Merge remote branch 'central/master'

db/GNUmakefile.in

@@ -80,7 +80,6 @@ boss-install: $(addprefix $(INSTALL_BINDIR)/, $(BIN_SCRIPTS)) \
         $(addprefix $(INSTALL_DIR)/opsdir/lib/, $(USERLIBS))
 
 post-install: 
-	chown root $(INSTALL_SBINDIR)/subboss_dhcpd_makeconf
 	chown root $(INSTALL_SBINDIR)/genelists
 	chmod u+s $(INSTALL_SBINDIR)/genelists
 	chown root $(INSTALL_SBINDIR)/dhcpd_makeconf

db/Node.pm.in

@@ -592,6 +592,46 @@ sub IsReserved($)
     return 1;
 }
 
+
+sub GetSubboss($$)
+{
+    my ($self, $service, $subboss_id) = @_;
+
+    return 0
+	if (! ref($self));
+
+    my $ref;
+
+    if (defined $self->{"SUBBOSSES"}) {
+	my $ref = $self->{"SUBBOSSES"}->{$service};
+    }
+
+    if (!defined $ref) {
+	my $nodeid = $self->node_id();
+
+	my $query_result =
+	    DBQueryWarn("select * from subbosses " .
+			"where node_id='$nodeid' and " .
+			"service = '$service'");
+
+	return 0
+	    if (!$query_result);
+	return 0
+	    if (!$query_result->numrows);
+
+	if (!defined($self->{"SUBBOSSES"})) {
+	    $self->{"SUBBOSSES"} = {};
+	}
+
+	$ref = $self->{"SUBBOSSES"}->{$service} =
+	    $query_result->fetchrow_hashref();
+    }
+
+    $$subboss_id = $ref->{'subboss_id'};
+
+    return 0;
+}
+
 #
 # Flush the reserve info so it reloads.
 #

protogeni/scripts/GNUmakefile.in

@@ -8,17 +8,22 @@ SRCDIR		= @srcdir@
 TESTBED_SRCDIR	= @top_srcdir@
 OBJDIR		= ../..
 SUBDIR		= protogeni/scripts
+ISMAINSITE	= @TBMAINSITE@
 
 include $(OBJDIR)/Makeconf
 
 SBIN_STUFF	= cleanupslice
 PSBIN_STUFF	= register_resources expire_daemon gencrl postcrl \
-		  createcerts initsite addauthority getcacerts \
+		  initsite addauthority getcacerts \
 		  gencrlbundle shutdownslice remauthority listusage \
 		  update reregister cleanupticket listhistory \
 		  register_sliver sa_daemon genadmincredential \
           genclrcredentials genallow_extcred
 
+ifeq ($(ISMAINSITE),1)
+SBIN_STUFF     += ch_daemon
+endif
+
 # These scripts installed setuid, with sudo. 
 SETUID_BIN_SCRIPTS   = 
 SETUID_SBIN_SCRIPTS  = cleanupslice

protogeni/scripts/ch_daemon.in

+#!/usr/bin/perl -w
+#
+# GENIPUBLIC-COPYRIGHT
+# Copyright (c) 2008-2010 University of Utah and the Flux Group.
+# All rights reserved.
+#
+use strict;
+use English;
+use Getopt::Std;
+
+#
+# Look for things that need to be expired and resources released. This is
+# incredibly primitive at the moment.
+#
+sub usage()
+{
+    print "Usage: ch_daemon [-d] [-n]\n";
+    exit(1);
+}
+my $optlist   = "dn";
+my $debug     = 0;
+my $impotent  = 0;
+
+#
+# Configure variables
+#
+my $TB		  = "@prefix@";
+my $TBOPS         = "@TBOPSEMAIL@";
+my $TBLOGS        = "@TBLOGSEMAIL@";
+my $PGENIDOMAIN   = "@PROTOGENI_DOMAIN@";
+my $PGENISUPPORT  = @PROTOGENI_SUPPORT@;
+my $PGENICLRHOUSE = @PROTOGENI_CLEARINGHOUSE@;
+my $LOGFILE       = "$TB/log/ch_daemon.log";
+my $EMULAB_PEMFILE = "$TB/etc/genich.pem";
+my $SLEEP_INTERVAL = 60;
+my $PROBE_INTERVAL = (60 * 60 * 24);
+
+# un-taint path
+$ENV{'PATH'} = '/bin:/usr/bin:/usr/local/bin:/usr/site/bin';
+delete @ENV{'IFS', 'CDPATH', 'ENV', 'BASH_ENV'};
+
+# Protos
+sub fatal($);
+sub notify($);
+
+#
+# Probe interval countdown. Set to zero so it runs right away.
+#
+my $probe_countdown = 0;
+	  
+#
+# Turn off line buffering on output
+#
+$| = 1; 
+
+if ($UID != 0) {
+    fatal("Must be root to run this script\n");
+}
+
+#
+# Exit if not a protogeni site.
+#
+if (! $PGENISUPPORT) {
+    exit(0);
+}
+
+#
+# Check args early so we get the right DB.
+#
+my %options = ();
+if (! getopts($optlist, \%options)) {
+    usage();
+}
+if (defined($options{"d"})) {
+    $debug = 1;
+}
+if (defined($options{"n"})) {
+    $impotent = 1;
+}
+
+# Do this early so that we talk to the right DB.
+use vars qw($GENI_DBNAME);
+$GENI_DBNAME = "geni-ch";
+
+# Load the Testbed support stuff.
+use lib "@prefix@/lib";
+require GeniDB;
+require GeniUtil;
+require GeniCertificate;
+require GeniAuthority;
+require libGeni;
+use libtestbed;
+use emutil;
+use POSIX qw(strftime ceil);
+
+#
+# So we know who/what we are acting as.
+#
+my $certificate = GeniCertificate->LoadFromFile($EMULAB_PEMFILE);
+if (!defined($certificate)) {
+    fatal("Could not get uuid/urn from $EMULAB_PEMFILE\n");
+}
+Genixmlrpc->SetContext(Genixmlrpc->Context($certificate));
+Genixmlrpc->SetTimeout(10);
+$ENV{'MYUUID'} = $certificate->uuid();
+$ENV{'MYURN'} = "urn:publicid:IDN+@OURDOMAIN@+authority+ch";
+
+if (CheckDaemonRunning("ch_daemon")) {
+    fatal("Not starting another ch daemon!");
+}
+# Go to ground.
+if (! $debug) {
+    if (TBBackGround($LOGFILE)) {
+	exit(0);
+    }
+}
+if (MarkDaemonRunning("ch_daemon")) {
+    fatal("Could not mark daemon as running!");
+}
+#
+# Setup a signal handler for newsyslog.
+#
+sub handler()
+{
+    my $SAVEEUID = $EUID;
+    
+    $EUID = 0;
+    ReOpenLog($LOGFILE);
+    $EUID = $SAVEEUID;
+}
+$SIG{HUP} = \&handler
+    if (!$debug);
+
+GeniUtil::FlipToGeniUser();
+
+# Do not wait long.
+Genixmlrpc->SetTimeout(10);
+
+#
+# Probe CMs
+#
+sub ProbeCMs()
+{
+    my @authorities = ();
+    my $up	    = "Up:\n";
+    my $down        = "Down:\n";
+    my $disabled    = "Disabled:\n";
+    
+    if (GeniAuthority->ListAll(\@authorities)) {
+	fatal("Could not get authorities list from the DB");
+    }
+    foreach my $authority (@authorities) {
+	my $urn = $authority->urn();
+	
+	next
+	    if ($authority->type() ne "cm");
+
+	if ($authority->disabled()) {
+	    $disabled .= "$urn\n";
+	    next;
+	}
+
+	print "Contacting $urn\n";
+	my $version = $authority->Version();
+	if (!defined($version)) {
+	    print "Error getting version from $urn\n";
+	    $down .= "$urn\n";
+	}
+	else {
+	    print "  Running version $version\n";
+	    $up .= "$urn, version $version\n";
+	}
+    }
+    SENDMAIL("geni-dev-utah\@flux.utah.edu",
+	     "ClearingHouse Daemon Message",
+	     "This is an automated message from the ClearingHouse Daemon\n\n".
+	     $up . "\n" . $down . "\n" . $disabled . "\n", $TBOPS);
+}
+
+while (1) {
+    next
+	if (NoLogins());
+
+    print "Running at ".
+	POSIX::strftime("20%y-%m-%d %H:%M:%S", localtime()) . "\n";
+
+    $probe_countdown -= $SLEEP_INTERVAL;
+    if ($probe_countdown <= 0) {
+	ProbeCMs();
+	$probe_countdown = $PROBE_INTERVAL;
+    }
+
+    # Be certain stale info is gone.
+    GeniUtil::FlushCaches();
+
+    sleep($SLEEP_INTERVAL);
+}
+exit(0);
+
+sub fatal($)
+{
+    my ($msg) = @_;
+
+    #
+    # Send a message to the testbed list. 
+    #
+    SENDMAIL($TBOPS,
+	     "ProtoGENI CH daemon died",
+	     $msg,
+	     $TBOPS);
+
+    MarkDaemonStopped("ch_daemon");
+    die("*** $0:\n".
+	"    $msg\n");
+}
+
+sub notify($)
+{
+    my ($msg) = @_;
+
+    print "$msg\n";
+    SENDMAIL($TBOPS, "ClearingHouse Daemon Message", $msg, $TBOPS);
+}
+

rc.d/3.testbed.sh.in

@@ -81,6 +81,11 @@ case "$1" in
 			@prefix@/sbin/protogeni/sa_daemon
 		fi
 
+		if [ -x @prefix@/sbin/protogeni/ch_daemon ]; then
+			echo -n " ch_daemon"
+			@prefix@/sbin/protogeni/ch_daemon
+		fi
+
 		if [ -x @prefix@/sbin/checknodes_daemon  ]; then
 			echo -n " checknodes_daemon"
 			@prefix@/sbin/checknodes_daemon
@@ -144,6 +149,9 @@ case "$1" in
 		if [ -r /var/run/sa_daemon.pid ]; then
 			kill `cat /var/run/sa_daemon.pid`
 		fi
+		if [ -r /var/run/ch_daemon.pid ]; then
+			kill `cat /var/run/ch_daemon.pid`
+		fi
 		if [ -r /var/run/checknodes.pid ]; then
 			kill `cat /var/run/checknodes.pid`
 		fi

tbsetup/GNUmakefile.in

@@ -43,7 +43,7 @@ SBIN_STUFF	= resetvlans console_setup.proxy sched_reload named_setup \
 		  elabinelab snmpit.proxy panic node_attributes \
 		  nfstrace plabinelab smbpasswd_setup smbpasswd_setup.proxy \
 		  rmproj snmpit.proxynew snmpit.proxyv2 pool_daemon \
-		  checknodes_daemon
+		  checknodes_daemon subboss_frisbeelauncher_wrapper
 ifeq ($(ISMAINSITE),1)
 SBIN_STUFF     += repos_daemon
 endif
@@ -95,7 +95,7 @@ SETUID_BIN_SCRIPTS   = node_reboot eventsys_control tarfiles_setup savelogs \
 SETUID_SBIN_SCRIPTS  = mkproj rmgroup mkgroup frisbeelauncher frisbeeimage \
 			rmuser idleswap named_setup exports_setup \
 			sfskey_update setgroups newnode_reboot vnode_setup \
-			elabinelab nfstrace rmproj
+			elabinelab nfstrace rmproj subboss_frisbeelauncher_wrapper
 SETUID_LIBX_SCRIPTS  = console_setup spewlogfile
 
 ifeq ($(SYSTEM),FreeBSD)
@@ -231,6 +231,8 @@ endif
 	chmod u+s $(INSTALL_SBINDIR)/mkgroup
 	chown root $(INSTALL_SBINDIR)/frisbeelauncher
 	chmod u+s $(INSTALL_SBINDIR)/frisbeelauncher
+	chown root $(INSTALL_SBINDIR)/subboss_frisbeelauncher_wrapper
+	chmod u+s $(INSTALL_SBINDIR)/subboss_frisbeelauncher_wrapper
 	chown root $(INSTALL_SBINDIR)/frisbeeimage
 	chmod u+s $(INSTALL_SBINDIR)/frisbeeimage
 	chown root $(INSTALL_SBINDIR)/rmuser

tbsetup/libosload.pm.in

@@ -45,6 +45,7 @@ my $PROJROOT    = "@PROJROOT_DIR@";
 my $MAXRETRIES  = 1;
 
 my $FRISBEELAUNCHER = "$TB/sbin/frisbeelauncher";
+my $SUBBOSS_FRISBEELAUNCHER = "$TB/sbin/subboss_frisbeelauncher_wrapper";
 my $osselect	    = "$TB/bin/os_select";
 my $TBUISP	    = "$TB/bin/tbuisp";
 
@@ -452,6 +453,7 @@ sub osload ($$) {
 	my $class = $nodeobject->class();
 	my $isremote = $nodeobject->isremotenode();
 	my $isvirtnode = $nodeobject->isvirtnode();
+	my $subboss;
 	my $reload_mode;
 	my $reload_func;
 	my $reboot_required;
@@ -465,6 +467,7 @@ sub osload ($$) {
 	    $reload_mode = "Frisbee";
 	    $reload_func = \&SetupReloadFrisbee;
 	    $reboot_required = !$noreboot; # Reboot unless $noreboot flag set
+	    $nodeobject->GetSubboss('frisbee', \$subboss);
 
 	    if (defined($nodeflags{$node}) 
 		&& defined($nodeflags{$node}{'noreboot'})) {
@@ -506,6 +509,7 @@ sub osload ($$) {
 	    'zerofree' => $zerofree,
 	    'prepare'  => $prepare,
 	    'maxwait'  => $maxwait,
+	    'subboss' => $subboss,
 	    'isremote' => $isremote,
 	    'isvirtnode' => $isvirtnode
 	};
@@ -981,6 +985,7 @@ sub SetupReloadFrisbee($)
     my $zerofree      = $reload_info->{'zerofree'};
     my $prepare       = $reload_info->{'prepare'};
     my $isvirtnode    = $reload_info->{'isvirtnode'};
+    my $subboss       = $reload_info->{'subboss'};
     my $osid          = TBNodeDiskloadOSID($node);
 
     #
@@ -1016,6 +1021,25 @@ sub SetupReloadFrisbee($)
     }
 
     foreach my $imageid (@$imageids) {
+	if (defined $subboss) {
+		my $image = Image->Lookup($imageid);
+		my $filename = $image->filename();
+		my @image_stats = stat $filename;
+
+		my $query_result = DBQueryWarn("select sync from subboss_images where " .
+			    "subboss_id = '$subboss' and imageid = '$imageid' and sync = 1");
+
+		if (!$query_result || !$query_result->numrows) {
+			system("$SUBBOSS_FRISBEELAUNCHER " .  ($debug ? "-d ": "") .
+				"$subboss " . "$imageid " . "\"$filename\" " .
+				stat[7] . ' ' . stat[9]);
+			if ($?) {
+			    tberror "Subboss Frisbee Launcher ($imageid on $subboss) failed!";
+			    return -1;
+			}
+			return 0;
+		}
+	}
 	system("$FRISBEELAUNCHER " . ($debug ? "-d ": "") . "$imageid");
 	if ($?) {
 	    tberror "Frisbee Launcher ($imageid) failed!";

tbsetup/libreboot.pm.in

@@ -843,12 +843,13 @@ sub RebootVNode($$) {
 	exit(GeniEmulab->StartSlivers($experiment, [ $nodeobj ]));
     }
 
-    my $addargs = "";
+    my $addargs = "-t ";	# Turn on timestamps.
     if ($plab) {
 	$addargs .= "-p ";
     }
     elsif ($jailed) {
-	$addargs .= "-j ";
+	# Use virtual control net routes. 
+	$addargs .= "-jV ";
     }
     else {
 	$addargs .= "-i ";

tbsetup/pool_daemon.in

@@ -15,17 +15,19 @@ use Data::Dumper;
 #
 sub usage()
 {
-    print STDOUT "Usage: pool_daemon [-d]\n" .
+    print STDOUT "Usage: pool_daemon [-d] [nsfile]\n" .
 	"Use the -d option to prevent daemonization\n";
     exit(-1);
 }
-my $optlist = "dne";
+my $optlist = "dne1";
 my $debug    = 0;
 my $impotent = 0;
 my $killme   = 0;
 my $nofree   = 1;
 my $gotlock  = 0;
 my $mailsent = 0;
+my $oneshot  = 0;
+my $nsfile;
 
 #
 # Configure variables
@@ -35,6 +37,7 @@ my $TBOPS    = "@TBOPSEMAIL@";
 my $BOSSNODE = "@BOSSNODE@";
 my $logfile  = "$TB/log/poollog";
 my $SWAPEXP  = "$TB/bin/swapexp";
+my $tmpfile  = "/tmp/pool-$$.nsfile";
 
 # Testbed Support library
 use lib "@prefix@/lib";
@@ -87,29 +90,35 @@ my %options = ();
 if (! getopts($optlist, \%options)) {
     usage();
 }
-if (@ARGV != 0) {
-    usage();
-}
 if (defined($options{"d"})) {
     $debug = 1;
 }
+if (defined($options{"1"})) {
+    $oneshot = 1;
+}
 if (defined($options{"n"})) {
     $impotent = 1;
 }
 if (defined($options{"e"})) {
     $nofree = 1;
 }
+usage()
+    if (@ARGV > 1);
+$nsfile = $ARGV[0]
+    if (@ARGV == 1);
+$oneshot = 1
+    if (defined($nsfile));
 
 if (!$impotent && CheckDaemonRunning("pool_daemon")) {
     fatal("Not starting another pool daemon!");
 }
 # Go to ground.
-if (! ($debug || $impotent)) {
+if (! ($oneshot || $debug || $impotent)) {
     if (TBBackGround($logfile)) {
 	exit(0);
     }
 }
-if (!$impotent && MarkDaemonRunning("pool_daemon")) {
+if (!($impotent || $oneshot) && MarkDaemonRunning("pool_daemon")) {
     fatal("Could not mark daemon as running!");
 }
 #
@@ -120,7 +129,7 @@ sub handler()
     ReOpenLog($logfile);
 }
 $SIG{HUP} = \&handler
-    if (!($debug || $impotent));
+    if (!($oneshot || $debug || $impotent));
 
 print "Pool Daemon starting... pid $$, at ".`date`;
 
@@ -226,6 +235,12 @@ while (!$killme) {
 	print "Skipping this loop cause the experiment is swapped\n";
 	goto loop;
     }
+    if (defined($nsfile)) {
+	print "Copying $nsfile to $tmpfile\n";
+	system("/bin/cp -f $nsfile $tmpfile") == 0
+	    or fatal("Could not copy nsfile");
+	goto skipns;
+    }
     
     my @nodelist = $experiment->NodeList();
     my %inuse    = ();
@@ -309,8 +324,6 @@ while (!$killme) {
     # Generate a new NS file. Be nice to not have to this, but not
     # having an NS file could confuse things for the web interface.
     #
-    my $tmpfile = "/tmp/pool-$$.nsfile";
-
     print "Generating a new NS file in $tmpfile\n";
     
     if (!open(NS, ">$tmpfile")) {
@@ -355,6 +368,7 @@ while (!$killme) {
     print NS "\$ns rtproto Static\n";
     print NS "\$ns run\n";
     close(NS);
+  skipns:
     chmod(0775, $tmpfile);
 
     last
@@ -405,6 +419,8 @@ while (!$killme) {
 	
 	$gotlock = 0;
     }
+    last
+	if ($oneshot);
     
     # Use a long period; we do not want the pool to change too fast.
     sleep(120);
@@ -458,7 +474,9 @@ sub fatal($)
 {
     my ($msg) = @_;
 
-    SENDMAIL($TBOPS, "Pool Daemon Died", $msg, $TBOPS);
+    SENDMAIL($TBOPS, "Pool Daemon Died", $msg, $TBOPS)
+	if (!($impotent || $oneshot));
+    
     $mailsent = 1;
     cleanup();
     die($msg);
@@ -481,13 +499,13 @@ sub cleanup()
     }
 
     MarkDaemonStopped("pool_daemon")
-	if (!$impotent);
+	if (!($impotent || $oneshot));
 }
 
 END {
     my $exitcode = $?;
 
-    if ($exitcode && !$mailsent) {
+    if ($exitcode && !($mailsent || $impotent || $oneshot)) {
 	SENDMAIL($TBOPS, "Pool Daemon Died",
 		 "Please look at $logfile", $TBOPS);
     }

tbsetup/subboss_frisbeeimage.in

@@ -24,7 +24,7 @@ my $debug    = 1;
 #
 my $TB		= "@prefix@";
 my $TBOPS       = "@TBOPSEMAIL@";
-my $RPCSERVER   = "@OUTERBOSS_NODENAME@";
+my $RPCSERVER   = "@BOSSNODE@";
 my $RPCPORT     = "@OUTERBOSS_XMLRPCPORT@";
 my $RPCCERT     = "@OUTERBOSS_SSLCERTNAME@";
 

tbsetup/subboss_frisbeelauncher.in

@@ -12,6 +12,8 @@ use Sys::Syslog;
 use English;
 use Socket;
 use Errno qw(EADDRINUSE);
+use File::Basename;
+use File::Path;
 
 #
 # This also kills a running frisbee.
@@ -34,7 +36,7 @@ my $MY_IP;
 my $BASEADDR    = "@FRISEBEEMCASTADDR@";
 my $BASEPORT    = "@FRISEBEEMCASTPORT@";
 
-my $RPCSERVER   = "@OUTERBOSS_NODENAME@";
+my $RPCSERVER   = "@BOSSNODE@";
 my $RPCPORT     = "@OUTERBOSS_XMLRPCPORT@";
 my $RPCCERT     = "@SUBBOSS_SSLCERTNAME@";
 
@@ -65,12 +67,13 @@ sub ClearPid();
 sub ClearAddress();
 sub Fatal($);
 sub debug($);
-sub GetImageInfo($$);
+sub GetLoadAddress($$);
 
 # Defines
 my $FRISBEED	= "$TB/sbin/frisbeed";
 my $LOGFILE	= "$TB/log/frisbeelauncher";
 my $FRISBEEIMAGE= "$TB/sbin/frisbeeimage";
+my $IMAGE_CACHE = "$TB/image_cache";
 my $child_pid   = 0;
 my $dlfilename  = 0;
 
@@ -123,6 +126,7 @@ usage()
 #
 # XXX hack: use map + regex to get around taint check
 my ($imageid, $filename, $size, $mtime) = map { /(.*)/; $1 } @ARGV;
+$filename = "$IMAGE_CACHE/$filename";
 
 #
 # Get node ID via tmcc
@@ -154,6 +158,13 @@ my $naddress;
 my $address;
 my $pid;