Diff for /loncom/loncnew between versions 1.60 and 1.76

version 1.60, 2004/09/22 11:11:54 version 1.76, 2006/08/25 21:12:19
Line 61  use LONCAPA::LondConnection; Line 61  use LONCAPA::LondConnection;
 use LONCAPA::LondTransaction;  use LONCAPA::LondTransaction;
 use LONCAPA::Configuration;  use LONCAPA::Configuration;
 use LONCAPA::HashIterator;  use LONCAPA::HashIterator;
   use Fcntl qw(:flock);
   
   
 # Read the httpd configuration file to get perl variables  # Read the httpd configuration file to get perl variables
Line 75  my %perlvar    = %{$perlvarref}; Line 76  my %perlvar    = %{$perlvarref};
 my %ChildHash; # by pid -> host.  my %ChildHash; # by pid -> host.
 my %HostToPid; # By host -> pid.  my %HostToPid; # By host -> pid.
 my %HostHash; # by loncapaname -> IP.  my %HostHash; # by loncapaname -> IP.
   my %listening_to; # Socket->host table for who the parent
                                   # is listening to.
   my %parent_dispatchers;         # host-> listener watcher events. 
   
   my %parent_handlers; # Parent signal handlers...
   
 my $MaxConnectionCount = 10; # Will get from config later.  my $MaxConnectionCount = 10; # Will get from config later.
 my $ClientConnection = 0; # Uniquifier for client events.  my $ClientConnection = 0; # Uniquifier for client events.
Line 85  my $NextDebugLevel= 2;  # So Sigint can Line 90  my $NextDebugLevel= 2;  # So Sigint can
 my $IdleTimeout= 600; # Wait 10 minutes before pruning connections.  my $IdleTimeout= 600; # Wait 10 minutes before pruning connections.
   
 my $LogTransactions = 0; # When True, all transactions/replies get logged.  my $LogTransactions = 0; # When True, all transactions/replies get logged.
   my $executable      = $0; # Get the full path to me.
   
 #  #
 #  The variables below are only used by the child processes.  #  The variables below are only used by the child processes.
Line 100  my $ConnectionCount = 0; Line 106  my $ConnectionCount = 0;
 my $IdleSeconds     = 0; # Number of seconds idle.  my $IdleSeconds     = 0; # Number of seconds idle.
 my $Status          = ""; # Current status string.  my $Status          = ""; # Current status string.
 my $RecentLogEntry  = "";  my $RecentLogEntry  = "";
 my $ConnectionRetries=2; # Number of connection retries allowed.  my $ConnectionRetries=5; # Number of connection retries allowed.
 my $ConnectionRetriesLeft=2; # Number of connection retries remaining.  my $ConnectionRetriesLeft=5; # Number of connection retries remaining.
 my $LondVersion     = "unknown"; # Version of lond we talk with.  my $LondVersion     = "unknown"; # Version of lond we talk with.
 my $KeyMode         = "";       # e.g. ssl, local, insecure from last connect.  my $KeyMode         = "";       # e.g. ssl, local, insecure from last connect.
 my $LondConnecting  = 0;       # True when a connection is being built.  my $LondConnecting  = 0;       # True when a connection is being built.
   
   
 # DO NOT SET THE NEXT VARIABLE TO NON ZERO!!!!!!!!!!!!!!!  
   
 my $DieWhenIdle     = 0; # When true children die when trimmed -> 0.  my $DieWhenIdle     = 1; # When true children die when trimmed -> 0.
   my $I_am_child      = 0; # True if this is the child process.
   
 #  #
 #   The hash below gives the HTML format for log messages  #   The hash below gives the HTML format for log messages
Line 146  sub UpdateStatus { Line 152  sub UpdateStatus {
 Makes an entry into the permanent log file.  Makes an entry into the permanent log file.
   
 =cut  =cut
   
 sub LogPerm {  sub LogPerm {
     my $message=shift;      my $message=shift;
     my $execdir=$perlvar{'lonDaemons'};      my $execdir=$perlvar{'lonDaemons'};
Line 191  sub Log { Line 198  sub Log {
     my $now   = time;      my $now   = time;
     my $local = localtime($now);      my $local = localtime($now);
     my $finalformat = "$local ($$) [$RemoteHost] [$Status] ";      my $finalformat = "$local ($$) [$RemoteHost] [$Status] ";
     my $finalformat = $finalformat.$format."\n";      $finalformat = $finalformat.$format."\n";
   
     # open the file and put the result.      # open the file and put the result.
   
Line 265  sub SocketDump { Line 272  sub SocketDump {
  and as what we return in a SIGUSR1   and as what we return in a SIGUSR1
   
 =cut  =cut
   
 sub ShowStatus {  sub ShowStatus {
     my $state = shift;      my $state = shift;
     my $now = time;      my $now = time;
Line 275  sub ShowStatus { Line 283  sub ShowStatus {
   
 =pod  =pod
   
 =head 2 SocketTimeout  =head2 SocketTimeout
   
     Called when an action on the socket times out.  The socket is       Called when an action on the socket times out.  The socket is 
    destroyed and any active transaction is failed.     destroyed and any active transaction is failed.
   
   
 =cut  =cut
   
 sub SocketTimeout {  sub SocketTimeout {
     my $Socket = shift;      my $Socket = shift;
     Log("WARNING", "A socket timeout was detected");      Log("WARNING", "A socket timeout was detected");
Line 299  sub SocketTimeout { Line 308  sub SocketTimeout {
     }      }
   
 }  }
   #
   #   This function should be called by the child in all cases where it must
   #   exit.  If the child process is running with the DieWhenIdle turned on
   #   it must create a lock file for the AF_UNIX socket in order to prevent
   #   connection requests from lonnet in the time between process exit
   #   and the parent picking up the listen again.
   # Parameters:
   #     exit_code           - Exit status value, however see the next parameter.
   #     message             - If this optional parameter is supplied, the exit
   #                           is via a die with this message.
   #
   sub child_exit {
       my ($exit_code, $message) = @_;
   
       # Regardless of how we exit, we may need to do the lock thing:
   
       if($DieWhenIdle) {
    #
    #  Create a lock file since there will be a time window
    #  between our exit and the parent's picking up the listen
    #  during which no listens will be done on the
    #  lonnet client socket.
    #
    my $lock_file = GetLoncSocketPath().".lock";
    open(LOCK,">$lock_file");
    print LOCK "Contents not important";
    close(LOCK);
   
    exit(0);
       }
       #  Now figure out how we exit:
   
       if($message) {
    die $message;
       } else {
    exit($exit_code);
       }
   }
 #----------------------------- Timer management ------------------------  #----------------------------- Timer management ------------------------
   
 =pod  =pod
Line 329  sub Tick { Line 376  sub Tick {
     $IdleSeconds = 0; # Otherwise all connections get trimmed to fast.      $IdleSeconds = 0; # Otherwise all connections get trimmed to fast.
     UpdateStatus();      UpdateStatus();
     if(($ConnectionCount == 0) && $DieWhenIdle) {      if(($ConnectionCount == 0) && $DieWhenIdle) {
  #   &child_exit(0);
  #  Create a lock file since there will be a time window  
  #  between our exit and the parent's picking up the listen  
  #  during which no listens will be done on the  
  #  lonnet client socket.  
  #  
  my $lock_file = GetLoncSocketPath().".lock";  
  open(LOCK,">$lock_file");  
  print LOCK "Contents not important";  
  close(LOCK);  
   
  exit(0);  
     }      }
  }   }
     } else {      } else {
Line 386  sub Tick { Line 423  sub Tick {
  $KeyMode = "";    $KeyMode = ""; 
  $clock_watcher->cancel();   $clock_watcher->cancel();
     }      }
       &UpdateStatus();
 }  }
   
 =pod  =pod
Line 639  sub FailTransaction { Line 677  sub FailTransaction {
   
     if ($ConnectionRetriesLeft > 0) {      if ($ConnectionRetriesLeft > 0) {
  Log("WARNING", "Failing transaction "   Log("WARNING", "Failing transaction "
     .$transaction->getRequest());      .$transaction->getLoggableRequest());
     }      }
     Debug(1, "Failing transaction: ".$transaction->getRequest());      Debug(1, "Failing transaction: ".$transaction->getLoggableRequest());
     if (!$transaction->isDeferred()) { # If the transaction is deferred we'll get to it.      if (!$transaction->isDeferred()) { # If the transaction is deferred we'll get to it.
  my $client  = $transaction->getClient();   my $client  = $transaction->getClient();
  Debug(1," Replying con_lost to ".$transaction->getRequest());   Debug(1," Replying con_lost to ".$transaction->getRequest());
Line 651  sub FailTransaction { Line 689  sub FailTransaction {
 }  }
   
 =pod  =pod
   
 =head1  EmptyQueue  =head1  EmptyQueue
   
   Fails all items in the work queue with con_lost.    Fails all items in the work queue with con_lost.
   Note that each item in the work queue is a transaction.    Note that each item in the work queue is a transaction.
   
 =cut  =cut
   
 sub EmptyQueue {  sub EmptyQueue {
     $ConnectionRetriesLeft--; # Counts as connection failure too.      $ConnectionRetriesLeft--; # Counts as connection failure too.
     while($WorkQueue->Count()) {      while($WorkQueue->Count()) {
Line 672  sub EmptyQueue { Line 712  sub EmptyQueue {
 Close all connections open on lond prior to exit e.g.  Close all connections open on lond prior to exit e.g.
   
 =cut  =cut
   
 sub CloseAllLondConnections {  sub CloseAllLondConnections {
     foreach my $Socket (keys %ActiveConnections) {      foreach my $Socket (keys %ActiveConnections) {
       if(exists($ActiveTransactions{$Socket})) {        if(exists($ActiveTransactions{$Socket})) {
Line 680  sub CloseAllLondConnections { Line 721  sub CloseAllLondConnections {
       KillSocket($Socket);        KillSocket($Socket);
     }      }
 }  }
 =cut  
   
 =pod  =pod
   
Line 702  Parameters: Line 742  Parameters:
   
 nonzero if we are allowed to create a new connection.  nonzero if we are allowed to create a new connection.
   
   
 =cut  =cut
   
 sub KillSocket {  sub KillSocket {
     my $Socket = shift;      my $Socket = shift;
   
Line 915  sub LondReadable { Line 955  sub LondReadable {
  #  We need to be writable for this and probably don't belong   #  We need to be writable for this and probably don't belong
  #  here inthe first place.   #  here inthe first place.
   
  Deubg(6, "SendingRequest state encountered in readable");   Debug(6, "SendingRequest state encountered in readable");
  $Watcher->poll("w");   $Watcher->poll("w");
  $Watcher->cb(\&LondWritable);   $Watcher->cb(\&LondWritable);
   
Line 1116  sub LondWritable { Line 1156  sub LondWritable {
 =pod  =pod
           
 =cut  =cut
   
 sub QueueDelayed {  sub QueueDelayed {
     Debug(3,"QueueDelayed called");      Debug(3,"QueueDelayed called");
   
Line 1123  sub QueueDelayed { Line 1164  sub QueueDelayed {
   
     Debug(4, "Delayed path: ".$path);      Debug(4, "Delayed path: ".$path);
     opendir(DIRHANDLE, $path);      opendir(DIRHANDLE, $path);
       
     my @alldelayed = grep /\.$RemoteHost$/, readdir DIRHANDLE;      my @all_host_ids;
       my $host_iterator =   &LondConnection::GetHostIterator();
       while (!$host_iterator->end()) {
    my ($host_id,$host_name) = @{$host_iterator->get()}[0,3];
    if ($host_name eq $RemoteHost) {
       push(@all_host_ids, $host_id);
    }
    $host_iterator->next();
       }
       my $host_id_re = '(?:'.join('|',@all_host_ids).')';
       my @alldelayed = grep(/\.$host_id_re$/, readdir(DIRHANDLE));
     closedir(DIRHANDLE);      closedir(DIRHANDLE);
     my $dfname;      foreach my $dfname (sort(@alldelayed)) {
     my $reqfile;   my $reqfile = "$path/$dfname";
     foreach $dfname (sort  @alldelayed) {   my ($host_id) = ($dfname =~ /\.([^.]*)$/);
  $reqfile = "$path/$dfname";   Debug(4, "queueing ".$reqfile." for $host_id");
  Debug(4, "queueing ".$reqfile);  
  my $Handle = IO::File->new($reqfile);   my $Handle = IO::File->new($reqfile);
  my $cmd    = <$Handle>;   my $cmd    = <$Handle>;
  chomp $cmd; # There may or may not be a newline...   chomp $cmd; # There may or may not be a newline...
  $cmd = $cmd."\n"; # now for sure there's exactly one newline.   $cmd = $cmd."\n"; # now for sure there's exactly one newline.
  my $Transaction = LondTransaction->new($cmd);   my $Transaction = LondTransaction->new("sethost:$host_id:$cmd");
  $Transaction->SetDeferred($reqfile);   $Transaction->SetDeferred($reqfile);
  QueueTransaction($Transaction);   QueueTransaction($Transaction);
     }      }
Line 1174  sub MakeLondConnection { Line 1224  sub MakeLondConnection {
  #   #
  my $Socket = $Connection->GetSocket();   my $Socket = $Connection->GetSocket();
  if($Socket eq undef) {   if($Socket eq undef) {
     die "did not get a socket from the connection";      &child_exit(-1, "did not get a socket from the connection");
  } else {   } else {
     &Debug(9,"MakeLondConnection got socket: ".$Socket);      &Debug(9,"MakeLondConnection got socket: ".$Socket);
  }   }
Line 1356  sub ClientRequest { Line 1406  sub ClientRequest {
   
 }  }
   
   #
   #     Accept a connection request for a client (lonc child) and
   #    start up an event watcher to keep an eye on input from that 
   #    Event.  This can be called both from NewClient and from
   #    ChildProcess if we are started in DieWhenIdle mode.
   # Parameters:
   #    $socket       - The listener socket.
   # Returns:
   #   NONE
   # Side Effects:
   #    An event is made to watch the accepted connection.
   #    Active clients hash is updated to reflect the new connection.
   #    The client connection count is incremented.
   #
   sub accept_client {
       my ($socket) = @_;
   
       Debug(8, "Entering accept for lonc UNIX socket\n");
       my $connection = $socket->accept(); # Accept the client connection.
       Debug(8,"Connection request accepted from "
     .GetPeername($connection, AF_UNIX));
   
   
       my $description = sprintf("Connection to lonc client %d",
         $ClientConnection);
       Debug(9, "Creating event named: ".$description);
       Event->io(cb      => \&ClientRequest,
         poll    => 'r',
         desc    => $description,
         data    => "",
         fd      => $connection);
       $ActiveClients{$connection} = $ClientConnection;
       $ClientConnection++;
   }
   
 =pod  =pod
   
Line 1374  sub NewClient { Line 1458  sub NewClient {
     my $event      = shift; # Get the event parameters.      my $event      = shift; # Get the event parameters.
     my $watcher    = $event->w;       my $watcher    = $event->w; 
     my $socket     = $watcher->fd; # Get the event' socket.      my $socket     = $watcher->fd; # Get the event' socket.
     my $connection = $socket->accept(); # Accept the client connection.  
     Debug(8,"Connection request accepted from "  
   .GetPeername($connection, AF_UNIX));  
   
       &accept_client($socket);
     my $description = sprintf("Connection to lonc client %d",  
       $ClientConnection);  
     Debug(9, "Creating event named: ".$description);  
     Event->io(cb      => \&ClientRequest,  
       poll    => 'r',  
       desc    => $description,  
       data    => "",  
       fd      => $connection);  
     $ActiveClients{$connection} = $ClientConnection;  
     $ClientConnection++;  
 }  }
   
 =pod  =pod
Line 1467  sub SetupLoncListener { Line 1538  sub SetupLoncListener {
     unless ($socket =IO::Socket::UNIX->new(Local  => $SocketName,      unless ($socket =IO::Socket::UNIX->new(Local  => $SocketName,
     Listen => 250,       Listen => 250, 
     Type   => SOCK_STREAM)) {      Type   => SOCK_STREAM)) {
  die "Failed to create a lonc listner socket";   if($I_am_child) {
       &child_exit(-1, "Failed to create a lonc listener socket");
    } else {
       die "Failed to create a lonc listner socket";
    }
     }      }
     return $socket;      return $socket;
 }  }
Line 1499  into the status file. Line 1574  into the status file.
   
 We also use this to reset the retries count in order to allow the  We also use this to reset the retries count in order to allow the
 client to retry connections with a previously dead server.  client to retry connections with a previously dead server.
   
 =cut  =cut
   
 sub ChildStatus {  sub ChildStatus {
Line 1507  sub ChildStatus { Line 1583  sub ChildStatus {
   
     Debug(2, "Reporting child status because : ".$watcher->data);      Debug(2, "Reporting child status because : ".$watcher->data);
     my $docdir = $perlvar{'lonDocRoot'};      my $docdir = $perlvar{'lonDocRoot'};
     my $fh = IO::File->new(">>$docdir/lon-status/loncstatus.txt");      
     print $fh $$."\t".$RemoteHost."\t".$Status."\t".      open(LOG,">>$docdir/lon-status/loncstatus.txt");
       flock(LOG,LOCK_EX);
       print LOG $$."\t".$RemoteHost."\t".$Status."\t".
  $RecentLogEntry."\n";   $RecentLogEntry."\n";
     #      #
     #  Write out information about each of the connections:      #  Write out information about each of the connections:
     #      #
     if ($DebugLevel > 2) {      if ($DebugLevel > 2) {
  print $fh "Active connection statuses: \n";   print LOG "Active connection statuses: \n";
  my $i = 1;   my $i = 1;
  print STDERR  "================================= Socket Status Dump:\n";   print STDERR  "================================= Socket Status Dump:\n";
  foreach my $item (keys %ActiveConnections) {   foreach my $item (keys %ActiveConnections) {
     my $Socket = $ActiveConnections{$item}->data;      my $Socket = $ActiveConnections{$item}->data;
     my $state  = $Socket->GetState();      my $state  = $Socket->GetState();
     print $fh "Connection $i State: $state\n";      print LOG "Connection $i State: $state\n";
     print STDERR "---------------------- Connection $i \n";      print STDERR "---------------------- Connection $i \n";
     $Socket->Dump(-1); # Ensure it gets dumped..      $Socket->Dump(-1); # Ensure it gets dumped..
     $i++;      $i++;
  }   }
     }      }
       flock(LOG,LOCK_UN);
       close(LOG);
     $ConnectionRetriesLeft = $ConnectionRetries;      $ConnectionRetriesLeft = $ConnectionRetries;
       UpdateStatus();
 }  }
   
 =pod  =pod
Line 1547  sub SignalledToDeath { Line 1628  sub SignalledToDeath {
     chomp($signal);      chomp($signal);
     Log("CRITICAL", "Abnormal exit.  Child $$ for $RemoteHost "      Log("CRITICAL", "Abnormal exit.  Child $$ for $RemoteHost "
  ."died through "."\"$signal\"");   ."died through "."\"$signal\"");
     LogPerm("F:lonc: $$ on $RemoteHost signalled to death: "      #LogPerm("F:lonc: $$ on $RemoteHost signalled to death: "
     ."\"$signal\"");  #    ."\"$signal\"");
     exit 0;      exit 0;
   
 }  }
   
   =pod
   
 =head2 ToggleDebug  =head2 ToggleDebug
   
 This sub toggles trace debugging on and off.  This sub toggles trace debugging on and off.
Line 1568  sub ToggleDebug { Line 1651  sub ToggleDebug {
   
 }  }
   
   =pod
   
 =head2 ChildProcess  =head2 ChildProcess
   
 This sub implements a child process for a single lonc daemon.  This sub implements a child process for a single lonc daemon.
   Optional parameter:
      $socket  - if provided, this is a socket already open for listen
                 on the client socket. Otherwise, a new listen is set up.
   
 =cut  =cut
   
 sub ChildProcess {  sub ChildProcess {
       #  If we are in DieWhenIdle mode, we've inherited all the
       #  events of our parent and those have to be cancelled or else
       #  all holy bloody chaos will result.. trust me, I already made
       #  >that< mistake.
   
       my $host = GetServerHost();
       foreach my $listener (keys %parent_dispatchers) {
    my $watcher = $parent_dispatchers{$listener};
    my $s       = $watcher->fd;
    if ($listener ne $host) { # Close everyone but me.
       Debug(5, "Closing listen socket for $listener");
       $s->close();
    }
    Debug(5, "Killing watcher for $listener");
   
    $watcher->cancel();
    delete($parent_dispatchers{$listener});
   
       }
   
       #  kill off the parent's signal handlers too!  
       #
   
       for my $handler (keys %parent_handlers) {
    my $watcher = $parent_handlers{$handler};
    $watcher->cancel();
    delete($parent_handlers{$handler});
       }
   
       $I_am_child    = 1; # Seems like in spite of it all I may still getting
                                   # parent event dispatches.. flag I'm a child.
   
   
     #      #
     #  Signals must be handled by the Event framework...      #  Signals must be handled by the Event framework...
 #      #
   
     Event->signal(signal   => "QUIT",      Event->signal(signal   => "QUIT",
   cb       => \&SignalledToDeath,    cb       => \&SignalledToDeath,
Line 1596  sub ChildProcess { Line 1715  sub ChildProcess {
   cb       => \&ToggleDebug,    cb       => \&ToggleDebug,
   data     => "INT");    data     => "INT");
   
           #  Figure out if we got passed a socket or need to open one to listen for
     my $socket =  SetupLoncListener();      #  client requests.
   
       my ($socket) = @_;
       if (!$socket) {
   
    $socket =  SetupLoncListener();
       }
       #  Establish an event to listen for client connection requests.
   
   
     Event->io(cb   => \&NewClient,      Event->io(cb   => \&NewClient,
       poll => 'r',        poll => 'r',
       desc => 'Lonc Listener Unix Socket',        desc => 'Lonc Listener Unix Socket',
       fd   => $socket);        fd   => $socket);
           
     $Event::Debuglevel = $DebugLevel;      $Event::DebugLevel = $DebugLevel;
           
     Debug(9, "Making initial lond connection for ".$RemoteHost);      Debug(9, "Making initial lond connection for ".$RemoteHost);
   
 # Setup the initial server connection:  # Setup the initial server connection:
           
      # &MakeLondConnection(); // let first work requirest do it.       # &MakeLondConnection(); // let first work request do it.
   
       #  If We are in diwhenidle, need to accept the connection since the
       #  event may  not fire.
   
       if ($DieWhenIdle) {
    &accept_client($socket);
       }
   
     Debug(9,"Entering event loop");      Debug(9,"Entering event loop");
     my $ret = Event::loop(); #  Start the main event loop.      my $ret = Event::loop(); #  Start the main event loop.
           
           
     die "Main event loop exited!!!";      &child_exit (-1,"Main event loop exited!!!");
 }  }
   
 #  Create a new child for host passed in:  #  Create a new child for host passed in:
   
 sub CreateChild {  sub CreateChild {
     my $host = shift;      my ($host, $socket) = @_;
   
     my $sigset = POSIX::SigSet->new(SIGINT);      my $sigset = POSIX::SigSet->new(SIGINT);
     sigprocmask(SIG_BLOCK, $sigset);      sigprocmask(SIG_BLOCK, $sigset);
Line 1639  sub CreateChild { Line 1773  sub CreateChild {
  ShowStatus("Connected to ".$RemoteHost);   ShowStatus("Connected to ".$RemoteHost);
  $SIG{INT} = 'DEFAULT';   $SIG{INT} = 'DEFAULT';
  sigprocmask(SIG_UNBLOCK, $sigset);   sigprocmask(SIG_UNBLOCK, $sigset);
  ChildProcess; # Does not return.   if(defined $socket) {
       &ChildProcess($socket);
    } else {
       ChildProcess; # Does not return.
    }
       }
   }
   
   # parent_client_connection:
   #    Event handler that processes client connections for the parent process.
   #    This sub is called when the parent is listening on a socket and
   #    a connection request arrives.  We must:
   #     Start a child process to accept the connection request.
   #     Kill our listen on the socket.
   # Parameter:
   #    event       - The event object that was created to monitor this socket.
   #                  event->w->fd is the socket.
   # Returns:
   #    NONE
   #
   sub parent_client_connection {
       if ($I_am_child) {
    #  Should not get here, but seem to anyway:
    &Debug(5," Child caught parent client connection event!!");
    my ($event) = @_;
    my $watcher = $event->w;
    $watcher->cancel(); # Try to kill it off again!!
       } else {
    &Debug(9, "parent_client_connection");
    my ($event)   = @_;
    my $watcher   = $event->w;
    my $socket    = $watcher->fd;
   
    # Lookup the host associated with this socket:
   
    my $host = $listening_to{$socket};
   
    # Start the child:
   
   
   
    &Debug(9,"Creating child for $host (parent_client_connection)");
    &CreateChild($host, $socket);
   
    # Clean up the listen since now the child takes over until it exits.
   
    $watcher->cancel(); # Nolonger listening to this event
    delete($listening_to{$socket});
    delete($parent_dispatchers{$host});
    $socket->close();
       }
   }
   
   # parent_listen:
   #    Opens a socket and starts a listen for the parent process on a client UNIX
   #    domain socket.
   #
   #    This involves:
   #       Creating a socket for listen.
   #       Removing any socket lock file
   #       Adding an event handler for this socket becoming readable
   #         To the parent's event dispatcher.
   # Parameters:
   #    loncapa_host    - LonCAPA cluster name of the host represented by the client
   #                      socket.
   # Returns:
   #    NONE
   #
   sub parent_listen {
       my ($loncapa_host) = @_;
       Debug(5, "parent_listen: $loncapa_host");
   
       my $socket    = &SetupLoncListener($loncapa_host);
       $listening_to{$socket} = $loncapa_host;
       if (!$socket) {
    die "Unable to create a listen socket for $loncapa_host";
       }
       
       my $lock_file = &GetLoncSocketPath($loncapa_host).".lock";
       unlink($lock_file); # No problem if it doesn't exist yet [startup e.g.]
   
       my $watcher = Event->io(cb    => \&parent_client_connection,
         poll  => 'r',
         desc  => "Parent listener unix socket ($loncapa_host)",
         fd    => $socket);
       $parent_dispatchers{$loncapa_host} = $watcher;
   
   }
   
   
   # listen_on_all_unix_sockets:
   #    This sub initiates a listen on all unix domain lonc client sockets.
   #    This will be called in the case where we are trimming idle processes.
   #    When idle processes are trimmed, loncnew starts up with no children,
   #    and only spawns off children when a connection request occurs on the
   #    client unix socket.  The spawned child continues to run until it has
   #    been idle a while at which point it eventually exits and once more
   #    the parent picks up the listen.
   #
   #  Parameters:
   #      NONE
   #  Implicit Inputs:
   #    The configuration file that has been read in by LondConnection.
   #  Returns:
   #     NONE
   #
   sub listen_on_all_unix_sockets {
       Debug(5, "listen_on_all_unix_sockets");
       my $host_iterator      =   &LondConnection::GetHostIterator();
       while (!$host_iterator->end()) {
    my $host_entry_ref =   $host_iterator->get();
    my $host_name      = $host_entry_ref->[3];
    Debug(9, "Listen for $host_name");
    &parent_listen($host_name);
    $host_iterator->next();
       }
   }
   
   #   server_died is called whenever a child process exits.
   #   Since this is dispatched via a signal, we must process all
   #   dead children until there are no more left.  The action
   #   is to:
   #      - Remove the child from the bookeeping hashes
   #      - Re-establish a listen on the unix domain socket associated
   #        with that host.
   # Parameters:
   #    The event, but we don't actually care about it.
   sub server_died {
       &Debug(9, "server_died called...");
       
       while(1) { # Loop until waitpid nowait fails.
    my $pid = waitpid(-1, WNOHANG);
    if($pid <= 0) {
       return; # Nothing left to wait for.
    }
    # need the host to restart:
   
    my $host = $ChildHash{$pid};
    if($host) { # It's for real...
       &Debug(9, "Caught sigchild for $host");
       delete($ChildHash{$pid});
       delete($HostToPid{$host});
       &parent_listen($host);
   
    } else {
       &Debug(5, "Caught sigchild for pid not in hosts hash: $pid");
    }
     }      }
   
 }  }
   
 #  #
 #  Parent process logic pass 1:  #  Parent process logic pass 1:
 #   For each entry in the hosts table, we will  #   For each entry in the hosts table, we will
Line 1694  LondConnection::ReadConfig; Line 1975  LondConnection::ReadConfig;
 my $HostIterator = LondConnection::GetHostIterator;  my $HostIterator = LondConnection::GetHostIterator;
   
 if ($DieWhenIdle) {  if ($DieWhenIdle) {
     print "Die when idle socket monitoring is not yet implemented\n";      $RemoteHost = "[parent]";
     exit(-1);      &listen_on_all_unix_sockets();
 } else {  } else {
           
     while (! $HostIterator->end()) {      while (! $HostIterator->end()) {
Line 1713  $RemoteHost = "Parent Server"; Line 1994  $RemoteHost = "Parent Server";
   
 ShowStatus("Parent keeping the flock");  ShowStatus("Parent keeping the flock");
   
 #  
 #   Set up parent signals:  
 #  
   
 if ($DieWhenIdle) {  if ($DieWhenIdle) {
     print "Die when idle main processing not yet implemented";      # We need to setup a SIGChild event to handle the exit (natural or otherwise)
     exit (-1);      # of the children.
   
       Event->signal(cb       => \&server_died,
      desc     => "Child exit handler",
      signal   => "CHLD");
   
   
       # Set up all the other signals we set up.  We'll vector them off to the
       # same subs as we would for DieWhenIdle false and, if necessary, conditionalize
       # the code there.
   
       $parent_handlers{INT} = Event->signal(cb       => \&Terminate,
     desc     => "Parent INT handler",
     signal   => "INT");
       $parent_handlers{TERM} = Event->signal(cb       => \&Terminate,
      desc     => "Parent TERM handler",
      signal   => "TERM");
       $parent_handlers{HUP}  = Event->signal(cb       => \&Restart,
      desc     => "Parent HUP handler.",
      signal   => "HUP");
       $parent_handlers{USR1} = Event->signal(cb       => \&CheckKids,
      desc     => "Parent USR1 handler",
      signal   => "USR1");
       $parent_handlers{USR2} = Event->signal(cb       => \&UpdateKids,
      desc     => "Parent USR2 handler.",
      signal   => "USR2");
       
       #  Start procdesing events.
   
       $Event::DebugLevel = $DebugLevel;
       Debug(9, "Parent entering event loop");
       my $ret = Event::loop();
       die "Main Event loop exited: $ret";
   
   
 } else {  } else {
       #
       #   Set up parent signals:
       #
           
     $SIG{INT}  = \&Terminate;      $SIG{INT}  = \&Terminate;
     $SIG{TERM} = \&Terminate;       $SIG{TERM} = \&Terminate; 
Line 1762  sub CheckKids { Line 2077  sub CheckKids {
     my $now=time;      my $now=time;
     my $local=localtime($now);      my $local=localtime($now);
     print $fh "LONC status $local - parent $$ \n\n";      print $fh "LONC status $local - parent $$ \n\n";
       foreach my $host (keys %parent_dispatchers) {
    print $fh "LONC Parent process listening for $host\n";
       }
     foreach my $pid (keys %ChildHash) {      foreach my $pid (keys %ChildHash) {
  Debug(2, "Sending USR1 -> $pid");   Debug(2, "Sending USR1 -> $pid");
  kill 'USR1' => $pid; # Tell Child to report status.   kill 'USR1' => $pid; # Tell Child to report status.
  sleep 1; # Wait so file doesn't intermix.  
     }      }
   
 }  }
   
 =pod  =pod
Line 1799  sub UpdateKids { Line 2117  sub UpdateKids {
   
     Log("INFO", "Updating connections via SIGUSR2");      Log("INFO", "Updating connections via SIGUSR2");
   
     #  Just in case we need to kill our own lonc, we wait a few seconds to      #  I'm not sure what I was thinking in the first implementation.
     #  give it a chance to receive and relay lond's response to the       # someone will have to work hard to convince me the effect is any
     #  re-init command.      # different than Restart, especially now that we don't start up 
     #      # per host servers automatically, may as well just restart.
       # The down side is transactions that are in flight will get timed out
       # (lost unless they are critical).
   
     sleep(2); # Wait a couple of seconds.      &Restart();
   
     my %hosts;                   # Indexed by loncapa hostname, value=ip.  
       
     # Need to re-read  the host table:  
       
       
     LondConnection::ReadConfig();  
     my $I = LondConnection::GetHostIterator;  
     while (! $I->end()) {  
  my $item = $I->get();  
  $hosts{$item->[0]} = $item->[4];  
  $I->next();  
     }  
   
     #  The logic below is written for clarity not for efficiency.  
     #  Since I anticipate that this function is only rarely called, that's  
     #  appropriate.  There are certainly ways to combine the loops below,  
     #  and anyone wishing to obscure the logic is welcome to go for it.  
     #  Note that we don't re-direct sigchild.  Instead we do what's needed  
     #  to the data structures that keep track of children to ensure that  
     #  when sigchild is honored, no new child is born.  
     #  
   
     #  For each existing child; if it's host doesn't exist, kill the child.  
   
     foreach my $child (keys %ChildHash) {  
  my $oldhost = $ChildHash{$child};  
  if (!(exists $hosts{$oldhost})) {  
     Log("CRITICAL", "Killing child for $oldhost  host no longer exists");  
     delete $ChildHash{$child};  
     delete $HostToPid{$oldhost};  
     kill 'QUIT' => $child;  
  }  
     }  
     # For each remaining existing child; if it's host's ip has changed,  
     # Restart the child on the new IP.  
   
     foreach my $child (keys %ChildHash) {  
  my $oldhost = $ChildHash{$child};  
  my $oldip   = $HostHash{$oldhost};  
  if ($hosts{$oldhost} ne $oldip) {  
   
     # kill the old child.  
   
     Log("CRITICAL", "Killing child for $oldhost host ip has changed...");  
     delete $ChildHash{$child};  
     delete $HostToPid{$oldhost};  
     kill 'QUIT' => $child;  
   
     # Do the book-keeping needed to start a new child on the  
     # new ip.  
   
     $HostHash{$oldhost} = $hosts{$oldhost};  
     CreateChild($oldhost);  
  }  
     }  
     # Finally, for each new host, not in the host hash, create a  
     # enter the host and create a new child.  
     # Force a status display of any existing process.  
   
     foreach my $host (keys %hosts) {  
  if(!(exists $HostHash{$host})) {  
     Log("INFO", "New host $host discovered in hosts.tab...");  
     $HostHash{$host} = $hosts{$host};  
     CreateChild($host);  
  } else {  
     kill 'HUP' => $HostToPid{$host};    # status display.  
  }  
     }  
 }  }
   
   
Line 1892  sub Restart { Line 2144  sub Restart {
     Log("CRITICAL", "Restarting");      Log("CRITICAL", "Restarting");
     my $execdir = $perlvar{'lonDaemons'};      my $execdir = $perlvar{'lonDaemons'};
     unlink("$execdir/logs/lonc.pid");      unlink("$execdir/logs/lonc.pid");
     exec("$execdir/loncnew");      exec("$executable");
 }  }
   
 =pod  =pod
Line 1936  sub really_kill_them_all_dammit Line 2188  sub really_kill_them_all_dammit
  unlink("$execdir/logs/lonc.pid");   unlink("$execdir/logs/lonc.pid");
     }      }
 }  }
   
 =pod  =pod
   
 =head1 Terminate  =head1 Terminate

Removed from v.1.60  
changed lines
  Added in v.1.76


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