher->cancel();
1208: delete($ActiveClients{$socket});
1209: return;
1210: }
1211: Debug(8,"Data: ".$data." this read: ".$thisread);
1212: $data = $data.$thisread; # Append new data.
1213: $watcher->data($data);
1214: if($data =~ /(.*\n)/) { # Request entirely read.
1215: if($data eq "close_connection_exit\n") {
1216: Log("CRITICAL",
1217: "Request Close Connection ... exiting");
1218: CloseAllLondConnections();
1219: exit;
1220: }
1221: Debug(8, "Complete transaction received: ".$data);
1222: my $Transaction = LondTransaction->new($data);
1223: $Transaction->SetClient($socket);
1224: QueueTransaction($Transaction);
1225: $watcher->cancel(); # Done looking for input data.
1226: }
1227:
1228: }
1229:
1230:
1231: =pod
1232:
1233: =head2 NewClient
1234:
1235: Callback that is called when a connection is received on the unix
1236: socket for a new client of lonc. The callback is parameterized by the
1237: event.. which is a-priori assumed to be an io event, and therefore has
1238: an fd member that is the Listener socket. We Accept the connection
1239: and register a new event on the readability of that socket:
1240:
1241: =cut
1242:
1243: sub NewClient {
1244: Debug(6, "NewClient");
1245: my $event = shift; # Get the event parameters.
1246: my $watcher = $event->w;
1247: my $socket = $watcher->fd; # Get the event' socket.
1248: my $connection = $socket->accept(); # Accept the client connection.
1249: Debug(8,"Connection request accepted from "
1250: .GetPeername($connection, AF_UNIX));
1251:
1252:
1253: my $description = sprintf("Connection to lonc client %d",
1254: $ClientConnection);
1255: Debug(9, "Creating event named: ".$description);
1256: Event->io(cb => \&ClientRequest,
1257: poll => 'r',
1258: desc => $description,
1259: data => "",
1260: fd => $connection);
1261: $ActiveClients{$connection} = $ClientConnection;
1262: $ClientConnection++;
1263: }
1264:
1265: =pod
1266:
1267: =head2 GetLoncSocketPath
1268:
1269: Returns the name of the UNIX socket on which to listen for client
1270: connections.
1271:
1272: =cut
1273:
1274: sub GetLoncSocketPath {
1275: return $UnixSocketDir."/".GetServerHost();
1276: }
1277:
1278: =pod
1279:
1280: =head2 GetServerHost
1281:
1282: Returns the host whose lond we talk with.
1283:
1284: =cut
1285:
1286: sub GetServerHost {
1287: return $RemoteHost; # Setup by the fork.
1288: }
1289:
1290: =pod
1291:
1292: =head2 GetServerPort
1293:
1294: Returns the lond port number.
1295:
1296: =cut
1297:
1298: sub GetServerPort {
1299: return $perlvar{londPort};
1300: }
1301:
1302: =pod
1303:
1304: =head2 SetupLoncListener
1305:
1306: Setup a lonc listener event. The event is called when the socket
1307: becomes readable.. that corresponds to the receipt of a new
1308: connection. The event handler established will accept the connection
1309: (creating a communcations channel), that int turn will establish
1310: another event handler to subess requests.
1311:
1312: =cut
1313:
1314: sub SetupLoncListener {
1315:
1316: my $socket;
1317: my $SocketName = GetLoncSocketPath();
1318: unlink($SocketName);
1319: unless ($socket =IO::Socket::UNIX->new(Local => $SocketName,
1320: Listen => 10,
1321: Type => SOCK_STREAM)) {
1322: die "Failed to create a lonc listner socket";
1323: }
1324: Event->io(cb => \&NewClient,
1325: poll => 'r',
1326: desc => 'Lonc listener Unix Socket',
1327: fd => $socket);
1328: }
1329:
1330: =pod
1331:
1332: =head2 ChildStatus
1333:
1334: Child USR1 signal handler to report the most recent status
1335: into the status file.
1336:
1337: We also use this to reset the retries count in order to allow the
1338: client to retry connections with a previously dead server.
1339: =cut
1340: sub ChildStatus {
1341: my $event = shift;
1342: my $watcher = $event->w;
1343:
1344: Debug(2, "Reporting child status because : ".$watcher->data);
1345: my $docdir = $perlvar{'lonDocRoot'};
1346: my $fh = IO::File->new(">>$docdir/lon-status/loncstatus.txt");
1347: print $fh $$."\t".$RemoteHost."\t".$Status."\t".
1348: $RecentLogEntry."\n";
1349: $ConnectionRetriesLeft = $ConnectionRetries;
1350: }
1351:
1352: =pod
1353:
1354: =head2 SignalledToDeath
1355:
1356: Called in response to a signal that causes a chid process to die.
1357:
1358: =cut
1359:
1360:
1361: sub SignalledToDeath {
1362: my $event = shift;
1363: my $watcher= $event->w;
1364:
1365: Debug(2,"Signalled to death! via ".$watcher->data);
1366: my ($signal) = $watcher->data;
1367: chomp($signal);
1368: Log("CRITICAL", "Abnormal exit. Child $$ for $RemoteHost "
1369: ."died through "."\"$signal\"");
1370: LogPerm("F:lonc: $$ on $RemoteHost signalled to death: "
1371: ."\"$signal\"");
1372: exit 0;
1373:
1374: }
1375:
1376: =head2 ToggleDebug
1377:
1378: This sub toggles trace debugging on and off.
1379:
1380: =cut
1381:
1382: sub ToggleDebug {
1383: my $Current = $DebugLevel;
1384: $DebugLevel = $NextDebugLevel;
1385: $NextDebugLevel = $Current;
1386:
1387: Log("SUCCESS", "New debugging level for $RemoteHost now $DebugLevel");
1388:
1389: }
1390:
1391: =head2 ChildProcess
1392:
1393: This sub implements a child process for a single lonc daemon.
1394:
1395: =cut
1396:
1397: sub ChildProcess {
1398:
1399:
1400: #
1401: # Signals must be handled by the Event framework...
1402: #
1403:
1404: Event->signal(signal => "QUIT",
1405: cb => \&SignalledToDeath,
1406: data => "QUIT");
1407: Event->signal(signal => "HUP",
1408: cb => \&ChildStatus,
1409: data => "HUP");
1410: Event->signal(signal => "USR1",
1411: cb => \&ChildStatus,
1412: data => "USR1");
1413: Event->signal(signal => "INT",
1414: cb => \&ToggleDebug,
1415: data => "INT");
1416:
1417: SetupTimer();
1418:
1419: SetupLoncListener();
1420:
1421: $Event::Debuglevel = $DebugLevel;
1422:
1423: Debug(9, "Making initial lond connection for ".$RemoteHost);
1424:
1425: # Setup the initial server connection:
1426:
1427: # &MakeLondConnection(); // let first work requirest do it.
1428:
1429:
1430: Debug(9,"Entering event loop");
1431: my $ret = Event::loop(); # Start the main event loop.
1432:
1433:
1434: die "Main event loop exited!!!";
1435: }
1436:
1437: # Create a new child for host passed in:
1438:
1439: sub CreateChild {
1440: my $sigset = POSIX::SigSet->new(SIGINT);
1441: sigprocmask(SIG_BLOCK, $sigset);
1442: my $host = shift;
1443: $RemoteHost = $host;
1444: Log("CRITICAL", "Forking server for ".$host);
1445: my $pid = fork;
1446: if($pid) { # Parent
1447: $RemoteHost = "Parent";
1448: $ChildHash{$pid} = $host;
1449: $HostToPid{$host}= $pid;
1450: sigprocmask(SIG_UNBLOCK, $sigset);
1451:
1452: } else { # child.
1453: ShowStatus("Connected to ".$RemoteHost);
1454: $SIG{INT} = 'DEFAULT';
1455: sigprocmask(SIG_UNBLOCK, $sigset);
1456: ChildProcess; # Does not return.
1457: }
1458:
1459: }
1460: #
1461: # Parent process logic pass 1:
1462: # For each entry in the hosts table, we will
1463: # fork off an instance of ChildProcess to service the transactions
1464: # to that host. Each pid will be entered in a global hash
1465: # with the value of the key, the host.
1466: # The parent will then enter a loop to wait for process exits.
1467: # Each exit gets logged and the child gets restarted.
1468: #
1469:
1470: #
1471: # Fork and start in new session so hang-up isn't going to
1472: # happen without intent.
1473: #
1474:
1475:
1476:
1477:
1478:
1479:
1480: ShowStatus("Forming new session");
1481: my $childpid = fork;
1482: if ($childpid != 0) {
1483: sleep 4; # Give child a chacne to break to
1484: exit 0; # a new sesion.
1485: }
1486: #
1487: # Write my pid into the pid file so I can be located
1488: #
1489:
1490: ShowStatus("Parent writing pid file:");
1491: my $execdir = $perlvar{'lonDaemons'};
1492: open (PIDSAVE, ">$execdir/logs/lonc.pid");
1493: print PIDSAVE "$$\n";
1494: close(PIDSAVE);
1495:
1496:
1497:
1498: if (POSIX::setsid() < 0) {
1499: print "Could not create new session\n";
1500: exit -1;
1501: }
1502:
1503: ShowStatus("Forking node servers");
1504:
1505: Log("CRITICAL", "--------------- Starting children ---------------");
1506:
1507: LondConnection::ReadConfig; # Read standard config files.
1508: my $HostIterator = LondConnection::GetHostIterator;
1509: while (! $HostIterator->end()) {
1510:
1511: my $hostentryref = $HostIterator->get();
1512: CreateChild($hostentryref->[0]);
1513: $HostHash{$hostentryref->[0]} = $hostentryref->[4];
1514: $HostIterator->next();
1515: }
1516: $RemoteHost = "Parent Server";
1517:
1518: # Maintain the population:
1519:
1520: ShowStatus("Parent keeping the flock");
1521:
1522: #
1523: # Set up parent signals:
1524: #
1525:
1526: $SIG{INT} = \&Terminate;
1527: $SIG{TERM} = \&Terminate;
1528: $SIG{HUP} = \&Restart;
1529: $SIG{USR1} = \&CheckKids;
1530: $SIG{USR2} = \&UpdateKids; # LonManage update request.
1531:
1532: while(1) {
1533: my $deadchild = wait();
1534: if(exists $ChildHash{$deadchild}) { # need to restart.
1535: my $deadhost = $ChildHash{$deadchild};
1536: delete($HostToPid{$deadhost});
1537: delete($ChildHash{$deadchild});
1538: Log("WARNING","Lost child pid= ".$deadchild.
1539: "Connected to host ".$deadhost);
1540: Log("INFO", "Restarting child procesing ".$deadhost);
1541: CreateChild($deadhost);
1542: }
1543: }
1544:
1545:
1546:
1547: =pod
1548:
1549: =head1 CheckKids
1550:
1551: Since kids do not die as easily in this implementation
1552: as the previous one, there is no need to restart the
1553: dead ones (all dead kids get restarted when they die!!)
1554: The only thing this function does is to pass USR1 to the
1555: kids so that they report their status.
1556:
1557: =cut
1558:
1559: sub CheckKids {
1560: Debug(2, "Checking status of children");
1561: my $docdir = $perlvar{'lonDocRoot'};
1562: my $fh = IO::File->new(">$docdir/lon-status/loncstatus.txt");
1563: my $now=time;
1564: my $local=localtime($now);
1565: print $fh "LONC status $local - parent $$ \n\n";
1566: foreach my $pid (keys %ChildHash) {
1567: Debug(2, "Sending USR1 -> $pid");
1568: kill 'USR1' => $pid; # Tell Child to report status.
1569: sleep 1; # Wait so file doesn't intermix.
1570: }
1571: }
1572:
1573: =pod
1574:
1575: =head1 UpdateKids
1576:
1577: parent's SIGUSR2 handler. This handler:
1578:
1579: =item
1580:
1581: Rereads the hosts file.
1582:
1583: =item
1584:
1585: Kills off (via sigint) children for hosts that have disappeared.
1586:
1587: =item
1588:
1589: QUITs children for hosts that already exist (this just forces a status display
1590: and resets the connection retry count for that host.
1591:
1592: =item
1593:
1594: Starts new children for hosts that have been added to the hosts.tab file since
1595: the start of the master program and maintains them.
1596:
1597: =cut
1598:
1599: sub UpdateKids {
1600:
1601: Log("INFO", "Updating connections via SIGUSR2");
1602:
1603: # Just in case we need to kill our own lonc, we wait a few seconds to
1604: # give it a chance to receive and relay lond's response to the
1605: # re-init command.
1606: #
1607:
1608: sleep(2); # Wait a couple of seconds.
1609:
1610: my %hosts; # Indexed by loncapa hostname, value=ip.
1611:
1612: # Need to re-read the host table:
1613:
1614:
1615: LondConnection::ReadConfig();
1616: my $I = LondConnection::GetHostIterator;
1617: while (! $I->end()) {
1618: my $item = $I->get();
1619: $hosts{$item->[0]} = $item->[4];
1620: $I->next();
1621: }
1622:
1623: # The logic below is written for clarity not for efficiency.
1624: # Since I anticipate that this function is only rarely called, that's
1625: # appropriate. There are certainly ways to combine the loops below,
1626: # and anyone wishing to obscure the logic is welcome to go for it.
1627: # Note that we don't re-direct sigchild. Instead we do what's needed
1628: # to the data structures that keep track of children to ensure that
1629: # when sigchild is honored, no new child is born.
1630: #
1631:
1632: # For each existing child; if it's host doesn't exist, kill the child.
1633:
1634: foreach my $child (keys %ChildHash) {
1635: my $oldhost = $ChildHash{$child};
1636: if (!(exists $hosts{$oldhost})) {
1637: Log("CRITICAL", "Killing child for $oldhost host no longer exists");
1638: delete $ChildHash{$child};
1639: delete $HostToPid{$oldhost};
1640: kill 'QUIT' => $child;
1641: }
1642: }
1643: # For each remaining existing child; if it's host's ip has changed,
1644: # Restart the child on the new IP.
1645:
1646: foreach my $child (keys %ChildHash) {
1647: my $oldhost = $ChildHash{$child};
1648: my $oldip = $HostHash{$oldhost};
1649: if ($hosts{$oldhost} ne $oldip) {
1650:
1651: # kill the old child.
1652:
1653: Log("CRITICAL", "Killing child for $oldhost host ip has changed...");
1654: delete $ChildHash{$child};
1655: delete $HostToPid{$oldhost};
1656: kill 'QUIT' => $child;
1657:
1658: # Do the book-keeping needed to start a new child on the
1659: # new ip.
1660:
1661: $HostHash{$oldhost} = $hosts{$oldhost};
1662: CreateChild($oldhost);
1663: }
1664: }
1665: # Finally, for each new host, not in the host hash, create a
1666: # enter the host and create a new child.
1667: # Force a status display of any existing process.
1668:
1669: foreach my $host (keys %hosts) {
1670: if(!(exists $HostHash{$host})) {
1671: Log("INFO", "New host $host discovered in hosts.tab...");
1672: $HostHash{$host} = $hosts{$host};
1673: CreateChild($host);
1674: } else {
1675: kill 'HUP' => $HostToPid{$host}; # status display.
1676: }
1677: }
1678: }
1679:
1680:
1681: =pod
1682:
1683: =head1 Restart
1684:
1685: Signal handler for HUP... all children are killed and
1686: we self restart. This is an el-cheapo way to re read
1687: the config file.
1688:
1689: =cut
1690:
1691: sub Restart {
1692: &KillThemAll; # First kill all the children.
1693: Log("CRITICAL", "Restarting");
1694: my $execdir = $perlvar{'lonDaemons'};
1695: unlink("$execdir/logs/lonc.pid");
1696: exec("$execdir/loncnew");
1697: }
1698:
1699: =pod
1700:
1701: =head1 KillThemAll
1702:
1703: Signal handler that kills all children by sending them a
1704: SIGHUP. Responds to sigint and sigterm.
1705:
1706: =cut
1707:
1708: sub KillThemAll {
1709: Debug(2, "Kill them all!!");
1710: local($SIG{CHLD}) = 'IGNORE'; # Our children >will< die.
1711: foreach my $pid (keys %ChildHash) {
1712: my $serving = $ChildHash{$pid};
1713: Debug(2, "Killing lonc for $serving pid = $pid");
1714: ShowStatus("Killing lonc for $serving pid = $pid");
1715: Log("CRITICAL", "Killing lonc for $serving pid = $pid");
1716: kill 'QUIT' => $pid;
1717: delete($ChildHash{$pid});
1718: }
1719: my $execdir = $perlvar{'lonDaemons'};
1720: unlink("$execdir/logs/lonc.pid");
1721:
1722: }
1723:
1724: =pod
1725:
1726: =head1 Terminate
1727:
1728: Terminate the system.
1729:
1730: =cut
1731:
1732: sub Terminate {
1733: KillThemAll;
1734: Log("CRITICAL","Master process exiting");
1735: exit 0;
1736:
1737: }
1738: =pod
1739:
1740: =head1 Theory
1741:
1742: The event class is used to build this as a single process with an
1743: event driven model. The following events are handled:
1744:
1745: =item UNIX Socket connection Received
1746:
1747: =item Request data arrives on UNIX data transfer socket.
1748:
1749: =item lond connection becomes writable.
1750:
1751: =item timer fires at 1 second intervals.
1752:
1753: All sockets are run in non-blocking mode. Timeouts managed by the timer
1754: handler prevents hung connections.
1755:
1756: Key data structures:
1757:
1758: =item RequestQueue
1759:
1760: A queue of requests received from UNIX sockets that are
1761: waiting for a chance to be forwarded on a lond connection socket.
1762:
1763: =item ActiveConnections
1764:
1765: A hash of lond connections that have transactions in process that are
1766: available to be timed out.
1767:
1768: =item ActiveTransactions
1769:
1770: A hash indexed by lond connections that contain the client reply
1771: socket for each connection that has an active transaction on it.
1772:
1773: =item IdleConnections
1774:
1775: A hash of lond connections that have no work to do. These connections
1776: can be closed if they are idle for a long enough time.
1777:
1778: =cut
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>