File:
[LON-CAPA] /
loncom /
loncnew
Revision
1.42:
download - view:
text,
annotated -
select for diffs
Tue Feb 17 09:43:21 2004 UTC (20 years, 7 months ago) by
foxr
Branches:
MAIN
CVS tags:
HEAD
- Clean up write logic and factor out common code.
- Fix up some error logic to ensure that:
o All failed transactions fromthe lonc point of view close the
associated connection
o All closures of a lond connection will fail any associated transaction
o A bit of defensive programming when closing a connection that should be
idle to ensure that any bad book-keeping gets cleaned up by failing
the associaed transaction.
1: #!/usr/bin/perl
2: # The LearningOnline Network with CAPA
3: # lonc maintains the connections to remote computers
4: #
5: # $Id: loncnew,v 1.42 2004/02/17 09:43:21 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. (done)
39: # - Add Signal handling - HUP restarts. USR1 status report. (done)
40: # - Add Configuration file I/O (done).
41: # - Add management/status request interface. (done)
42: # - Add deferred request capability. (done)
43: # - Detect transmission timeouts. (done)
44: #
45:
46: use strict;
47: use lib "/home/httpd/lib/perl/";
48: use Event qw(:DEFAULT );
49: use POSIX qw(:signal_h);
50: use POSIX;
51: use IO::Socket;
52: use IO::Socket::INET;
53: use IO::Socket::UNIX;
54: use IO::File;
55: use IO::Handle;
56: use Socket;
57: use Crypt::IDEA;
58: use LONCAPA::Queue;
59: use LONCAPA::Stack;
60: use LONCAPA::LondConnection;
61: use LONCAPA::LondTransaction;
62: use LONCAPA::Configuration;
63: use LONCAPA::HashIterator;
64:
65:
66: # Read the httpd configuration file to get perl variables
67: # normally set in apache modules:
68:
69: my $perlvarref = LONCAPA::Configuration::read_conf('loncapa.conf');
70: my %perlvar = %{$perlvarref};
71:
72: #
73: # parent and shared variables.
74:
75: my %ChildHash; # by pid -> host.
76: my %HostToPid; # By host -> pid.
77: my %HostHash; # by loncapaname -> IP.
78:
79:
80: my $MaxConnectionCount = 10; # Will get from config later.
81: my $ClientConnection = 0; # Uniquifier for client events.
82:
83: my $DebugLevel = 0;
84: my $NextDebugLevel= 2; # So Sigint can toggle this.
85: my $IdleTimeout= 3600; # Wait an hour before pruning connections.
86:
87: my $LogTransactions = 0; # When True, all transactions/replies get logged.
88:
89: #
90: # The variables below are only used by the child processes.
91: #
92: my $RemoteHost; # Name of host child is talking to.
93: my $UnixSocketDir= $perlvar{'lonSockDir'};
94: my $IdleConnections = Stack->new(); # Set of idle connections
95: my %ActiveConnections; # Connections to the remote lond.
96: my %ActiveTransactions; # LondTransactions in flight.
97: my %ActiveClients; # Serial numbers of active clients by socket.
98: my $WorkQueue = Queue->new(); # Queue of pending transactions.
99: my $ConnectionCount = 0;
100: my $IdleSeconds = 0; # Number of seconds idle.
101: my $Status = ""; # Current status string.
102: my $RecentLogEntry = "";
103: my $ConnectionRetries=2; # Number of connection retries allowed.
104: my $ConnectionRetriesLeft=2; # Number of connection retries remaining.
105: my $LondVersion = "unknown"; # Version of lond we talk with.
106:
107: #
108: # The hash below gives the HTML format for log messages
109: # given a severity.
110: #
111: my %LogFormats;
112:
113: $LogFormats{"CRITICAL"} = "<font color=red>CRITICAL: %s</font>";
114: $LogFormats{"SUCCESS"} = "<font color=green>SUCCESS: %s</font>";
115: $LogFormats{"INFO"} = "<font color=yellow>INFO: %s</font>";
116: $LogFormats{"WARNING"} = "<font color=blue>WARNING: %s</font>";
117: $LogFormats{"DEFAULT"} = " %s ";
118:
119:
120:
121: =pod
122:
123: =head2 LogPerm
124:
125: Makes an entry into the permanent log file.
126:
127: =cut
128: sub LogPerm {
129: my $message=shift;
130: my $execdir=$perlvar{'lonDaemons'};
131: my $now=time;
132: my $local=localtime($now);
133: my $fh=IO::File->new(">>$execdir/logs/lonnet.perm.log");
134: print $fh "$now:$message:$local\n";
135: }
136:
137: =pod
138:
139: =head2 Log
140:
141: Logs a message to the log file.
142: Parameters:
143:
144: =item severity
145:
146: One of CRITICAL, WARNING, INFO, SUCCESS used to select the
147: format string used to format the message. if the severity is
148: not a defined severity the Default format string is used.
149:
150: =item message
151:
152: The base message. In addtion to the format string, the message
153: will be appended to a string containing the name of our remote
154: host and the time will be formatted into the message.
155:
156: =cut
157:
158: sub Log {
159: my $severity = shift;
160: my $message = shift;
161:
162: if(!$LogFormats{$severity}) {
163: $severity = "DEFAULT";
164: }
165:
166: my $format = $LogFormats{$severity};
167:
168: # Put the window dressing in in front of the message format:
169:
170: my $now = time;
171: my $local = localtime($now);
172: my $finalformat = "$local ($$) [$RemoteHost] [$Status] ";
173: my $finalformat = $finalformat.$format."\n";
174:
175: # open the file and put the result.
176:
177: my $execdir = $perlvar{'lonDaemons'};
178: my $fh = IO::File->new(">>$execdir/logs/lonc.log");
179: my $msg = sprintf($finalformat, $message);
180: $RecentLogEntry = $msg;
181: print $fh $msg;
182:
183:
184: }
185:
186:
187: =pod
188:
189: =head2 GetPeerName
190:
191: Returns the name of the host that a socket object is connected to.
192:
193: =cut
194:
195: sub GetPeername {
196: my $connection = shift;
197: my $AdrFamily = shift;
198: my $peer = $connection->peername();
199: my $peerport;
200: my $peerip;
201: if($AdrFamily == AF_INET) {
202: ($peerport, $peerip) = sockaddr_in($peer);
203: my $peername = gethostbyaddr($peerip, $AdrFamily);
204: return $peername;
205: } elsif ($AdrFamily == AF_UNIX) {
206: my $peerfile;
207: ($peerfile) = sockaddr_un($peer);
208: return $peerfile;
209: }
210: }
211: =pod
212:
213: =head2 Debug
214:
215: Invoked to issue a debug message.
216:
217: =cut
218:
219: sub Debug {
220: my $level = shift;
221: my $message = shift;
222: if ($level <= $DebugLevel) {
223: Log("INFO", "-Debug- $message host = $RemoteHost");
224: }
225: }
226:
227: sub SocketDump {
228: my $level = shift;
229: my $socket= shift;
230: if($level <= $DebugLevel) {
231: $socket->Dump();
232: }
233: }
234:
235: =pod
236:
237: =head2 ShowStatus
238:
239: Place some text as our pid status.
240: and as what we return in a SIGUSR1
241:
242: =cut
243: sub ShowStatus {
244: my $state = shift;
245: my $now = time;
246: my $local = localtime($now);
247: $Status = $local.": ".$state;
248: $0='lonc: '.$state.' '.$local;
249: }
250:
251: =pod
252:
253: =head 2 SocketTimeout
254:
255: Called when an action on the socket times out. The socket is
256: destroyed and any active transaction is failed.
257:
258:
259: =cut
260: sub SocketTimeout {
261: my $Socket = shift;
262: Log("WARNING", "A socket timeout was detected");
263: Debug(0, " SocketTimeout called: ");
264: $Socket->Dump();
265: if(exists($ActiveTransactions{$Socket})) {
266: FailTransaction($ActiveTransactions{$Socket});
267: }
268: KillSocket($Socket); # A transaction timeout also counts as
269: # a connection failure:
270: $ConnectionRetriesLeft--;
271: if($ConnectionRetriesLeft <= 0) {
272: Log("CRITICAL", "Host marked dead: ".GetServerHost());
273: }
274:
275: }
276: #----------------------------- Timer management ------------------------
277:
278: =pod
279:
280: =head2 Tick
281:
282: Invoked each timer tick.
283:
284: =cut
285:
286:
287: sub Tick {
288: my $client;
289: if($ConnectionRetriesLeft > 0) {
290: ShowStatus(GetServerHost()." Connection count: ".$ConnectionCount
291: ." Retries remaining: ".$ConnectionRetriesLeft);
292: } else {
293: ShowStatus(GetServerHost()." >> DEAD <<");
294: }
295: # Is it time to prune connection count:
296:
297:
298: if($IdleConnections->Count() &&
299: ($WorkQueue->Count() == 0)) { # Idle connections and nothing to do?
300: $IdleSeconds++;
301: if($IdleSeconds > $IdleTimeout) { # Prune a connection...
302: my $Socket = $IdleConnections->pop();
303: KillSocket($Socket);
304: }
305: } else {
306: $IdleSeconds = 0; # Reset idle count if not idle.
307: }
308: #
309: # For each inflight transaction, tick down its timeout counter.
310: #
311:
312: foreach my $item (keys %ActiveConnections) {
313: my $State = $ActiveConnections{$item}->data->GetState();
314: if ($State ne 'Idle') {
315: Debug(5,"Ticking Socket $State $item");
316: $ActiveConnections{$item}->data->Tick();
317: }
318: }
319: # Do we have work in the queue, but no connections to service them?
320: # If so, try to make some new connections to get things going again.
321: #
322:
323: my $Requests = $WorkQueue->Count();
324: if (($ConnectionCount == 0) && ($Requests > 0)) {
325: if ($ConnectionRetriesLeft > 0) {
326: my $Connections = ($Requests <= $MaxConnectionCount) ?
327: $Requests : $MaxConnectionCount;
328: Debug(5,"Work but no connections, start ".$Connections." of them");
329: my $successCount = 0;
330: for (my $i =0; $i < $Connections; $i++) {
331: $successCount += MakeLondConnection();
332: }
333: if($successCount == 0) { # All connections failed:
334: Debug(5,"Work in queue failed to make any connectiouns\n");
335: EmptyQueue(); # Fail pending transactions with con_lost.
336: CloseAllLondConnections(); # Should all be closed but....
337: }
338: } else {
339: ShowStatus(GetServerHost()." >>> DEAD!!! <<<");
340: Debug(5,"Work in queue, but gave up on connections..flushing\n");
341: EmptyQueue(); # Connections can't be established.
342: CloseAllLondConnections(); # Should all already be closed but...
343: }
344:
345: }
346: }
347:
348: =pod
349:
350: =head2 SetupTimer
351:
352: Sets up a 1 per sec recurring timer event. The event handler is used to:
353:
354: =item
355:
356: Trigger timeouts on communications along active sockets.
357:
358: =item
359:
360: Trigger disconnections of idle sockets.
361:
362: =cut
363:
364: sub SetupTimer {
365: Debug(6, "SetupTimer");
366: Event->timer(interval => 1, cb => \&Tick );
367: }
368:
369: =pod
370:
371: =head2 ServerToIdle
372:
373: This function is called when a connection to the server is
374: ready for more work.
375:
376: If there is work in the Work queue the top element is dequeued
377: and the connection will start to work on it. If the work queue is
378: empty, the connection is pushed on the idle connection stack where
379: it will either get another work unit, or alternatively, if it sits there
380: long enough, it will be shut down and released.
381:
382: =cut
383:
384: sub ServerToIdle {
385: my $Socket = shift; # Get the socket.
386: delete($ActiveTransactions{$Socket}); # Server has no transaction
387:
388: &Debug(5, "Server to idle");
389:
390: # If there's work to do, start the transaction:
391:
392: my $reqdata = $WorkQueue->dequeue(); # This is a LondTransaction
393: if ($reqdata ne undef) {
394: Debug(5, "Queue gave request data: ".$reqdata->getRequest());
395: &StartRequest($Socket, $reqdata);
396:
397: } else {
398:
399: # There's no work waiting, so push the server to idle list.
400: &Debug(5, "No new work requests, server connection going idle");
401: $IdleConnections->push($Socket);
402: }
403: }
404:
405: =pod
406:
407: =head2 ClientWritable
408:
409: Event callback for when a client socket is writable.
410:
411: This callback is established when a transaction reponse is
412: avaiable from lond. The response is forwarded to the unix socket
413: as it becomes writable in this sub.
414:
415: Parameters:
416:
417: =item Event
418:
419: The event that has been triggered. Event->w->data is
420: the data and Event->w->fd is the socket to write.
421:
422: =cut
423:
424: sub ClientWritable {
425: my $Event = shift;
426: my $Watcher = $Event->w;
427: my $Data = $Watcher->data;
428: my $Socket = $Watcher->fd;
429:
430: # Try to send the data:
431:
432: &Debug(6, "ClientWritable writing".$Data);
433: &Debug(9, "Socket is: ".$Socket);
434:
435: if($Socket->connected) {
436: my $result = $Socket->send($Data, 0);
437:
438: # $result undefined: the write failed.
439: # otherwise $result is the number of bytes written.
440: # Remove that preceding string from the data.
441: # If the resulting data is empty, destroy the watcher
442: # and set up a read event handler to accept the next
443: # request.
444:
445: &Debug(9,"Send result is ".$result." Defined: ".defined($result));
446: if($result ne undef) {
447: &Debug(9, "send result was defined");
448: if($result == length($Data)) { # Entire string sent.
449: &Debug(9, "ClientWritable data all written");
450: $Watcher->cancel();
451: #
452: # Set up to read next request from socket:
453:
454: my $descr = sprintf("Connection to lonc client %d",
455: $ActiveClients{$Socket});
456: Event->io(cb => \&ClientRequest,
457: poll => 'r',
458: desc => $descr,
459: data => "",
460: fd => $Socket);
461:
462: } else { # Partial string sent.
463: $Watcher->data(substr($Data, $result));
464: if($result == 0) { # client hung up on us!!
465: Log("INFO", "lonc pipe client hung up on us!");
466: $Watcher->cancel;
467: $Socket->shutdown(2);
468: $Socket->close();
469: }
470: }
471:
472: } else { # Error of some sort...
473:
474: # Some errnos are possible:
475: my $errno = $!;
476: if($errno == POSIX::EWOULDBLOCK ||
477: $errno == POSIX::EAGAIN ||
478: $errno == POSIX::EINTR) {
479: # No action taken?
480: } else { # Unanticipated errno.
481: &Debug(5,"ClientWritable error or peer shutdown".$RemoteHost);
482: $Watcher->cancel; # Stop the watcher.
483: $Socket->shutdown(2); # Kill connection
484: $Socket->close(); # Close the socket.
485: }
486:
487: }
488: } else {
489: $Watcher->cancel(); # A delayed request...just cancel.
490: }
491: }
492:
493: =pod
494:
495: =head2 CompleteTransaction
496:
497: Called when the reply data has been received for a lond
498: transaction. The reply data must now be sent to the
499: ultimate client on the other end of the Unix socket. This is
500: done by setting up a writable event for the socket with the
501: data the reply data.
502:
503: Parameters:
504:
505: =item Socket
506:
507: Socket on which the lond transaction occured. This is a
508: LondConnection. The data received is in the TransactionReply member.
509:
510: =item Transaction
511:
512: The transaction that is being completed.
513:
514: =cut
515:
516: sub CompleteTransaction {
517: &Debug(5,"Complete transaction");
518: my $Socket = shift;
519: my $Transaction = shift;
520:
521: if (!$Transaction->isDeferred()) { # Normal transaction
522: my $data = $Socket->GetReply(); # Data to send.
523: if($LogTransactions) {
524: Log("SUCCESS", "Reply from lond: '$data'");
525: }
526: StartClientReply($Transaction, $data);
527: } else { # Delete deferred transaction file.
528: Log("SUCCESS", "A delayed transaction was completed");
529: LogPerm("S:$Transaction->getClient() :".$Transaction->getRequest());
530: unlink $Transaction->getFile();
531: }
532: }
533:
534: =pod
535:
536: =head1 StartClientReply
537:
538: Initiates a reply to a client where the reply data is a parameter.
539:
540: =head2 parameters:
541:
542: =item Transaction
543:
544: The transaction for which we are responding to the client.
545:
546: =item data
547:
548: The data to send to apached client.
549:
550: =cut
551:
552: sub StartClientReply {
553: my $Transaction = shift;
554: my $data = shift;
555:
556:
557: my $Client = $Transaction->getClient();
558:
559: &Debug(8," Reply was: ".$data);
560: my $Serial = $ActiveClients{$Client};
561: my $desc = sprintf("Connection to lonc client %d",
562: $Serial);
563: Event->io(fd => $Client,
564: poll => "w",
565: desc => $desc,
566: cb => \&ClientWritable,
567: data => $data);
568: }
569:
570: =pod
571:
572: =head2 FailTransaction
573:
574: Finishes a transaction with failure because the associated lond socket
575: disconnected. There are two possibilities:
576: - The transaction is deferred: in which case we just quietly
577: delete the transaction since there is no client connection.
578: - The transaction is 'live' in which case we initiate the sending
579: of "con_lost" to the client.
580:
581: Deleting the transaction means killing it from the %ActiveTransactions hash.
582:
583: Parameters:
584:
585: =item client
586:
587: The LondTransaction we are failing.
588:
589:
590: =cut
591:
592: sub FailTransaction {
593: my $transaction = shift;
594: Log("WARNING", "Failing transaction ".$transaction->getRequest());
595: Debug(1, "Failing transaction: ".$transaction->getRequest());
596: if (!$transaction->isDeferred()) { # If the transaction is deferred we'll get to it.
597: my $client = $transaction->getClient();
598: Debug(1," Replying con_lost to ".$transaction->getRequest());
599: StartClientReply($transaction, "con_lost\n");
600: }
601:
602: }
603:
604: =pod
605: =head1 EmptyQueue
606:
607: Fails all items in the work queue with con_lost.
608: Note that each item in the work queue is a transaction.
609:
610: =cut
611: sub EmptyQueue {
612: $ConnectionRetriesLeft--; # Counts as connection failure too.
613: while($WorkQueue->Count()) {
614: my $request = $WorkQueue->dequeue(); # This is a transaction
615: FailTransaction($request);
616: }
617: }
618:
619: =pod
620:
621: =head2 CloseAllLondConnections
622:
623: Close all connections open on lond prior to exit e.g.
624:
625: =cut
626: sub CloseAllLondConnections {
627: foreach my $Socket (keys %ActiveConnections) {
628: if(exists($ActiveTransactions{$Socket})) {
629: FailTransaction($ActiveTransactions{$Socket});
630: }
631: KillSocket($Socket);
632: }
633: }
634: =cut
635:
636: =pod
637:
638: =head2 KillSocket
639:
640: Destroys a socket. This function can be called either when a socket
641: has died of 'natural' causes or because a socket needs to be pruned due to
642: idleness. If the socket has died naturally, if there are no longer any
643: live connections a new connection is created (in case there are transactions
644: in the queue). If the socket has been pruned, it is never re-created.
645:
646: Parameters:
647:
648: =item Socket
649:
650: The socket to kill off.
651:
652: =item Restart
653:
654: nonzero if we are allowed to create a new connection.
655:
656:
657: =cut
658: sub KillSocket {
659: my $Socket = shift;
660:
661: Log("WARNING", "Shutting down a socket");
662: $Socket->Shutdown();
663:
664: # If the socket came from the active connection set,
665: # delete its transaction... note that FailTransaction should
666: # already have been called!!!
667: # otherwise it came from the idle set.
668: #
669:
670: if(exists($ActiveTransactions{$Socket})) {
671: delete ($ActiveTransactions{$Socket});
672: }
673: if(exists($ActiveConnections{$Socket})) {
674: delete($ActiveConnections{$Socket});
675: $ConnectionCount--;
676: if ($ConnectionCount < 0) { $ConnectionCount = 0; }
677: }
678: # If the connection count has gone to zero and there is work in the
679: # work queue, the work all gets failed with con_lost.
680: #
681: if($ConnectionCount == 0) {
682: EmptyQueue();
683: CloseAllLondConnections; # Should all already be closed but...
684: }
685: }
686:
687: =pod
688:
689: =head2 LondReadable
690:
691: This function is called whenever a lond connection
692: is readable. The action is state dependent:
693:
694: =head3 State=Initialized
695:
696: We''re waiting for the challenge, this is a no-op until the
697: state changes.
698:
699: =head3 State=Challenged
700:
701: The challenge has arrived we need to transition to Writable.
702: The connection must echo the challenge back.
703:
704: =head3 State=ChallengeReplied
705:
706: The challenge has been replied to. The we are receiveing the
707: 'ok' from the partner.
708:
709: =head3 State=ReadingVersionString
710:
711: We have requested the lond version and are reading the
712: version back. Upon completion, we'll store the version away
713: for future use(?).
714:
715: =head3 State=HostSet
716:
717: We have selected the domain name of our peer (multhomed hosts)
718: and are getting the reply (presumably ok) back.
719:
720: =head3 State=RequestingKey
721:
722: The ok has been received and we need to send the request for
723: an encryption key. Transition to writable for that.
724:
725: =head3 State=ReceivingKey
726:
727: The the key has been requested, now we are reading the new key.
728:
729: =head3 State=Idle
730:
731: The encryption key has been negotiated or we have finished
732: reading data from the a transaction. If the callback data has
733: a client as well as the socket iformation, then we are
734: doing a transaction and the data received is relayed to the client
735: before the socket is put on the idle list.
736:
737: =head3 State=SendingRequest
738:
739: I do not think this state can be received here, but if it is,
740: the appropriate thing to do is to transition to writable, and send
741: the request.
742:
743: =head3 State=ReceivingReply
744:
745: We finished sending the request to the server and now transition
746: to readable to receive the reply.
747:
748: The parameter to this function are:
749:
750: The event. Implicit in this is the watcher and its data. The data
751: contains at least the lond connection object and, if a
752: transaction is in progress, the socket attached to the local client.
753:
754: =cut
755:
756: sub LondReadable {
757:
758: my $Event = shift;
759: my $Watcher = $Event->w;
760: my $Socket = $Watcher->data;
761: my $client = undef;
762:
763: &Debug(6,"LondReadable called state = ".$Socket->GetState());
764:
765:
766: my $State = $Socket->GetState(); # All action depends on the state.
767:
768: SocketDump(6, $Socket);
769: my $status = $Socket->Readable();
770:
771: &Debug(2, "Socket->Readable returned: $status");
772:
773: if($status != 0) {
774: # bad return from socket read. Currently this means that
775: # The socket has become disconnected. We fail the transaction.
776:
777: Log("WARNING",
778: "Lond connection lost.");
779: if(exists($ActiveTransactions{$Socket})) {
780: FailTransaction($ActiveTransactions{$Socket});
781: }
782: $Watcher->cancel();
783: KillSocket($Socket);
784: $ConnectionRetriesLeft--; # Counts as connection failure
785: return;
786: }
787: SocketDump(6,$Socket);
788:
789: $State = $Socket->GetState(); # Update in case of transition.
790: &Debug(6, "After read, state is ".$State);
791:
792: if($State eq "Initialized") {
793:
794:
795: } elsif ($State eq "ChallengeReceived") {
796: # The challenge must be echoed back; The state machine
797: # in the connection takes care of setting that up. Just
798: # need to transition to writable:
799:
800: $Watcher->cb(\&LondWritable);
801: $Watcher->poll("w");
802:
803: } elsif ($State eq "ChallengeReplied") {
804:
805: } elsif ($State eq "RequestingVersion") {
806: # Need to ask for the version... that is writiability:
807:
808: $Watcher->cb(\&LondWritable);
809: $Watcher->poll("w");
810:
811: } elsif ($State eq "ReadingVersionString") {
812: # Read the rest of the version string...
813: } elsif ($State eq "SetHost") {
814: # Need to request the actual domain get set...
815:
816: $Watcher->cb(\&LondWritable);
817: $Watcher->poll("w");
818: } elsif ($State eq "HostSet") {
819: # Reading the 'ok' from the peer.
820:
821: } elsif ($State eq "RequestingKey") {
822: # The ok was received. Now we need to request the key
823: # That requires us to be writable:
824:
825: $Watcher->cb(\&LondWritable);
826: $Watcher->poll("w");
827:
828: } elsif ($State eq "ReceivingKey") {
829:
830: } elsif ($State eq "Idle") {
831:
832: # This is as good a spot as any to get the peer version
833: # string:
834:
835: if($LondVersion eq "unknown") {
836: $LondVersion = $Socket->PeerVersion();
837: Log("INFO", "Connected to lond version: $LondVersion");
838: }
839: # If necessary, complete a transaction and then go into the
840: # idle queue.
841: # Note that a trasition to idle indicates a live lond
842: # on the other end so reset the connection retries.
843: #
844: $ConnectionRetriesLeft = $ConnectionRetries; # success resets the count
845: $Watcher->cancel();
846: if(exists($ActiveTransactions{$Socket})) {
847: Debug(5,"Completing transaction!!");
848: CompleteTransaction($Socket,
849: $ActiveTransactions{$Socket});
850: } else {
851: Log("SUCCESS", "Connection ".$ConnectionCount." to "
852: .$RemoteHost." now ready for action");
853: }
854: ServerToIdle($Socket); # Next work unit or idle.
855:
856: } elsif ($State eq "SendingRequest") {
857: # We need to be writable for this and probably don't belong
858: # here inthe first place.
859:
860: Deubg(6, "SendingRequest state encountered in readable");
861: $Watcher->poll("w");
862: $Watcher->cb(\&LondWritable);
863:
864: } elsif ($State eq "ReceivingReply") {
865:
866:
867: } else {
868: # Invalid state.
869: Debug(4, "Invalid state in LondReadable");
870: }
871: }
872:
873: =pod
874:
875: =head2 LondWritable
876:
877: This function is called whenever a lond connection
878: becomes writable while there is a writeable monitoring
879: event. The action taken is very state dependent:
880:
881: =head3 State = Connected
882:
883: The connection is in the process of sending the 'init' hailing to the
884: lond on the remote end. The connection object''s Writable member is
885: called. On error, ConnectionError is called to destroy the connection
886: and remove it from the ActiveConnections hash
887:
888: =head3 Initialized
889:
890: 'init' has been sent, writability monitoring is removed and
891: readability monitoring is started with LondReadable as the callback.
892:
893: =head3 ChallengeReceived
894:
895: The connection has received the who are you challenge from the remote
896: system, and is in the process of sending the challenge
897: response. Writable is called.
898:
899: =head3 ChallengeReplied
900:
901: The connection has replied to the initial challenge The we switch to
902: monitoring readability looking for the server to reply with 'ok'.
903:
904: =head3 RequestingKey
905:
906: The connection is in the process of requesting its encryption key.
907: Writable is called.
908:
909: =head3 ReceivingKey
910:
911: The connection has sent the request for a key. Switch to readability
912: monitoring to accept the key
913:
914: =head3 SendingRequest
915:
916: The connection is in the process of sending a request to the server.
917: This request is part of a client transaction. All the states until
918: now represent the client setup protocol. Writable is called.
919:
920: =head3 ReceivingReply
921:
922: The connection has sent a request. Now it must receive a reply.
923: Readability monitoring is requested.
924:
925: This function is an event handler and therefore receives as
926: a parameter the event that has fired. The data for the watcher
927: of this event is a reference to a list of one or two elements,
928: depending on state. The first (and possibly only) element is the
929: socket. The second (present only if a request is in progress)
930: is the socket on which to return a reply to the caller.
931:
932: =cut
933:
934: sub LondWritable {
935: my $Event = shift;
936: my $Watcher = $Event->w;
937: my $Socket = $Watcher->data;
938: my $State = $Socket->GetState();
939:
940: Debug(6,"LondWritable State = ".$State."\n");
941:
942:
943: # Figure out what to do depending on the state of the socket:
944:
945:
946:
947:
948: SocketDump(6,$Socket);
949:
950: # If the socket is writable, we must always write.
951: # Only by writing will we undergo state transitions.
952: # Old logic wrote in state specific code below, however
953: # That forces us at least through another invocation of
954: # this function after writability is possible again.
955: # This logic also factors out common code for handling
956: # write failures... in all cases, write failures
957: # Kill the socket.
958: # This logic makes the branches of the >big< if below
959: # so that the writing states are actually NO-OPs.
960:
961: if ($Socket->Writable() != 0) {
962: # The write resulted in an error.
963: # We'll treat this as if the socket got disconnected:
964: Log("WARNING", "Connection to ".$RemoteHost.
965: " has been disconnected");
966: if(exists($ActiveTransactions{$Socket})) {
967: FailTransaction($ActiveTransactions{$Socket});
968: }
969: $Watcher->cancel();
970: KillSocket($Socket);
971: return;
972: }
973:
974:
975:
976: if ($State eq "Connected") {
977:
978: # "init" is being sent...
979:
980: } elsif ($State eq "Initialized") {
981:
982: # Now that init was sent, we switch
983: # to watching for readability:
984:
985: $Watcher->cb(\&LondReadable);
986: $Watcher->poll("r");
987:
988: } elsif ($State eq "ChallengeReceived") {
989: # We received the challenge, now we
990: # are echoing it back. This is a no-op,
991: # we're waiting for the state to change
992:
993: } elsif ($State eq "ChallengeReplied") {
994: # The echo was sent back, so we switch
995: # to watching readability.
996:
997: $Watcher->cb(\&LondReadable);
998: $Watcher->poll("r");
999: } elsif ($State eq "RequestingVersion") {
1000: # Sending the peer a version request...
1001:
1002: } elsif ($State eq "ReadingVersionString") {
1003: # Transition to read since we have sent the
1004: # version command and now just need to read the
1005: # version string from the peer:
1006:
1007: $Watcher->cb(\&LondReadable);
1008: $Watcher->poll("r");
1009:
1010: } elsif ($State eq "SetHost") {
1011: # Setting the remote domain...
1012:
1013: } elsif ($State eq "HostSet") {
1014: # Back to readable to get the ok.
1015:
1016: $Watcher->cb(\&LondReadable);
1017: $Watcher->poll("r");
1018:
1019:
1020: } elsif ($State eq "RequestingKey") {
1021: # At this time we're requesting the key.
1022: # again, this is essentially a no-op.
1023:
1024: } elsif ($State eq "ReceivingKey") {
1025: # Now we need to wait for the key
1026: # to come back from the peer:
1027:
1028: $Watcher->cb(\&LondReadable);
1029: $Watcher->poll("r");
1030:
1031: } elsif ($State eq "SendingRequest") {
1032:
1033: # At this time we are sending a request to the
1034: # peer... write the next chunk:
1035:
1036:
1037: } elsif ($State eq "ReceivingReply") {
1038: # The send has completed. Wait for the
1039: # data to come in for a reply.
1040: Debug(8,"Writable sent request/receiving reply");
1041: $Watcher->cb(\&LondReadable);
1042: $Watcher->poll("r");
1043:
1044: } else {
1045: # Control only passes here on an error:
1046: # the socket state does not match any
1047: # of the known states... so an error
1048: # must be logged.
1049:
1050: &Debug(4, "Invalid socket state ".$State."\n");
1051: }
1052:
1053: }
1054: =pod
1055:
1056: =cut
1057: sub QueueDelayed {
1058: Debug(3,"QueueDelayed called");
1059:
1060: my $path = "$perlvar{'lonSockDir'}/delayed";
1061:
1062: Debug(4, "Delayed path: ".$path);
1063: opendir(DIRHANDLE, $path);
1064:
1065: my @alldelayed = grep /\.$RemoteHost$/, readdir DIRHANDLE;
1066: closedir(DIRHANDLE);
1067: my $dfname;
1068: my $reqfile;
1069: foreach $dfname (sort @alldelayed) {
1070: $reqfile = "$path/$dfname";
1071: Debug(4, "queueing ".$reqfile);
1072: my $Handle = IO::File->new($reqfile);
1073: my $cmd = <$Handle>;
1074: chomp $cmd; # There may or may not be a newline...
1075: $cmd = $cmd."\n"; # now for sure there's exactly one newline.
1076: my $Transaction = LondTransaction->new($cmd);
1077: $Transaction->SetDeferred($reqfile);
1078: QueueTransaction($Transaction);
1079: }
1080:
1081: }
1082:
1083: =pod
1084:
1085: =head2 MakeLondConnection
1086:
1087: Create a new lond connection object, and start it towards its initial
1088: idleness. Once idle, it becomes elligible to receive transactions
1089: from the work queue. If the work queue is not empty when the
1090: connection is completed and becomes idle, it will dequeue an entry and
1091: start off on it.
1092:
1093: =cut
1094:
1095: sub MakeLondConnection {
1096: Debug(4,"MakeLondConnection to ".GetServerHost()." on port "
1097: .GetServerPort());
1098:
1099: my $Connection = LondConnection->new(&GetServerHost(),
1100: &GetServerPort());
1101:
1102: if($Connection eq undef) { # Needs to be more robust later.
1103: Log("CRITICAL","Failed to make a connection with lond.");
1104: $ConnectionRetriesLeft--;
1105: return 0; # Failure.
1106: } else {
1107:
1108: # The connection needs to have writability
1109: # monitored in order to send the init sequence
1110: # that starts the whole authentication/key
1111: # exchange underway.
1112: #
1113: my $Socket = $Connection->GetSocket();
1114: if($Socket eq undef) {
1115: die "did not get a socket from the connection";
1116: } else {
1117: &Debug(9,"MakeLondConnection got socket: ".$Socket);
1118: }
1119:
1120: $Connection->SetTimeoutCallback(\&SocketTimeout);
1121:
1122: my $event = Event->io(fd => $Socket,
1123: poll => 'w',
1124: cb => \&LondWritable,
1125: data => $Connection,
1126: desc => 'Connection to lond server');
1127: $ActiveConnections{$Connection} = $event;
1128:
1129: $ConnectionCount++;
1130: Debug(4, "Connection count = ".$ConnectionCount);
1131: if($ConnectionCount == 1) { # First Connection:
1132: QueueDelayed;
1133: }
1134: Log("SUCESS", "Created connection ".$ConnectionCount
1135: ." to host ".GetServerHost());
1136: return 1; # Return success.
1137: }
1138:
1139: }
1140:
1141: =pod
1142:
1143: =head2 StartRequest
1144:
1145: Starts a lond request going on a specified lond connection.
1146: parameters are:
1147:
1148: =item $Lond
1149:
1150: Connection to the lond that will send the transaction and receive the
1151: reply.
1152:
1153: =item $Client
1154:
1155: Connection to the client that is making this request We got the
1156: request from this socket, and when the request has been relayed to
1157: lond and we get a reply back from lond it will get sent to this
1158: socket.
1159:
1160: =item $Request
1161:
1162: The text of the request to send.
1163:
1164: =cut
1165:
1166: sub StartRequest {
1167: my $Lond = shift;
1168: my $Request = shift; # This is a LondTransaction.
1169:
1170: Debug(6, "StartRequest: ".$Request->getRequest());
1171:
1172: my $Socket = $Lond->GetSocket();
1173:
1174: $Request->Activate($Lond);
1175: $ActiveTransactions{$Lond} = $Request;
1176:
1177: $Lond->InitiateTransaction($Request->getRequest());
1178: my $event = Event->io(fd => $Socket,
1179: poll => "w",
1180: cb => \&LondWritable,
1181: data => $Lond,
1182: desc => "lond transaction connection");
1183: $ActiveConnections{$Lond} = $event;
1184: Debug(8," Start Request made watcher data with ".$event->data."\n");
1185: }
1186:
1187: =pod
1188:
1189: =head2 QueueTransaction
1190:
1191: If there is an idle lond connection, it is put to work doing this
1192: transaction. Otherwise, the transaction is placed in the work queue.
1193: If placed in the work queue and the maximum number of connections has
1194: not yet been created, a new connection will be started. Our goal is
1195: to eventually have a sufficient number of connections that the work
1196: queue will typically be empty. parameters are:
1197:
1198: =item Socket
1199:
1200: open on the lonc client.
1201:
1202: =item Request
1203:
1204: data to send to the lond.
1205:
1206: =cut
1207:
1208: sub QueueTransaction {
1209:
1210: my $requestData = shift; # This is a LondTransaction.
1211: my $cmd = $requestData->getRequest();
1212:
1213: Debug(6,"QueueTransaction: ".$cmd);
1214:
1215: my $LondSocket = $IdleConnections->pop();
1216: if(!defined $LondSocket) { # Need to queue request.
1217: Debug(5,"Must queue...");
1218: $WorkQueue->enqueue($requestData);
1219: if($ConnectionCount < $MaxConnectionCount) {
1220: if($ConnectionRetriesLeft > 0) {
1221: Debug(5,"Starting additional lond connection");
1222: if(MakeLondConnection() == 0) {
1223: EmptyQueue(); # Fail transactions, can't make connection.
1224: CloseAllLondConnections; # Should all be closed but...
1225: }
1226: } else {
1227: ShowStatus(GetServerHost()." >>> DEAD !!!! <<<");
1228: EmptyQueue(); # It's worse than that ... he's dead Jim.
1229: CloseAllLondConnections; # Should all be closed but..
1230: }
1231: }
1232: } else { # Can start the request:
1233: Debug(8,"Can start...");
1234: StartRequest($LondSocket, $requestData);
1235: }
1236: }
1237:
1238: #-------------------------- Lonc UNIX socket handling ---------------------
1239:
1240: =pod
1241:
1242: =head2 ClientRequest
1243: Callback that is called when data can be read from the UNIX domain
1244: socket connecting us with an apache server process.
1245:
1246: =cut
1247:
1248: sub ClientRequest {
1249: Debug(6, "ClientRequest");
1250: my $event = shift;
1251: my $watcher = $event->w;
1252: my $socket = $watcher->fd;
1253: my $data = $watcher->data;
1254: my $thisread;
1255:
1256: Debug(9, " Watcher named: ".$watcher->desc);
1257:
1258: my $rv = $socket->recv($thisread, POSIX::BUFSIZ, 0);
1259: Debug(8, "rcv: data length = ".length($thisread)
1260: ." read =".$thisread);
1261: unless (defined $rv && length($thisread)) {
1262: # Likely eof on socket.
1263: Debug(5,"Client Socket closed on lonc for ".$RemoteHost);
1264: close($socket);
1265: $watcher->cancel();
1266: delete($ActiveClients{$socket});
1267: return;
1268: }
1269: Debug(8,"Data: ".$data." this read: ".$thisread);
1270: $data = $data.$thisread; # Append new data.
1271: $watcher->data($data);
1272: if($data =~ /(.*\n)/) { # Request entirely read.
1273: if($data eq "close_connection_exit\n") {
1274: Log("CRITICAL",
1275: "Request Close Connection ... exiting");
1276: CloseAllLondConnections();
1277: exit;
1278: }
1279: Debug(8, "Complete transaction received: ".$data);
1280: if($LogTransactions) {
1281: Log("SUCCESS", "Transaction: '$data'"); # Transaction has \n.
1282: }
1283: my $Transaction = LondTransaction->new($data);
1284: $Transaction->SetClient($socket);
1285: QueueTransaction($Transaction);
1286: $watcher->cancel(); # Done looking for input data.
1287: }
1288:
1289: }
1290:
1291:
1292: =pod
1293:
1294: =head2 NewClient
1295:
1296: Callback that is called when a connection is received on the unix
1297: socket for a new client of lonc. The callback is parameterized by the
1298: event.. which is a-priori assumed to be an io event, and therefore has
1299: an fd member that is the Listener socket. We Accept the connection
1300: and register a new event on the readability of that socket:
1301:
1302: =cut
1303:
1304: sub NewClient {
1305: Debug(6, "NewClient");
1306: my $event = shift; # Get the event parameters.
1307: my $watcher = $event->w;
1308: my $socket = $watcher->fd; # Get the event' socket.
1309: my $connection = $socket->accept(); # Accept the client connection.
1310: Debug(8,"Connection request accepted from "
1311: .GetPeername($connection, AF_UNIX));
1312:
1313:
1314: my $description = sprintf("Connection to lonc client %d",
1315: $ClientConnection);
1316: Debug(9, "Creating event named: ".$description);
1317: Event->io(cb => \&ClientRequest,
1318: poll => 'r',
1319: desc => $description,
1320: data => "",
1321: fd => $connection);
1322: $ActiveClients{$connection} = $ClientConnection;
1323: $ClientConnection++;
1324: }
1325:
1326: =pod
1327:
1328: =head2 GetLoncSocketPath
1329:
1330: Returns the name of the UNIX socket on which to listen for client
1331: connections.
1332:
1333: =cut
1334:
1335: sub GetLoncSocketPath {
1336: return $UnixSocketDir."/".GetServerHost();
1337: }
1338:
1339: =pod
1340:
1341: =head2 GetServerHost
1342:
1343: Returns the host whose lond we talk with.
1344:
1345: =cut
1346:
1347: sub GetServerHost {
1348: return $RemoteHost; # Setup by the fork.
1349: }
1350:
1351: =pod
1352:
1353: =head2 GetServerPort
1354:
1355: Returns the lond port number.
1356:
1357: =cut
1358:
1359: sub GetServerPort {
1360: return $perlvar{londPort};
1361: }
1362:
1363: =pod
1364:
1365: =head2 SetupLoncListener
1366:
1367: Setup a lonc listener event. The event is called when the socket
1368: becomes readable.. that corresponds to the receipt of a new
1369: connection. The event handler established will accept the connection
1370: (creating a communcations channel), that int turn will establish
1371: another event handler to subess requests.
1372:
1373: =cut
1374:
1375: sub SetupLoncListener {
1376:
1377: my $socket;
1378: my $SocketName = GetLoncSocketPath();
1379: unlink($SocketName);
1380: unless ($socket =IO::Socket::UNIX->new(Local => $SocketName,
1381: Listen => 10,
1382: Type => SOCK_STREAM)) {
1383: die "Failed to create a lonc listner socket";
1384: }
1385: Event->io(cb => \&NewClient,
1386: poll => 'r',
1387: desc => 'Lonc listener Unix Socket',
1388: fd => $socket);
1389: }
1390:
1391: #
1392: # Toggle transaction logging.
1393: # Implicit inputs:
1394: # LogTransactions
1395: # Implicit Outputs:
1396: # LogTransactions
1397: sub ToggleTransactionLogging {
1398: print STDERR "Toggle transaction logging...\n";
1399: if(!$LogTransactions) {
1400: $LogTransactions = 1;
1401: } else {
1402: $LogTransactions = 0;
1403: }
1404:
1405:
1406: Log("SUCCESS", "Toggled transaction logging: $LogTransactions \n");
1407: }
1408:
1409: =pod
1410:
1411: =head2 ChildStatus
1412:
1413: Child USR1 signal handler to report the most recent status
1414: into the status file.
1415:
1416: We also use this to reset the retries count in order to allow the
1417: client to retry connections with a previously dead server.
1418: =cut
1419: sub ChildStatus {
1420: my $event = shift;
1421: my $watcher = $event->w;
1422:
1423: Debug(2, "Reporting child status because : ".$watcher->data);
1424: my $docdir = $perlvar{'lonDocRoot'};
1425: my $fh = IO::File->new(">>$docdir/lon-status/loncstatus.txt");
1426: print $fh $$."\t".$RemoteHost."\t".$Status."\t".
1427: $RecentLogEntry."\n";
1428: #
1429: # Write out information about each of the connections:
1430: #
1431: print $fh "Active connection statuses: \n";
1432: my $i = 1;
1433: print STDERR "================================= Socket Status Dump:\n";
1434: foreach my $item (keys %ActiveConnections) {
1435: my $Socket = $ActiveConnections{$item}->data;
1436: my $state = $Socket->GetState();
1437: print $fh "Connection $i State: $state\n";
1438: print STDERR "---------------------- Connection $i \n";
1439: $Socket->Dump();
1440: $i++;
1441: }
1442: $ConnectionRetriesLeft = $ConnectionRetries;
1443: }
1444:
1445: =pod
1446:
1447: =head2 SignalledToDeath
1448:
1449: Called in response to a signal that causes a chid process to die.
1450:
1451: =cut
1452:
1453:
1454: sub SignalledToDeath {
1455: my $event = shift;
1456: my $watcher= $event->w;
1457:
1458: Debug(2,"Signalled to death! via ".$watcher->data);
1459: my ($signal) = $watcher->data;
1460: chomp($signal);
1461: Log("CRITICAL", "Abnormal exit. Child $$ for $RemoteHost "
1462: ."died through "."\"$signal\"");
1463: LogPerm("F:lonc: $$ on $RemoteHost signalled to death: "
1464: ."\"$signal\"");
1465: exit 0;
1466:
1467: }
1468:
1469: =head2 ToggleDebug
1470:
1471: This sub toggles trace debugging on and off.
1472:
1473: =cut
1474:
1475: sub ToggleDebug {
1476: my $Current = $DebugLevel;
1477: $DebugLevel = $NextDebugLevel;
1478: $NextDebugLevel = $Current;
1479:
1480: Log("SUCCESS", "New debugging level for $RemoteHost now $DebugLevel");
1481:
1482: }
1483:
1484: =head2 ChildProcess
1485:
1486: This sub implements a child process for a single lonc daemon.
1487:
1488: =cut
1489:
1490: sub ChildProcess {
1491:
1492:
1493: #
1494: # Signals must be handled by the Event framework...
1495: #
1496:
1497: Event->signal(signal => "QUIT",
1498: cb => \&SignalledToDeath,
1499: data => "QUIT");
1500: Event->signal(signal => "HUP",
1501: cb => \&ChildStatus,
1502: data => "HUP");
1503: Event->signal(signal => "USR1",
1504: cb => \&ChildStatus,
1505: data => "USR1");
1506: Event->signal(signal => "USR2",
1507: cb => \&ToggleTransactionLogging);
1508: Event->signal(signal => "INT",
1509: cb => \&ToggleDebug,
1510: data => "INT");
1511:
1512: SetupTimer();
1513:
1514: SetupLoncListener();
1515:
1516: $Event::Debuglevel = $DebugLevel;
1517:
1518: Debug(9, "Making initial lond connection for ".$RemoteHost);
1519:
1520: # Setup the initial server connection:
1521:
1522: # &MakeLondConnection(); // let first work requirest do it.
1523:
1524:
1525: Debug(9,"Entering event loop");
1526: my $ret = Event::loop(); # Start the main event loop.
1527:
1528:
1529: die "Main event loop exited!!!";
1530: }
1531:
1532: # Create a new child for host passed in:
1533:
1534: sub CreateChild {
1535: my $sigset = POSIX::SigSet->new(SIGINT);
1536: sigprocmask(SIG_BLOCK, $sigset);
1537: my $host = shift;
1538: $RemoteHost = $host;
1539: Log("CRITICAL", "Forking server for ".$host);
1540: my $pid = fork;
1541: if($pid) { # Parent
1542: $RemoteHost = "Parent";
1543: $ChildHash{$pid} = $host;
1544: $HostToPid{$host}= $pid;
1545: sigprocmask(SIG_UNBLOCK, $sigset);
1546:
1547: } else { # child.
1548: ShowStatus("Connected to ".$RemoteHost);
1549: $SIG{INT} = 'DEFAULT';
1550: sigprocmask(SIG_UNBLOCK, $sigset);
1551: ChildProcess; # Does not return.
1552: }
1553:
1554: }
1555: #
1556: # Parent process logic pass 1:
1557: # For each entry in the hosts table, we will
1558: # fork off an instance of ChildProcess to service the transactions
1559: # to that host. Each pid will be entered in a global hash
1560: # with the value of the key, the host.
1561: # The parent will then enter a loop to wait for process exits.
1562: # Each exit gets logged and the child gets restarted.
1563: #
1564:
1565: #
1566: # Fork and start in new session so hang-up isn't going to
1567: # happen without intent.
1568: #
1569:
1570:
1571:
1572:
1573:
1574:
1575: ShowStatus("Forming new session");
1576: my $childpid = fork;
1577: if ($childpid != 0) {
1578: sleep 4; # Give child a chacne to break to
1579: exit 0; # a new sesion.
1580: }
1581: #
1582: # Write my pid into the pid file so I can be located
1583: #
1584:
1585: ShowStatus("Parent writing pid file:");
1586: my $execdir = $perlvar{'lonDaemons'};
1587: open (PIDSAVE, ">$execdir/logs/lonc.pid");
1588: print PIDSAVE "$$\n";
1589: close(PIDSAVE);
1590:
1591:
1592:
1593: if (POSIX::setsid() < 0) {
1594: print "Could not create new session\n";
1595: exit -1;
1596: }
1597:
1598: ShowStatus("Forking node servers");
1599:
1600: Log("CRITICAL", "--------------- Starting children ---------------");
1601:
1602: LondConnection::ReadConfig; # Read standard config files.
1603: my $HostIterator = LondConnection::GetHostIterator;
1604: while (! $HostIterator->end()) {
1605:
1606: my $hostentryref = $HostIterator->get();
1607: CreateChild($hostentryref->[0]);
1608: $HostHash{$hostentryref->[0]} = $hostentryref->[4];
1609: $HostIterator->next();
1610: }
1611: $RemoteHost = "Parent Server";
1612:
1613: # Maintain the population:
1614:
1615: ShowStatus("Parent keeping the flock");
1616:
1617: #
1618: # Set up parent signals:
1619: #
1620:
1621: $SIG{INT} = \&Terminate;
1622: $SIG{TERM} = \&Terminate;
1623: $SIG{HUP} = \&Restart;
1624: $SIG{USR1} = \&CheckKids;
1625: $SIG{USR2} = \&UpdateKids; # LonManage update request.
1626:
1627: while(1) {
1628: my $deadchild = wait();
1629: if(exists $ChildHash{$deadchild}) { # need to restart.
1630: my $deadhost = $ChildHash{$deadchild};
1631: delete($HostToPid{$deadhost});
1632: delete($ChildHash{$deadchild});
1633: Log("WARNING","Lost child pid= ".$deadchild.
1634: "Connected to host ".$deadhost);
1635: Log("INFO", "Restarting child procesing ".$deadhost);
1636: CreateChild($deadhost);
1637: }
1638: }
1639:
1640:
1641:
1642: =pod
1643:
1644: =head1 CheckKids
1645:
1646: Since kids do not die as easily in this implementation
1647: as the previous one, there is no need to restart the
1648: dead ones (all dead kids get restarted when they die!!)
1649: The only thing this function does is to pass USR1 to the
1650: kids so that they report their status.
1651:
1652: =cut
1653:
1654: sub CheckKids {
1655: Debug(2, "Checking status of children");
1656: my $docdir = $perlvar{'lonDocRoot'};
1657: my $fh = IO::File->new(">$docdir/lon-status/loncstatus.txt");
1658: my $now=time;
1659: my $local=localtime($now);
1660: print $fh "LONC status $local - parent $$ \n\n";
1661: foreach my $pid (keys %ChildHash) {
1662: Debug(2, "Sending USR1 -> $pid");
1663: kill 'USR1' => $pid; # Tell Child to report status.
1664: sleep 1; # Wait so file doesn't intermix.
1665: }
1666: }
1667:
1668: =pod
1669:
1670: =head1 UpdateKids
1671:
1672: parent's SIGUSR2 handler. This handler:
1673:
1674: =item
1675:
1676: Rereads the hosts file.
1677:
1678: =item
1679:
1680: Kills off (via sigint) children for hosts that have disappeared.
1681:
1682: =item
1683:
1684: QUITs children for hosts that already exist (this just forces a status display
1685: and resets the connection retry count for that host.
1686:
1687: =item
1688:
1689: Starts new children for hosts that have been added to the hosts.tab file since
1690: the start of the master program and maintains them.
1691:
1692: =cut
1693:
1694: sub UpdateKids {
1695:
1696: Log("INFO", "Updating connections via SIGUSR2");
1697:
1698: # Just in case we need to kill our own lonc, we wait a few seconds to
1699: # give it a chance to receive and relay lond's response to the
1700: # re-init command.
1701: #
1702:
1703: sleep(2); # Wait a couple of seconds.
1704:
1705: my %hosts; # Indexed by loncapa hostname, value=ip.
1706:
1707: # Need to re-read the host table:
1708:
1709:
1710: LondConnection::ReadConfig();
1711: my $I = LondConnection::GetHostIterator;
1712: while (! $I->end()) {
1713: my $item = $I->get();
1714: $hosts{$item->[0]} = $item->[4];
1715: $I->next();
1716: }
1717:
1718: # The logic below is written for clarity not for efficiency.
1719: # Since I anticipate that this function is only rarely called, that's
1720: # appropriate. There are certainly ways to combine the loops below,
1721: # and anyone wishing to obscure the logic is welcome to go for it.
1722: # Note that we don't re-direct sigchild. Instead we do what's needed
1723: # to the data structures that keep track of children to ensure that
1724: # when sigchild is honored, no new child is born.
1725: #
1726:
1727: # For each existing child; if it's host doesn't exist, kill the child.
1728:
1729: foreach my $child (keys %ChildHash) {
1730: my $oldhost = $ChildHash{$child};
1731: if (!(exists $hosts{$oldhost})) {
1732: Log("CRITICAL", "Killing child for $oldhost host no longer exists");
1733: delete $ChildHash{$child};
1734: delete $HostToPid{$oldhost};
1735: kill 'QUIT' => $child;
1736: }
1737: }
1738: # For each remaining existing child; if it's host's ip has changed,
1739: # Restart the child on the new IP.
1740:
1741: foreach my $child (keys %ChildHash) {
1742: my $oldhost = $ChildHash{$child};
1743: my $oldip = $HostHash{$oldhost};
1744: if ($hosts{$oldhost} ne $oldip) {
1745:
1746: # kill the old child.
1747:
1748: Log("CRITICAL", "Killing child for $oldhost host ip has changed...");
1749: delete $ChildHash{$child};
1750: delete $HostToPid{$oldhost};
1751: kill 'QUIT' => $child;
1752:
1753: # Do the book-keeping needed to start a new child on the
1754: # new ip.
1755:
1756: $HostHash{$oldhost} = $hosts{$oldhost};
1757: CreateChild($oldhost);
1758: }
1759: }
1760: # Finally, for each new host, not in the host hash, create a
1761: # enter the host and create a new child.
1762: # Force a status display of any existing process.
1763:
1764: foreach my $host (keys %hosts) {
1765: if(!(exists $HostHash{$host})) {
1766: Log("INFO", "New host $host discovered in hosts.tab...");
1767: $HostHash{$host} = $hosts{$host};
1768: CreateChild($host);
1769: } else {
1770: kill 'HUP' => $HostToPid{$host}; # status display.
1771: }
1772: }
1773: }
1774:
1775:
1776: =pod
1777:
1778: =head1 Restart
1779:
1780: Signal handler for HUP... all children are killed and
1781: we self restart. This is an el-cheapo way to re read
1782: the config file.
1783:
1784: =cut
1785:
1786: sub Restart {
1787: &KillThemAll; # First kill all the children.
1788: Log("CRITICAL", "Restarting");
1789: my $execdir = $perlvar{'lonDaemons'};
1790: unlink("$execdir/logs/lonc.pid");
1791: exec("$execdir/loncnew");
1792: }
1793:
1794: =pod
1795:
1796: =head1 KillThemAll
1797:
1798: Signal handler that kills all children by sending them a
1799: SIGHUP. Responds to sigint and sigterm.
1800:
1801: =cut
1802:
1803: sub KillThemAll {
1804: Debug(2, "Kill them all!!");
1805: local($SIG{CHLD}) = 'IGNORE'; # Our children >will< die.
1806: foreach my $pid (keys %ChildHash) {
1807: my $serving = $ChildHash{$pid};
1808: Debug(2, "Killing lonc for $serving pid = $pid");
1809: ShowStatus("Killing lonc for $serving pid = $pid");
1810: Log("CRITICAL", "Killing lonc for $serving pid = $pid");
1811: kill 'QUIT' => $pid;
1812: delete($ChildHash{$pid});
1813: }
1814: my $execdir = $perlvar{'lonDaemons'};
1815: unlink("$execdir/logs/lonc.pid");
1816:
1817: }
1818:
1819: =pod
1820:
1821: =head1 Terminate
1822:
1823: Terminate the system.
1824:
1825: =cut
1826:
1827: sub Terminate {
1828: KillThemAll;
1829: Log("CRITICAL","Master process exiting");
1830: exit 0;
1831:
1832: }
1833: =pod
1834:
1835: =head1 Theory
1836:
1837: The event class is used to build this as a single process with an
1838: event driven model. The following events are handled:
1839:
1840: =item UNIX Socket connection Received
1841:
1842: =item Request data arrives on UNIX data transfer socket.
1843:
1844: =item lond connection becomes writable.
1845:
1846: =item timer fires at 1 second intervals.
1847:
1848: All sockets are run in non-blocking mode. Timeouts managed by the timer
1849: handler prevents hung connections.
1850:
1851: Key data structures:
1852:
1853: =item RequestQueue
1854:
1855: A queue of requests received from UNIX sockets that are
1856: waiting for a chance to be forwarded on a lond connection socket.
1857:
1858: =item ActiveConnections
1859:
1860: A hash of lond connections that have transactions in process that are
1861: available to be timed out.
1862:
1863: =item ActiveTransactions
1864:
1865: A hash indexed by lond connections that contain the client reply
1866: socket for each connection that has an active transaction on it.
1867:
1868: =item IdleConnections
1869:
1870: A hash of lond connections that have no work to do. These connections
1871: can be closed if they are idle for a long enough time.
1872:
1873: =cut
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>