1: #!/usr/bin/perl
2: # The LearningOnline Network with CAPA
3: # lonc maintains the connections to remote computers
4: #
5: # $Id: loncnew,v 1.30 2003/10/27 10:09:21 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.
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: # Change log:
47: # $Log: loncnew,v $
48: # Revision 1.30 2003/10/27 10:09:21 foxr
49: # Tighten up a few compares to eq and flip a few debug levels around... nothing
50: # critical
51: #
52: # Revision 1.29 2003/10/21 14:24:42 foxr
53: # Fix little typo that may explain growth of connections
54: #
55: # Revision 1.28 2003/10/14 15:36:21 albertel
56: # - making it easier to run loncnew,
57: # /etc/init.d/loncontrol startnew
58: # /etc/init.d/loncontrol restartnew
59: # will now start loncnew in place of lonc
60: #
61: # Revision 1.27 2003/10/07 11:23:03 foxr
62: # Installed and tested code to process reinit in parent server.
63: #
64: # Revision 1.26 2003/09/30 11:11:17 foxr
65: # Add book-keeping hashes to support the re-init procedure.
66: #
67: # Revision 1.25 2003/09/23 11:22:14 foxr
68: # Tested ability to receive sigusr2 This is now logged and must be
69: # properly implemented as a re-read of hosts and re-init of appropriate
70: # children.
71: #
72: # Revision 1.24 2003/09/16 09:46:42 foxr
73: # Added skeletal infrastructure to support SIGUSR2 update hosts request.
74: #
75: # Revision 1.23 2003/09/15 09:24:49 foxr
76: # Add use strict and fix all the fallout from that.
77: #
78: # Revision 1.22 2003/09/02 10:34:47 foxr
79: # - Fix errors in host dead detection logic (too many cases where the
80: # retries left were not getting incremented or just not checked).
81: # - Added some additional status to the ps axuww display:
82: # o Remaining retries on a host.
83: # o >>> DEAD <<< indicator if I've given up on a host.
84: # - Tested the SIGHUP will reset the retries remaining count (thanks to
85: # the above status stuff, and get allow the loncnew to re-try again
86: # on the host (thanks to the log).
87: #
88: # Revision 1.21 2003/08/26 09:19:51 foxr
89: # How embarrassing... put in the SocketTimeout function in loncnew and forgot
90: # to actually hook it into the LondTransaction. Added this to MakeLondConnection
91: # where it belongs... hopefully transactions (not just connection attempts) will
92: # timeout more speedily than the socket errors will catch it.
93: #
94: # Revision 1.20 2003/08/25 18:48:11 albertel
95: # - fixing a forgotten ;
96: #
97: # Revision 1.19 2003/08/19 09:31:46 foxr
98: # Get socket directory from configuration rather than the old hard coded test
99: # way that I forgot to un-hard code.
100: #
101: # Revision 1.18 2003/08/06 09:52:29 foxr
102: # Also needed to remember to fail in-flight transactions if their sends fail.
103: #
104: # Revision 1.17 2003/08/03 00:44:31 foxr
105: # 1. Correct handling of connection failure: Assume it means the host is
106: # unreachable and fail all of the queued transactions. Note that the
107: # inflight transactions should fail on their own time due either to timeout
108: # or send/receive failures.
109: # 2. Correct handling of logs for forced death signals. Pull the signal
110: # from the event watcher.
111: #
112: # Revision 1.16 2003/07/29 02:33:05 foxr
113: # Add SIGINT processing to child processes to toggle annoying trace mode
114: # on/off.. will try to use this to isolate the compute boud process issue.
115: #
116: # Revision 1.15 2003/07/15 02:07:05 foxr
117: # Added code for lonc/lond transaction timeouts. Who knows if it works right.
118: # The intent is for a timeout to fail any transaction in progress and kill
119: # off the sockt that timed out.
120: #
121: # Revision 1.14 2003/07/03 02:10:18 foxr
122: # Get all of the signals to work correctly.
123: #
124: # Revision 1.13 2003/07/02 01:31:55 foxr
125: # Added kill -HUP logic (restart).
126: #
127: # Revision 1.11 2003/06/25 01:54:44 foxr
128: # Fix more problems with transaction failure.
129: #
130: # Revision 1.10 2003/06/24 02:46:04 foxr
131: # Put a limit on the number of times we'll retry a connection.
132: # Start getting the signal stuff put in as well...note that need to get signals
133: # going or else the client will permanently give up on dead servers.
134: #
135: # Revision 1.9 2003/06/13 02:38:43 foxr
136: # Add logging in 'expected format'
137: #
138: # Revision 1.8 2003/06/11 02:04:35 foxr
139: # Support delayed transactions... this is done uniformly by encapsulating
140: # transactions in an object ... a LondTransaction that is implemented by
141: # LondTransaction.pm
142: #
143: # Revision 1.7 2003/06/03 01:59:39 foxr
144: # complete coding to support deferred transactions.
145: #
146: #
147: use strict;
148: use lib "/home/httpd/lib/perl/";
149: use lib "/home/foxr/newloncapa/types";
150: use Event qw(:DEFAULT );
151: use POSIX qw(:signal_h);
152: use POSIX;
153: use IO::Socket;
154: use IO::Socket::INET;
155: use IO::Socket::UNIX;
156: use IO::File;
157: use IO::Handle;
158: use Socket;
159: use Crypt::IDEA;
160: use LONCAPA::Queue;
161: use LONCAPA::Stack;
162: use LONCAPA::LondConnection;
163: use LONCAPA::LondTransaction;
164: use LONCAPA::Configuration;
165: use LONCAPA::HashIterator;
166:
167:
168: #
169: # Disable all signals we might receive from outside for now.
170: #
171:
172:
173: # Read the httpd configuration file to get perl variables
174: # normally set in apache modules:
175:
176: my $perlvarref = LONCAPA::Configuration::read_conf('loncapa.conf');
177: my %perlvar = %{$perlvarref};
178:
179: #
180: # parent and shared variables.
181:
182: my %ChildHash; # by pid -> host.
183: my %HostToPid; # By host -> pid.
184: my %HostHash; # by loncapaname -> IP.
185:
186:
187: my $MaxConnectionCount = 10; # Will get from config later.
188: my $ClientConnection = 0; # Uniquifier for client events.
189:
190: my $DebugLevel = 0;
191: my $NextDebugLevel= 2; # So Sigint can toggle this.
192: my $IdleTimeout= 3600; # Wait an hour before pruning connections.
193:
194: #
195: # The variables below are only used by the child processes.
196: #
197: my $RemoteHost; # Name of host child is talking to.
198: my $UnixSocketDir= $perlvar{'lonSockDir'};
199: my $IdleConnections = Stack->new(); # Set of idle connections
200: my %ActiveConnections; # Connections to the remote lond.
201: my %ActiveTransactions; # LondTransactions in flight.
202: my %ActiveClients; # Serial numbers of active clients by socket.
203: my $WorkQueue = Queue->new(); # Queue of pending transactions.
204: my $ConnectionCount = 0;
205: my $IdleSeconds = 0; # Number of seconds idle.
206: my $Status = ""; # Current status string.
207: my $RecentLogEntry = "";
208: my $ConnectionRetries=2; # Number of connection retries allowed.
209: my $ConnectionRetriesLeft=2; # Number of connection retries remaining.
210:
211: #
212: # The hash below gives the HTML format for log messages
213: # given a severity.
214: #
215: my %LogFormats;
216:
217: $LogFormats{"CRITICAL"} = "<font color=red>CRITICAL: %s</font>";
218: $LogFormats{"SUCCESS"} = "<font color=green>SUCCESS: %s</font>";
219: $LogFormats{"INFO"} = "<font color=yellow>INFO: %s</font>";
220: $LogFormats{"WARNING"} = "<font color=blue>WARNING: %s</font>";
221: $LogFormats{"DEFAULT"} = " %s ";
222:
223:
224:
225: =pod
226:
227: =head2 LogPerm
228:
229: Makes an entry into the permanent log file.
230:
231: =cut
232: sub LogPerm {
233: my $message=shift;
234: my $execdir=$perlvar{'lonDaemons'};
235: my $now=time;
236: my $local=localtime($now);
237: my $fh=IO::File->new(">>$execdir/logs/lonnet.perm.log");
238: print $fh "$now:$message:$local\n";
239: }
240:
241: =pod
242:
243: =head2 Log
244:
245: Logs a message to the log file.
246: Parameters:
247:
248: =item severity
249:
250: One of CRITICAL, WARNING, INFO, SUCCESS used to select the
251: format string used to format the message. if the severity is
252: not a defined severity the Default format string is used.
253:
254: =item message
255:
256: The base message. In addtion to the format string, the message
257: will be appended to a string containing the name of our remote
258: host and the time will be formatted into the message.
259:
260: =cut
261:
262: sub Log {
263: my $severity = shift;
264: my $message = shift;
265:
266: if(!$LogFormats{$severity}) {
267: $severity = "DEFAULT";
268: }
269:
270: my $format = $LogFormats{$severity};
271:
272: # Put the window dressing in in front of the message format:
273:
274: my $now = time;
275: my $local = localtime($now);
276: my $finalformat = "$local ($$) [$RemoteHost] [$Status] ";
277: my $finalformat = $finalformat.$format."\n";
278:
279: # open the file and put the result.
280:
281: my $execdir = $perlvar{'lonDaemons'};
282: my $fh = IO::File->new(">>$execdir/logs/lonc.log");
283: my $msg = sprintf($finalformat, $message);
284: $RecentLogEntry = $msg;
285: print $fh $msg;
286:
287:
288: }
289:
290:
291: =pod
292:
293: =head2 GetPeerName
294:
295: Returns the name of the host that a socket object is connected to.
296:
297: =cut
298:
299: sub GetPeername {
300: my $connection = shift;
301: my $AdrFamily = shift;
302: my $peer = $connection->peername();
303: my $peerport;
304: my $peerip;
305: if($AdrFamily == AF_INET) {
306: ($peerport, $peerip) = sockaddr_in($peer);
307: my $peername = gethostbyaddr($peerip, $AdrFamily);
308: return $peername;
309: } elsif ($AdrFamily == AF_UNIX) {
310: my $peerfile;
311: ($peerfile) = sockaddr_un($peer);
312: return $peerfile;
313: }
314: }
315: #----------------------------- Timer management ------------------------
316: =pod
317:
318: =head2 Debug
319:
320: Invoked to issue a debug message.
321:
322: =cut
323:
324: sub Debug {
325: my $level = shift;
326: my $message = shift;
327: if ($level <= $DebugLevel) {
328: Log("INFO", "-Debug- $message host = $RemoteHost");
329: }
330: }
331:
332: sub SocketDump {
333: my $level = shift;
334: my $socket= shift;
335: if($level <= $DebugLevel) {
336: $socket->Dump();
337: }
338: }
339:
340: =pod
341:
342: =head2 ShowStatus
343:
344: Place some text as our pid status.
345: and as what we return in a SIGUSR1
346:
347: =cut
348: sub ShowStatus {
349: my $state = shift;
350: my $now = time;
351: my $local = localtime($now);
352: $Status = $local.": ".$state;
353: $0='lonc: '.$state.' '.$local;
354: }
355:
356: =pod
357:
358: =head 2 SocketTimeout
359:
360: Called when an action on the socket times out. The socket is
361: destroyed and any active transaction is failed.
362:
363:
364: =cut
365: sub SocketTimeout {
366: my $Socket = shift;
367:
368: KillSocket($Socket); # A transaction timeout also counts as
369: # a connection failure:
370: $ConnectionRetriesLeft--;
371: }
372:
373: =pod
374:
375: =head2 Tick
376:
377: Invoked each timer tick.
378:
379: =cut
380:
381:
382: sub Tick {
383: my $client;
384: if($ConnectionRetriesLeft > 0) {
385: ShowStatus(GetServerHost()." Connection count: ".$ConnectionCount
386: ." Retries remaining: ".$ConnectionRetriesLeft);
387: } else {
388: ShowStatus(GetServerHost()." >> DEAD <<");
389: }
390: # Is it time to prune connection count:
391:
392:
393: if($IdleConnections->Count() &&
394: ($WorkQueue->Count() == 0)) { # Idle connections and nothing to do?
395: $IdleSeconds++;
396: if($IdleSeconds > $IdleTimeout) { # Prune a connection...
397: my $Socket = $IdleConnections->pop();
398: KillSocket($Socket);
399: }
400: } else {
401: $IdleSeconds = 0; # Reset idle count if not idle.
402: }
403: #
404: # For each inflight transaction, tick down its timeout counter.
405: #
406: foreach my $item (keys %ActiveTransactions) {
407: my $Socket = $ActiveTransactions{$item}->getServer();
408: $Socket->Tick();
409: }
410: # Do we have work in the queue, but no connections to service them?
411: # If so, try to make some new connections to get things going again.
412: #
413:
414: my $Requests = $WorkQueue->Count();
415: if (($ConnectionCount == 0) && ($Requests > 0)) {
416: if ($ConnectionRetriesLeft > 0) {
417: my $Connections = ($Requests <= $MaxConnectionCount) ?
418: $Requests : $MaxConnectionCount;
419: Debug(5,"Work but no connections, start ".$Connections." of them");
420: my $successCount = 0;
421: for (my $i =0; $i < $Connections; $i++) {
422: $successCount += MakeLondConnection();
423: }
424: if($successCount == 0) { # All connections failed:
425: Debug(5,"Work in queue failed to make any connectiouns\n");
426: EmptyQueue(); # Fail pending transactions with con_lost.
427: }
428: } else {
429: ShowStatus(GetServerHost()." >>> DEAD!!! <<<");
430: Debug(5,"Work in queue, but gave up on connections..flushing\n");
431: EmptyQueue(); # Connections can't be established.
432: }
433:
434: }
435: }
436:
437: =pod
438:
439: =head2 SetupTimer
440:
441: Sets up a 1 per sec recurring timer event. The event handler is used to:
442:
443: =item
444:
445: Trigger timeouts on communications along active sockets.
446:
447: =item
448:
449: Trigger disconnections of idle sockets.
450:
451: =cut
452:
453: sub SetupTimer {
454: Debug(6, "SetupTimer");
455: Event->timer(interval => 1, debug => 1, cb => \&Tick );
456: }
457:
458: =pod
459:
460: =head2 ServerToIdle
461:
462: This function is called when a connection to the server is
463: ready for more work.
464:
465: If there is work in the Work queue the top element is dequeued
466: and the connection will start to work on it. If the work queue is
467: empty, the connection is pushed on the idle connection stack where
468: it will either get another work unit, or alternatively, if it sits there
469: long enough, it will be shut down and released.
470:
471: =cut
472:
473: sub ServerToIdle {
474: my $Socket = shift; # Get the socket.
475: delete($ActiveTransactions{$Socket}); # Server has no transaction
476:
477: &Debug(5, "Server to idle");
478:
479: # If there's work to do, start the transaction:
480:
481: my $reqdata = $WorkQueue->dequeue(); # This is a LondTransaction
482: if ($reqdata ne undef) {
483: Debug(5, "Queue gave request data: ".$reqdata->getRequest());
484: &StartRequest($Socket, $reqdata);
485:
486: } else {
487:
488: # There's no work waiting, so push the server to idle list.
489: &Debug(5, "No new work requests, server connection going idle");
490: $IdleConnections->push($Socket);
491: }
492: }
493:
494: =pod
495:
496: =head2 ClientWritable
497:
498: Event callback for when a client socket is writable.
499:
500: This callback is established when a transaction reponse is
501: avaiable from lond. The response is forwarded to the unix socket
502: as it becomes writable in this sub.
503:
504: Parameters:
505:
506: =item Event
507:
508: The event that has been triggered. Event->w->data is
509: the data and Event->w->fd is the socket to write.
510:
511: =cut
512:
513: sub ClientWritable {
514: my $Event = shift;
515: my $Watcher = $Event->w;
516: my $Data = $Watcher->data;
517: my $Socket = $Watcher->fd;
518:
519: # Try to send the data:
520:
521: &Debug(6, "ClientWritable writing".$Data);
522: &Debug(9, "Socket is: ".$Socket);
523:
524: if($Socket->connected) {
525: my $result = $Socket->send($Data, 0);
526:
527: # $result undefined: the write failed.
528: # otherwise $result is the number of bytes written.
529: # Remove that preceding string from the data.
530: # If the resulting data is empty, destroy the watcher
531: # and set up a read event handler to accept the next
532: # request.
533:
534: &Debug(9,"Send result is ".$result." Defined: ".defined($result));
535: if($result ne undef) {
536: &Debug(9, "send result was defined");
537: if($result == length($Data)) { # Entire string sent.
538: &Debug(9, "ClientWritable data all written");
539: $Watcher->cancel();
540: #
541: # Set up to read next request from socket:
542:
543: my $descr = sprintf("Connection to lonc client %d",
544: $ActiveClients{$Socket});
545: Event->io(cb => \&ClientRequest,
546: poll => 'r',
547: desc => $descr,
548: data => "",
549: fd => $Socket);
550:
551: } else { # Partial string sent.
552: $Watcher->data(substr($Data, $result));
553: if($result == 0) { # client hung up on us!!
554: Log("INFO", "lonc pipe client hung up on us!");
555: $Watcher->cancel;
556: $Socket->shutdown(2);
557: $Socket->close();
558: }
559: }
560:
561: } else { # Error of some sort...
562:
563: # Some errnos are possible:
564: my $errno = $!;
565: if($errno == POSIX::EWOULDBLOCK ||
566: $errno == POSIX::EAGAIN ||
567: $errno == POSIX::EINTR) {
568: # No action taken?
569: } else { # Unanticipated errno.
570: &Debug(5,"ClientWritable error or peer shutdown".$RemoteHost);
571: $Watcher->cancel; # Stop the watcher.
572: $Socket->shutdown(2); # Kill connection
573: $Socket->close(); # Close the socket.
574: }
575:
576: }
577: } else {
578: $Watcher->cancel(); # A delayed request...just cancel.
579: }
580: }
581:
582: =pod
583:
584: =head2 CompleteTransaction
585:
586: Called when the reply data has been received for a lond
587: transaction. The reply data must now be sent to the
588: ultimate client on the other end of the Unix socket. This is
589: done by setting up a writable event for the socket with the
590: data the reply data.
591:
592: Parameters:
593:
594: =item Socket
595:
596: Socket on which the lond transaction occured. This is a
597: LondConnection. The data received is in the TransactionReply member.
598:
599: =item Transaction
600:
601: The transaction that is being completed.
602:
603: =cut
604:
605: sub CompleteTransaction {
606: &Debug(5,"Complete transaction");
607: my $Socket = shift;
608: my $Transaction = shift;
609:
610: if (!$Transaction->isDeferred()) { # Normal transaction
611: my $data = $Socket->GetReply(); # Data to send.
612: StartClientReply($Transaction, $data);
613: } else { # Delete deferred transaction file.
614: Log("SUCCESS", "A delayed transaction was completed");
615: LogPerm("S:$Transaction->getClient() :".$Transaction->getRequest());
616: unlink $Transaction->getFile();
617: }
618: }
619: =pod
620: =head1 StartClientReply
621:
622: Initiates a reply to a client where the reply data is a parameter.
623:
624: =head2 parameters:
625:
626: =item Transaction
627:
628: The transaction for which we are responding to the client.
629:
630: =item data
631:
632: The data to send to apached client.
633:
634: =cut
635: sub StartClientReply {
636: my $Transaction = shift;
637: my $data = shift;
638:
639:
640: my $Client = $Transaction->getClient();
641:
642: &Debug(8," Reply was: ".$data);
643: my $Serial = $ActiveClients{$Client};
644: my $desc = sprintf("Connection to lonc client %d",
645: $Serial);
646: Event->io(fd => $Client,
647: poll => "w",
648: desc => $desc,
649: cb => \&ClientWritable,
650: data => $data);
651: }
652: =pod
653: =head2 FailTransaction
654:
655: Finishes a transaction with failure because the associated lond socket
656: disconnected. There are two possibilities:
657: - The transaction is deferred: in which case we just quietly
658: delete the transaction since there is no client connection.
659: - The transaction is 'live' in which case we initiate the sending
660: of "con_lost" to the client.
661:
662: Deleting the transaction means killing it from the
663: %ActiveTransactions hash.
664:
665: Parameters:
666:
667: =item client
668:
669: The LondTransaction we are failing.
670:
671: =cut
672:
673: sub FailTransaction {
674: my $transaction = shift;
675: Log("WARNING", "Failing transaction ".$transaction->getRequest());
676: Debug(1, "Failing transaction: ".$transaction->getRequest());
677: if (!$transaction->isDeferred()) { # If the transaction is deferred we'll get to it.
678: my $client = $transaction->getClient();
679: Debug(1," Replying con_lost to ".$transaction->getRequest());
680: StartClientReply($transaction, "con_lost\n");
681: }
682: if($ConnectionRetriesLeft <= 0) {
683: Log("CRITICAL", "Host marked dead: ".GetServerHost());
684: }
685:
686: }
687:
688: =pod
689: =head1 EmptyQueue
690:
691: Fails all items in the work queue with con_lost.
692: Note that each item in the work queue is a transaction.
693:
694: =cut
695: sub EmptyQueue {
696: $ConnectionRetriesLeft--; # Counts as connection failure too.
697: while($WorkQueue->Count()) {
698: my $request = $WorkQueue->dequeue(); # This is a transaction
699: FailTransaction($request);
700: }
701: }
702:
703: =pod
704:
705: =head2 CloseAllLondConnections
706:
707: Close all connections open on lond prior to exit e.g.
708:
709: =cut
710: sub CloseAllLondConnections {
711: foreach my $Socket (keys %ActiveConnections) {
712: KillSocket($Socket);
713: }
714: }
715: =cut
716:
717: =pod
718:
719: =head2 KillSocket
720:
721: Destroys a socket. This function can be called either when a socket
722: has died of 'natural' causes or because a socket needs to be pruned due to
723: idleness. If the socket has died naturally, if there are no longer any
724: live connections a new connection is created (in case there are transactions
725: in the queue). If the socket has been pruned, it is never re-created.
726:
727: Parameters:
728:
729: =item Socket
730:
731: The socket to kill off.
732:
733: =item Restart
734:
735: nonzero if we are allowed to create a new connection.
736:
737:
738: =cut
739: sub KillSocket {
740: my $Socket = shift;
741:
742: Log("WARNING", "Shutting down a socket");
743: $Socket->Shutdown();
744:
745: # If the socket came from the active connection set,
746: # delete its transaction... note that FailTransaction should
747: # already have been called!!!
748: # otherwise it came from the idle set.
749: #
750:
751: if(exists($ActiveTransactions{$Socket})) {
752: delete ($ActiveTransactions{$Socket});
753: }
754: if(exists($ActiveConnections{$Socket})) {
755: delete($ActiveConnections{$Socket});
756: }
757: $ConnectionCount--;
758:
759: # If the connection count has gone to zero and there is work in the
760: # work queue, the work all gets failed with con_lost.
761: #
762: if($ConnectionCount == 0) {
763: EmptyQueue();
764: }
765: }
766:
767: =pod
768:
769: =head2 LondReadable
770:
771: This function is called whenever a lond connection
772: is readable. The action is state dependent:
773:
774: =head3 State=Initialized
775:
776: We''re waiting for the challenge, this is a no-op until the
777: state changes.
778:
779: =head3 State=Challenged
780:
781: The challenge has arrived we need to transition to Writable.
782: The connection must echo the challenge back.
783:
784: =head3 State=ChallengeReplied
785:
786: The challenge has been replied to. The we are receiveing the
787: 'ok' from the partner.
788:
789: =head3 State=RequestingKey
790:
791: The ok has been received and we need to send the request for
792: an encryption key. Transition to writable for that.
793:
794: =head3 State=ReceivingKey
795:
796: The the key has been requested, now we are reading the new key.
797:
798: =head3 State=Idle
799:
800: The encryption key has been negotiated or we have finished
801: reading data from the a transaction. If the callback data has
802: a client as well as the socket iformation, then we are
803: doing a transaction and the data received is relayed to the client
804: before the socket is put on the idle list.
805:
806: =head3 State=SendingRequest
807:
808: I do not think this state can be received here, but if it is,
809: the appropriate thing to do is to transition to writable, and send
810: the request.
811:
812: =head3 State=ReceivingReply
813:
814: We finished sending the request to the server and now transition
815: to readable to receive the reply.
816:
817: The parameter to this function are:
818:
819: The event. Implicit in this is the watcher and its data. The data
820: contains at least the lond connection object and, if a
821: transaction is in progress, the socket attached to the local client.
822:
823: =cut
824:
825: sub LondReadable {
826:
827: my $Event = shift;
828: my $Watcher = $Event->w;
829: my $Socket = $Watcher->data;
830: my $client = undef;
831:
832: &Debug(6,"LondReadable called state = ".$Socket->GetState());
833:
834:
835: my $State = $Socket->GetState(); # All action depends on the state.
836:
837: SocketDump(6, $Socket);
838: my $status = $Socket->Readable();
839:
840: &Debug(2, "Socket->Readable returned: $status");
841:
842: if($status != 0) {
843: # bad return from socket read. Currently this means that
844: # The socket has become disconnected. We fail the transaction.
845:
846: Log("WARNING",
847: "Lond connection lost.");
848: if(exists($ActiveTransactions{$Socket})) {
849: FailTransaction($ActiveTransactions{$Socket});
850: }
851: $Watcher->cancel();
852: KillSocket($Socket);
853: $ConnectionRetriesLeft--; # Counts as connection failure
854: return;
855: }
856: SocketDump(6,$Socket);
857:
858: $State = $Socket->GetState(); # Update in case of transition.
859: &Debug(6, "After read, state is ".$State);
860:
861: if($State eq "Initialized") {
862:
863:
864: } elsif ($State eq "ChallengeReceived") {
865: # The challenge must be echoed back; The state machine
866: # in the connection takes care of setting that up. Just
867: # need to transition to writable:
868:
869: $Watcher->cb(\&LondWritable);
870: $Watcher->poll("w");
871:
872: } elsif ($State eq "ChallengeReplied") {
873:
874:
875: } elsif ($State eq "RequestingKey") {
876: # The ok was received. Now we need to request the key
877: # That requires us to be writable:
878:
879: $Watcher->cb(\&LondWritable);
880: $Watcher->poll("w");
881:
882: } elsif ($State eq "ReceivingKey") {
883:
884: } elsif ($State eq "Idle") {
885: # If necessary, complete a transaction and then go into the
886: # idle queue.
887: # Note that a trasition to idle indicates a live lond
888: # on the other end so reset the connection retries.
889: #
890: $ConnectionRetriesLeft = $ConnectionRetries; # success resets the count
891: $Watcher->cancel();
892: if(exists($ActiveTransactions{$Socket})) {
893: Debug(5,"Completing transaction!!");
894: CompleteTransaction($Socket,
895: $ActiveTransactions{$Socket});
896: } else {
897: Log("SUCCESS", "Connection ".$ConnectionCount." to "
898: .$RemoteHost." now ready for action");
899: }
900: ServerToIdle($Socket); # Next work unit or idle.
901:
902: } elsif ($State eq "SendingRequest") {
903: # We need to be writable for this and probably don't belong
904: # here inthe first place.
905:
906: Deubg(6, "SendingRequest state encountered in readable");
907: $Watcher->poll("w");
908: $Watcher->cb(\&LondWritable);
909:
910: } elsif ($State eq "ReceivingReply") {
911:
912:
913: } else {
914: # Invalid state.
915: Debug(4, "Invalid state in LondReadable");
916: }
917: }
918:
919: =pod
920:
921: =head2 LondWritable
922:
923: This function is called whenever a lond connection
924: becomes writable while there is a writeable monitoring
925: event. The action taken is very state dependent:
926:
927: =head3 State = Connected
928:
929: The connection is in the process of sending the 'init' hailing to the
930: lond on the remote end. The connection object''s Writable member is
931: called. On error, ConnectionError is called to destroy the connection
932: and remove it from the ActiveConnections hash
933:
934: =head3 Initialized
935:
936: 'init' has been sent, writability monitoring is removed and
937: readability monitoring is started with LondReadable as the callback.
938:
939: =head3 ChallengeReceived
940:
941: The connection has received the who are you challenge from the remote
942: system, and is in the process of sending the challenge
943: response. Writable is called.
944:
945: =head3 ChallengeReplied
946:
947: The connection has replied to the initial challenge The we switch to
948: monitoring readability looking for the server to reply with 'ok'.
949:
950: =head3 RequestingKey
951:
952: The connection is in the process of requesting its encryption key.
953: Writable is called.
954:
955: =head3 ReceivingKey
956:
957: The connection has sent the request for a key. Switch to readability
958: monitoring to accept the key
959:
960: =head3 SendingRequest
961:
962: The connection is in the process of sending a request to the server.
963: This request is part of a client transaction. All the states until
964: now represent the client setup protocol. Writable is called.
965:
966: =head3 ReceivingReply
967:
968: The connection has sent a request. Now it must receive a reply.
969: Readability monitoring is requested.
970:
971: This function is an event handler and therefore receives as
972: a parameter the event that has fired. The data for the watcher
973: of this event is a reference to a list of one or two elements,
974: depending on state. The first (and possibly only) element is the
975: socket. The second (present only if a request is in progress)
976: is the socket on which to return a reply to the caller.
977:
978: =cut
979:
980: sub LondWritable {
981: my $Event = shift;
982: my $Watcher = $Event->w;
983: my $Socket = $Watcher->data;
984: my $State = $Socket->GetState();
985:
986: Debug(6,"LondWritable State = ".$State."\n");
987:
988:
989: # Figure out what to do depending on the state of the socket:
990:
991:
992:
993:
994: SocketDump(6,$Socket);
995:
996: if ($State eq "Connected") {
997:
998: if ($Socket->Writable() != 0) {
999: # The write resulted in an error.
1000: # We'll treat this as if the socket got disconnected:
1001: Log("WARNING", "Connection to ".$RemoteHost.
1002: " has been disconnected");
1003: FailTransaction($ActiveTransactions{$Socket});
1004: $Watcher->cancel();
1005: KillSocket($Socket);
1006: return;
1007: }
1008: # "init" is being sent...
1009:
1010:
1011: } elsif ($State eq "Initialized") {
1012:
1013: # Now that init was sent, we switch
1014: # to watching for readability:
1015:
1016: $Watcher->cb(\&LondReadable);
1017: $Watcher->poll("r");
1018:
1019: } elsif ($State eq "ChallengeReceived") {
1020: # We received the challenge, now we
1021: # are echoing it back. This is a no-op,
1022: # we're waiting for the state to change
1023:
1024: if($Socket->Writable() != 0) {
1025:
1026: $Watcher->cancel();
1027: KillSocket($Socket);
1028: return;
1029: }
1030:
1031: } elsif ($State eq "ChallengeReplied") {
1032: # The echo was sent back, so we switch
1033: # to watching readability.
1034:
1035: $Watcher->cb(\&LondReadable);
1036: $Watcher->poll("r");
1037:
1038: } elsif ($State eq "RequestingKey") {
1039: # At this time we're requesting the key.
1040: # again, this is essentially a no-op.
1041: # we'll write the next chunk until the
1042: # state changes.
1043:
1044: if($Socket->Writable() != 0) {
1045: # Write resulted in an error.
1046:
1047: $Watcher->cancel();
1048: KillSocket($Socket);
1049: return;
1050:
1051: }
1052: } elsif ($State eq "ReceivingKey") {
1053: # Now we need to wait for the key
1054: # to come back from the peer:
1055:
1056: $Watcher->cb(\&LondReadable);
1057: $Watcher->poll("r");
1058:
1059: } elsif ($State eq "SendingRequest") {
1060: # At this time we are sending a request to the
1061: # peer... write the next chunk:
1062:
1063: if($Socket->Writable() != 0) {
1064:
1065: if(exists($ActiveTransactions{$Socket})) {
1066: Debug(3, "Lond connection lost, failing transactions");
1067: FailTransaction($ActiveTransactions{$Socket});
1068: }
1069: $Watcher->cancel();
1070: KillSocket($Socket);
1071: return;
1072:
1073: }
1074:
1075: } elsif ($State eq "ReceivingReply") {
1076: # The send has completed. Wait for the
1077: # data to come in for a reply.
1078: Debug(8,"Writable sent request/receiving reply");
1079: $Watcher->cb(\&LondReadable);
1080: $Watcher->poll("r");
1081:
1082: } else {
1083: # Control only passes here on an error:
1084: # the socket state does not match any
1085: # of the known states... so an error
1086: # must be logged.
1087:
1088: &Debug(4, "Invalid socket state ".$State."\n");
1089: }
1090:
1091: }
1092: =pod
1093:
1094: =cut
1095: sub QueueDelayed {
1096: Debug(3,"QueueDelayed called");
1097:
1098: my $path = "$perlvar{'lonSockDir'}/delayed";
1099:
1100: Debug(4, "Delayed path: ".$path);
1101: opendir(DIRHANDLE, $path);
1102:
1103: my @alldelayed = grep /\.$RemoteHost$/, readdir DIRHANDLE;
1104: closedir(DIRHANDLE);
1105: my $dfname;
1106: my $reqfile;
1107: foreach $dfname (sort @alldelayed) {
1108: $reqfile = "$path/$dfname";
1109: Debug(4, "queueing ".$reqfile);
1110: my $Handle = IO::File->new($reqfile);
1111: my $cmd = <$Handle>;
1112: chomp $cmd; # There may or may not be a newline...
1113: $cmd = $cmd."\n"; # now for sure there's exactly one newline.
1114: my $Transaction = LondTransaction->new($cmd);
1115: $Transaction->SetDeferred($reqfile);
1116: QueueTransaction($Transaction);
1117: }
1118:
1119: }
1120:
1121: =pod
1122:
1123: =head2 MakeLondConnection
1124:
1125: Create a new lond connection object, and start it towards its initial
1126: idleness. Once idle, it becomes elligible to receive transactions
1127: from the work queue. If the work queue is not empty when the
1128: connection is completed and becomes idle, it will dequeue an entry and
1129: start off on it.
1130:
1131: =cut
1132:
1133: sub MakeLondConnection {
1134: Debug(4,"MakeLondConnection to ".GetServerHost()." on port "
1135: .GetServerPort());
1136:
1137: my $Connection = LondConnection->new(&GetServerHost(),
1138: &GetServerPort());
1139:
1140: if($Connection eq undef) { # Needs to be more robust later.
1141: Log("CRITICAL","Failed to make a connection with lond.");
1142: $ConnectionRetriesLeft--;
1143: return 0; # Failure.
1144: } else {
1145:
1146: # The connection needs to have writability
1147: # monitored in order to send the init sequence
1148: # that starts the whole authentication/key
1149: # exchange underway.
1150: #
1151: my $Socket = $Connection->GetSocket();
1152: if($Socket eq undef) {
1153: die "did not get a socket from the connection";
1154: } else {
1155: &Debug(9,"MakeLondConnection got socket: ".$Socket);
1156: }
1157:
1158: $Connection->SetTimeoutCallback(\&SocketTimeout);
1159:
1160: my $event = Event->io(fd => $Socket,
1161: poll => 'w',
1162: cb => \&LondWritable,
1163: data => $Connection,
1164: desc => 'Connection to lond server');
1165: $ActiveConnections{$Connection} = $event;
1166:
1167: $ConnectionCount++;
1168: Debug(4, "Connection count = ".$ConnectionCount);
1169: if($ConnectionCount == 1) { # First Connection:
1170: QueueDelayed;
1171: }
1172: Log("SUCESS", "Created connection ".$ConnectionCount
1173: ." to host ".GetServerHost());
1174: return 1; # Return success.
1175: }
1176:
1177: }
1178:
1179: =pod
1180:
1181: =head2 StartRequest
1182:
1183: Starts a lond request going on a specified lond connection.
1184: parameters are:
1185:
1186: =item $Lond
1187:
1188: Connection to the lond that will send the transaction and receive the
1189: reply.
1190:
1191: =item $Client
1192:
1193: Connection to the client that is making this request We got the
1194: request from this socket, and when the request has been relayed to
1195: lond and we get a reply back from lond it will get sent to this
1196: socket.
1197:
1198: =item $Request
1199:
1200: The text of the request to send.
1201:
1202: =cut
1203:
1204: sub StartRequest {
1205: my $Lond = shift;
1206: my $Request = shift; # This is a LondTransaction.
1207:
1208: Debug(6, "StartRequest: ".$Request->getRequest());
1209:
1210: my $Socket = $Lond->GetSocket();
1211:
1212: $Request->Activate($Lond);
1213: $ActiveTransactions{$Lond} = $Request;
1214:
1215: $Lond->InitiateTransaction($Request->getRequest());
1216: my $event = Event->io(fd => $Socket,
1217: poll => "w",
1218: cb => \&LondWritable,
1219: data => $Lond,
1220: desc => "lond transaction connection");
1221: $ActiveConnections{$Lond} = $event;
1222: Debug(8," Start Request made watcher data with ".$event->data."\n");
1223: }
1224:
1225: =pod
1226:
1227: =head2 QueueTransaction
1228:
1229: If there is an idle lond connection, it is put to work doing this
1230: transaction. Otherwise, the transaction is placed in the work queue.
1231: If placed in the work queue and the maximum number of connections has
1232: not yet been created, a new connection will be started. Our goal is
1233: to eventually have a sufficient number of connections that the work
1234: queue will typically be empty. parameters are:
1235:
1236: =item Socket
1237:
1238: open on the lonc client.
1239:
1240: =item Request
1241:
1242: data to send to the lond.
1243:
1244: =cut
1245:
1246: sub QueueTransaction {
1247:
1248: my $requestData = shift; # This is a LondTransaction.
1249: my $cmd = $requestData->getRequest();
1250:
1251: Debug(6,"QueueTransaction: ".$cmd);
1252:
1253: my $LondSocket = $IdleConnections->pop();
1254: if(!defined $LondSocket) { # Need to queue request.
1255: Debug(5,"Must queue...");
1256: $WorkQueue->enqueue($requestData);
1257: if($ConnectionCount < $MaxConnectionCount) {
1258: if($ConnectionRetriesLeft > 0) {
1259: Debug(5,"Starting additional lond connection");
1260: if(MakeLondConnection() == 0) {
1261: EmptyQueue(); # Fail transactions, can't make connection.
1262: }
1263: } else {
1264: ShowStatus(GetServerHost()." >>> DEAD !!!! <<<");
1265: EmptyQueue(); # It's worse than that ... he's dead Jim.
1266: }
1267: }
1268: } else { # Can start the request:
1269: Debug(8,"Can start...");
1270: StartRequest($LondSocket, $requestData);
1271: }
1272: }
1273:
1274: #-------------------------- Lonc UNIX socket handling ---------------------
1275:
1276: =pod
1277:
1278: =head2 ClientRequest
1279: Callback that is called when data can be read from the UNIX domain
1280: socket connecting us with an apache server process.
1281:
1282: =cut
1283:
1284: sub ClientRequest {
1285: Debug(6, "ClientRequest");
1286: my $event = shift;
1287: my $watcher = $event->w;
1288: my $socket = $watcher->fd;
1289: my $data = $watcher->data;
1290: my $thisread;
1291:
1292: Debug(9, " Watcher named: ".$watcher->desc);
1293:
1294: my $rv = $socket->recv($thisread, POSIX::BUFSIZ, 0);
1295: Debug(8, "rcv: data length = ".length($thisread)
1296: ." read =".$thisread);
1297: unless (defined $rv && length($thisread)) {
1298: # Likely eof on socket.
1299: Debug(5,"Client Socket closed on lonc for ".$RemoteHost);
1300: close($socket);
1301: $watcher->cancel();
1302: delete($ActiveClients{$socket});
1303: return;
1304: }
1305: Debug(8,"Data: ".$data." this read: ".$thisread);
1306: $data = $data.$thisread; # Append new data.
1307: $watcher->data($data);
1308: if($data =~ /(.*\n)/) { # Request entirely read.
1309: if($data eq "close_connection_exit\n") {
1310: Log("CRITICAL",
1311: "Request Close Connection ... exiting");
1312: CloseAllLondConnections();
1313: exit;
1314: }
1315: Debug(8, "Complete transaction received: ".$data);
1316: my $Transaction = LondTransaction->new($data);
1317: $Transaction->SetClient($socket);
1318: QueueTransaction($Transaction);
1319: $watcher->cancel(); # Done looking for input data.
1320: }
1321:
1322: }
1323:
1324:
1325: =pod
1326:
1327: =head2 NewClient
1328:
1329: Callback that is called when a connection is received on the unix
1330: socket for a new client of lonc. The callback is parameterized by the
1331: event.. which is a-priori assumed to be an io event, and therefore has
1332: an fd member that is the Listener socket. We Accept the connection
1333: and register a new event on the readability of that socket:
1334:
1335: =cut
1336:
1337: sub NewClient {
1338: Debug(6, "NewClient");
1339: my $event = shift; # Get the event parameters.
1340: my $watcher = $event->w;
1341: my $socket = $watcher->fd; # Get the event' socket.
1342: my $connection = $socket->accept(); # Accept the client connection.
1343: Debug(8,"Connection request accepted from "
1344: .GetPeername($connection, AF_UNIX));
1345:
1346:
1347: my $description = sprintf("Connection to lonc client %d",
1348: $ClientConnection);
1349: Debug(9, "Creating event named: ".$description);
1350: Event->io(cb => \&ClientRequest,
1351: poll => 'r',
1352: desc => $description,
1353: data => "",
1354: fd => $connection);
1355: $ActiveClients{$connection} = $ClientConnection;
1356: $ClientConnection++;
1357: }
1358:
1359: =pod
1360:
1361: =head2 GetLoncSocketPath
1362:
1363: Returns the name of the UNIX socket on which to listen for client
1364: connections.
1365:
1366: =cut
1367:
1368: sub GetLoncSocketPath {
1369: return $UnixSocketDir."/".GetServerHost();
1370: }
1371:
1372: =pod
1373:
1374: =head2 GetServerHost
1375:
1376: Returns the host whose lond we talk with.
1377:
1378: =cut
1379:
1380: sub GetServerHost {
1381: return $RemoteHost; # Setup by the fork.
1382: }
1383:
1384: =pod
1385:
1386: =head2 GetServerPort
1387:
1388: Returns the lond port number.
1389:
1390: =cut
1391:
1392: sub GetServerPort {
1393: return $perlvar{londPort};
1394: }
1395:
1396: =pod
1397:
1398: =head2 SetupLoncListener
1399:
1400: Setup a lonc listener event. The event is called when the socket
1401: becomes readable.. that corresponds to the receipt of a new
1402: connection. The event handler established will accept the connection
1403: (creating a communcations channel), that int turn will establish
1404: another event handler to subess requests.
1405:
1406: =cut
1407:
1408: sub SetupLoncListener {
1409:
1410: my $socket;
1411: my $SocketName = GetLoncSocketPath();
1412: unlink($SocketName);
1413: unless ($socket =IO::Socket::UNIX->new(Local => $SocketName,
1414: Listen => 10,
1415: Type => SOCK_STREAM)) {
1416: die "Failed to create a lonc listner socket";
1417: }
1418: Event->io(cb => \&NewClient,
1419: poll => 'r',
1420: desc => 'Lonc listener Unix Socket',
1421: fd => $socket);
1422: }
1423:
1424: =pod
1425:
1426: =head2 ChildStatus
1427:
1428: Child USR1 signal handler to report the most recent status
1429: into the status file.
1430:
1431: We also use this to reset the retries count in order to allow the
1432: client to retry connections with a previously dead server.
1433: =cut
1434: sub ChildStatus {
1435: my $event = shift;
1436: my $watcher = $event->w;
1437:
1438: Debug(2, "Reporting child status because : ".$watcher->data);
1439: my $docdir = $perlvar{'lonDocRoot'};
1440: my $fh = IO::File->new(">>$docdir/lon-status/loncstatus.txt");
1441: print $fh $$."\t".$RemoteHost."\t".$Status."\t".
1442: $RecentLogEntry."\n";
1443: $ConnectionRetriesLeft = $ConnectionRetries;
1444: }
1445:
1446: =pod
1447:
1448: =head2 SignalledToDeath
1449:
1450: Called in response to a signal that causes a chid process to die.
1451:
1452: =cut
1453:
1454:
1455: sub SignalledToDeath {
1456: my $event = shift;
1457: my $watcher= $event->w;
1458:
1459: Debug(2,"Signalled to death! via ".$watcher->data);
1460: my ($signal) = $watcher->data;
1461: chomp($signal);
1462: Log("CRITICAL", "Abnormal exit. Child $$ for $RemoteHost "
1463: ."died through "."\"$signal\"");
1464: LogPerm("F:lonc: $$ on $RemoteHost signalled to death: "
1465: ."\"$signal\"");
1466: exit 0;
1467:
1468: }
1469:
1470: =head2 ToggleDebug
1471:
1472: This sub toggles trace debugging on and off.
1473:
1474: =cut
1475:
1476: sub ToggleDebug {
1477: my $Current = $DebugLevel;
1478: $DebugLevel = $NextDebugLevel;
1479: $NextDebugLevel = $Current;
1480:
1481: Log("SUCCESS", "New debugging level for $RemoteHost now $DebugLevel");
1482:
1483: }
1484:
1485: =head2 ChildProcess
1486:
1487: This sub implements a child process for a single lonc daemon.
1488:
1489: =cut
1490:
1491: sub ChildProcess {
1492:
1493:
1494: #
1495: # Signals must be handled by the Event framework...
1496: #
1497:
1498: Event->signal(signal => "QUIT",
1499: cb => \&SignalledToDeath,
1500: data => "QUIT");
1501: Event->signal(signal => "HUP",
1502: cb => \&ChildStatus,
1503: data => "HUP");
1504: Event->signal(signal => "USR1",
1505: cb => \&ChildStatus,
1506: data => "USR1");
1507: Event->signal(signal => "INT",
1508: cb => \&ToggleDebug,
1509: data => "INT");
1510:
1511: SetupTimer();
1512:
1513: SetupLoncListener();
1514:
1515: $Event::Debuglevel = $DebugLevel;
1516:
1517: Debug(9, "Making initial lond connection for ".$RemoteHost);
1518:
1519: # Setup the initial server connection:
1520:
1521: # &MakeLondConnection(); // let first work requirest do it.
1522:
1523:
1524: Debug(9,"Entering event loop");
1525: my $ret = Event::loop(); # Start the main event loop.
1526:
1527:
1528: die "Main event loop exited!!!";
1529: }
1530:
1531: # Create a new child for host passed in:
1532:
1533: sub CreateChild {
1534: my $sigset = POSIX::SigSet->new(SIGINT);
1535: sigprocmask(SIG_BLOCK, $sigset);
1536: my $host = shift;
1537: $RemoteHost = $host;
1538: Log("CRITICAL", "Forking server for ".$host);
1539: my $pid = fork;
1540: if($pid) { # Parent
1541: $RemoteHost = "Parent";
1542: $ChildHash{$pid} = $host;
1543: $HostToPid{$host}= $pid;
1544: sigprocmask(SIG_UNBLOCK, $sigset);
1545:
1546: } else { # child.
1547: ShowStatus("Connected to ".$RemoteHost);
1548: $SIG{INT} = 'DEFAULT';
1549: sigprocmask(SIG_UNBLOCK, $sigset);
1550: ChildProcess; # Does not return.
1551: }
1552:
1553: }
1554: #
1555: # Parent process logic pass 1:
1556: # For each entry in the hosts table, we will
1557: # fork off an instance of ChildProcess to service the transactions
1558: # to that host. Each pid will be entered in a global hash
1559: # with the value of the key, the host.
1560: # The parent will then enter a loop to wait for process exits.
1561: # Each exit gets logged and the child gets restarted.
1562: #
1563:
1564: #
1565: # Fork and start in new session so hang-up isn't going to
1566: # happen without intent.
1567: #
1568:
1569:
1570:
1571:
1572:
1573:
1574: ShowStatus("Forming new session");
1575: my $childpid = fork;
1576: if ($childpid != 0) {
1577: sleep 4; # Give child a chacne to break to
1578: exit 0; # a new sesion.
1579: }
1580: #
1581: # Write my pid into the pid file so I can be located
1582: #
1583:
1584: ShowStatus("Parent writing pid file:");
1585: my $execdir = $perlvar{'lonDaemons'};
1586: open (PIDSAVE, ">$execdir/logs/lonc.pid");
1587: print PIDSAVE "$$\n";
1588: close(PIDSAVE);
1589:
1590:
1591:
1592: if (POSIX::setsid() < 0) {
1593: print "Could not create new session\n";
1594: exit -1;
1595: }
1596:
1597: ShowStatus("Forking node servers");
1598:
1599: Log("CRITICAL", "--------------- Starting children ---------------");
1600:
1601: my $HostIterator = LondConnection::GetHostIterator;
1602: while (! $HostIterator->end()) {
1603:
1604: my $hostentryref = $HostIterator->get();
1605: CreateChild($hostentryref->[0]);
1606: $HostHash{$hostentryref->[0]} = $hostentryref->[4];
1607: $HostIterator->next();
1608: }
1609: $RemoteHost = "Parent Server";
1610:
1611: # Maintain the population:
1612:
1613: ShowStatus("Parent keeping the flock");
1614:
1615: #
1616: # Set up parent signals:
1617: #
1618:
1619: $SIG{INT} = \&Terminate;
1620: $SIG{TERM} = \&Terminate;
1621: $SIG{HUP} = \&Restart;
1622: $SIG{USR1} = \&CheckKids;
1623: $SIG{USR2} = \&UpdateKids; # LonManage update request.
1624:
1625: while(1) {
1626: my $deadchild = wait();
1627: if(exists $ChildHash{$deadchild}) { # need to restart.
1628: my $deadhost = $ChildHash{$deadchild};
1629: delete($HostToPid{$deadhost});
1630: delete($ChildHash{$deadchild});
1631: Log("WARNING","Lost child pid= ".$deadchild.
1632: "Connected to host ".$deadhost);
1633: Log("INFO", "Restarting child procesing ".$deadhost);
1634: CreateChild($deadhost);
1635: }
1636: }
1637:
1638:
1639:
1640: =pod
1641:
1642: =head1 CheckKids
1643:
1644: Since kids do not die as easily in this implementation
1645: as the previous one, there is no need to restart the
1646: dead ones (all dead kids get restarted when they die!!)
1647: The only thing this function does is to pass USR1 to the
1648: kids so that they report their status.
1649:
1650: =cut
1651:
1652: sub CheckKids {
1653: Debug(2, "Checking status of children");
1654: my $docdir = $perlvar{'lonDocRoot'};
1655: my $fh = IO::File->new(">$docdir/lon-status/loncstatus.txt");
1656: my $now=time;
1657: my $local=localtime($now);
1658: print $fh "LONC status $local - parent $$ \n\n";
1659: foreach my $pid (keys %ChildHash) {
1660: Debug(2, "Sending USR1 -> $pid");
1661: kill 'USR1' => $pid; # Tell Child to report status.
1662: sleep 1; # Wait so file doesn't intermix.
1663: }
1664: }
1665:
1666: =pod
1667:
1668: =head1 UpdateKids
1669:
1670: parent's SIGUSR2 handler. This handler:
1671:
1672: =item
1673:
1674: Rereads the hosts file.
1675:
1676: =item
1677:
1678: Kills off (via sigint) children for hosts that have disappeared.
1679:
1680: =item
1681:
1682: QUITs children for hosts that already exist (this just forces a status display
1683: and resets the connection retry count for that host.
1684:
1685: =item
1686:
1687: Starts new children for hosts that have been added to the hosts.tab file since
1688: the start of the master program and maintains them.
1689:
1690: =cut
1691:
1692: sub UpdateKids {
1693:
1694: Log("INFO", "Updating connections via SIGUSR2");
1695:
1696: # Just in case we need to kill our own lonc, we wait a few seconds to
1697: # give it a chance to receive and relay lond's response to the
1698: # re-init command.
1699: #
1700:
1701: sleep(2); # Wait a couple of seconds.
1702:
1703: my %hosts; # Indexed by loncapa hostname, value=ip.
1704:
1705: # Need to re-read the host table:
1706:
1707:
1708: LondConnection::ReadConfig();
1709: my $I = LondConnection::GetHostIterator;
1710: while (! $I->end()) {
1711: my $item = $I->get();
1712: $hosts{$item->[0]} = $item->[4];
1713: $I->next();
1714: }
1715:
1716: # The logic below is written for clarity not for efficiency.
1717: # Since I anticipate that this function is only rarely called, that's
1718: # appropriate. There are certainly ways to combine the loops below,
1719: # and anyone wishing to obscure the logic is welcome to go for it.
1720: # Note that we don't re-direct sigchild. Instead we do what's needed
1721: # to the data structures that keep track of children to ensure that
1722: # when sigchild is honored, no new child is born.
1723: #
1724:
1725: # For each existing child; if it's host doesn't exist, kill the child.
1726:
1727: foreach my $child (keys %ChildHash) {
1728: my $oldhost = $ChildHash{$child};
1729: if (!(exists $hosts{$oldhost})) {
1730: Log("CRITICAL", "Killing child for $oldhost host no longer exists");
1731: delete $ChildHash{$child};
1732: delete $HostToPid{$oldhost};
1733: kill 'QUIT' => $child;
1734: }
1735: }
1736: # For each remaining existing child; if it's host's ip has changed,
1737: # Restart the child on the new IP.
1738:
1739: foreach my $child (keys %ChildHash) {
1740: my $oldhost = $ChildHash{$child};
1741: my $oldip = $HostHash{$oldhost};
1742: if ($hosts{$oldhost} ne $oldip) {
1743:
1744: # kill the old child.
1745:
1746: Log("CRITICAL", "Killing child for $oldhost host ip has changed...");
1747: delete $ChildHash{$child};
1748: delete $HostToPid{$oldhost};
1749: kill 'QUIT' => $child;
1750:
1751: # Do the book-keeping needed to start a new child on the
1752: # new ip.
1753:
1754: $HostHash{$oldhost} = $hosts{$oldhost};
1755: CreateChild($oldhost);
1756: }
1757: }
1758: # Finally, for each new host, not in the host hash, create a
1759: # enter the host and create a new child.
1760: # Force a status display of any existing process.
1761:
1762: foreach my $host (keys %hosts) {
1763: if(!(exists $HostHash{$host})) {
1764: Log("INFO", "New host $host discovered in hosts.tab...");
1765: $HostHash{$host} = $hosts{$host};
1766: CreateChild($host);
1767: } else {
1768: kill 'HUP' => $HostToPid{$host}; # status display.
1769: }
1770: }
1771: }
1772:
1773:
1774: =pod
1775:
1776: =head1 Restart
1777:
1778: Signal handler for HUP... all children are killed and
1779: we self restart. This is an el-cheapo way to re read
1780: the config file.
1781:
1782: =cut
1783:
1784: sub Restart {
1785: &KillThemAll; # First kill all the children.
1786: Log("CRITICAL", "Restarting");
1787: my $execdir = $perlvar{'lonDaemons'};
1788: unlink("$execdir/logs/lonc.pid");
1789: exec("$execdir/loncnew");
1790: }
1791:
1792: =pod
1793:
1794: =head1 KillThemAll
1795:
1796: Signal handler that kills all children by sending them a
1797: SIGHUP. Responds to sigint and sigterm.
1798:
1799: =cut
1800:
1801: sub KillThemAll {
1802: Debug(2, "Kill them all!!");
1803: local($SIG{CHLD}) = 'IGNORE'; # Our children >will< die.
1804: foreach my $pid (keys %ChildHash) {
1805: my $serving = $ChildHash{$pid};
1806: Debug(2, "Killing lonc for $serving pid = $pid");
1807: ShowStatus("Killing lonc for $serving pid = $pid");
1808: Log("CRITICAL", "Killing lonc for $serving pid = $pid");
1809: kill 'QUIT' => $pid;
1810: delete($ChildHash{$pid});
1811: }
1812: my $execdir = $perlvar{'lonDaemons'};
1813: unlink("$execdir/logs/lonc.pid");
1814:
1815: }
1816:
1817: =pod
1818:
1819: =head1 Terminate
1820:
1821: Terminate the system.
1822:
1823: =cut
1824:
1825: sub Terminate {
1826: KillThemAll;
1827: Log("CRITICAL","Master process exiting");
1828: exit 0;
1829:
1830: }
1831: =pod
1832:
1833: =head1 Theory
1834:
1835: The event class is used to build this as a single process with an
1836: event driven model. The following events are handled:
1837:
1838: =item UNIX Socket connection Received
1839:
1840: =item Request data arrives on UNIX data transfer socket.
1841:
1842: =item lond connection becomes writable.
1843:
1844: =item timer fires at 1 second intervals.
1845:
1846: All sockets are run in non-blocking mode. Timeouts managed by the timer
1847: handler prevents hung connections.
1848:
1849: Key data structures:
1850:
1851: =item RequestQueue
1852:
1853: A queue of requests received from UNIX sockets that are
1854: waiting for a chance to be forwarded on a lond connection socket.
1855:
1856: =item ActiveConnections
1857:
1858: A hash of lond connections that have transactions in process that are
1859: available to be timed out.
1860:
1861: =item ActiveTransactions
1862:
1863: A hash indexed by lond connections that contain the client reply
1864: socket for each connection that has an active transaction on it.
1865:
1866: =item IdleConnections
1867:
1868: A hash of lond connections that have no work to do. These connections
1869: can be closed if they are idle for a long enough time.
1870:
1871: =cut
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>