newnode.in 13.5 KB
Newer Older
1 2 3 4
#!/usr/bin/perl -w

#
# EMULAB-COPYRIGHT
5
# Copyright (c) 2003, 2004, 2005 University of Utah and the Flux Group.
6 7 8 9 10 11 12
# All rights reserved.
#

#
# newnode - a script for moving nodes from the new_ tables into production.
#

13
use lib '@prefix@/lib';
14 15

use libdb;
16
use libxmlrpc;
Robert Ricci's avatar
Robert Ricci committed
17
use English;
18
use Getopt::Std;
19 20 21

use strict;

22 23 24 25 26
my $TB         = "@prefix@";
my $ELABINELAB = @ELABINELAB@;
my $RPCSERVER  = "@OUTERBOSS_NODENAME@";
my $RPCPORT    = "@OUTERBOSS_XMLRPCPORT@";
my $RPCCERT    = "@OUTERBOSS_SSLCERTNAME@";
27 28

my $switchmac = "$TB/sbin/switchmac";
29 30 31 32
my $os_load = "$TB/bin/os_load";
my $os_select = "$TB/bin/os_select";
my $newnode_reboot = "$TB/sbin/newnode_reboot";
my $named_setup = "$TB/sbin/named_setup";
33 34 35
my $nalloc = "$TB/bin/nalloc";
my $nfree = "$TB/bin/nfree";
my $dhcpd_makeconf = "$TB/sbin/dhcpd_makeconf";
36
my $exports_setup = "$TB/sbin/exports_setup";
37 38 39

my $dhcpd_rc = "/usr/local/etc/rc.d/2.dhcpd.sh";

40
my $sudo = "/usr/local/bin/sudo -S";
41 42

#
43
# MFS to boot the nodes into initially
44
#
45 46
my $MFS_INITIAL = TB_OSID_FREEBSD_MFS();
my $OS_OPSNODE  = "OPSNODE-BSD";
47 48

#
49
# Initial event system state to put the nodes into
50
#
51 52
my $STATE_INITIAL = TBDB_NODESTATE_SHUTDOWN;
my $STATE_OPSNODE = TBDB_NODESTATE_ISUP;
53 54 55 56 57 58

#
# Number of vnodes to create for each physical node
#
my $NUM_VNODES = 50;

59 60 61 62 63
#
# Default number of vnodes that can be hosted on any pnode
#
my $DEFAULT_PCVM_COUNT = 20;

64 65 66 67 68 69 70 71
#
# Experiments we might put nodes into
#
my $PID_HWDOWN = NODEDEAD_PID();
my $EID_HWDOWN = NODEDEAD_EID();
my $PID_OPSNODE = TBOPSPID();
my $EID_OPSNODE = "opsnodes";

72 73
# Turn off line buffering on output
$| = 1;
74

75 76 77 78
if (!TBAdmin()) {
    die "Sorry, only testbed administrators can run this script!\n";
}

79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95
#
# Handle command-line options
#
my $force_unconnected_interfaces = 0;
my %options = ();
if (! getopts("f", \%options)) {
    usage();
}

if ($options{f}) {
    $force_unconnected_interfaces = 1;
}

sub usage() {
    die "Usage: $0 [-f] <node_id> ...\n";
}

96 97 98 99 100 101
#
# We seem to have to do this so that ssh gets proper root permissions to read
# the key file. Argh.
#
$UID = $EUID = 0;

102
if (@ARGV < 1) {
103
    usage();
104 105
}

106 107 108
#
# The user has to be able to run sudo, so they can restart dhcpd.
#
109
if (system "$sudo /bin/pwd < /dev/null") {
110 111 112
    die "You must be able to sudo to root to use this script\n";
}

113 114 115 116 117 118 119 120 121 122
#
# For vnodes - figure out the jail IP base
#
my $IPBASE;
if (TBDB_JAILIPBASE =~ /^(\d+).(\d+).(\d+).(\d+)/) {
    $IPBASE = "$1.$2";
} else {
    die "Problem with JAILIPBASE\n"
}

123 124 125 126 127
my @node_ids = @ARGV;

#
# Now, loop through the nodes given, and add each one
#
Robert Ricci's avatar
Robert Ricci committed
128
my (@succeeded_nodes, @reboot_IPs);
129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144
NODE: foreach my $node_id (@node_ids) {
    my $query_result;

    #
    # Check to make sure said node does not already exist!
    #
    $query_result = DBQueryFatal("SELECT node_id FROM nodes WHERE " .
	"node_id='$node_id'");
    if ($query_result->num_rows()) {
	warn "Node $node_id failed: a node with that name already exists!\n";
	next NODE;
    }

    #
    # Grab information about the node from the new_nodes table
    #
145
    $query_result = DBQueryFatal("SELECT new_node_id, type, IP, temporary_IP, ".
146
	"floor, building, loc_x, loc_y, contact, phone, room, role " .
147
	"FROM new_nodes WHERE node_id='$node_id'");
148 149 150 151
    if (!$query_result->num_rows()) {
	warn "Node $node_id failed: No pending node with that name exists!\n";
	next NODE;
    }
152
    my ($new_node_id, $type, $IP, $tempIP, $floor, $building, $loc_x, $loc_y,
153
	$contact, $phone, $room, $role) = $query_result->fetchrow();
154

155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175
    #
    # Make sure they've given it an IP address that looks valid
    #
    if ($IP !~ /^(\d+).(\d+).(\d+).(\d+)/) {
	warn "Node $node_id has an invalid IP address ($IP) - skipping\n";
	next NODE;
    }

    #
    # Make sure some other node doesn't already have this IP address on its
    # control interface
    #
    $query_result = DBQueryFatal("SELECT node_id FROM interfaces WHERE " .
	"IP='$IP' AND role='" . TBDB_IFACEROLE_CONTROL() . "'");
    if ($query_result->num_rows()) {
	my ($existing_node_id) = $query_result->fetchrow();
	warn "Node $node_id failed: $existing_node_id is already using IP " .
		"address $IP\n";
	next NODE;
    }

176 177 178 179
    #
    # Make sure that the new node is of a valid type, and grab a few other
    # things to fill in as initial values
    #
180
    $query_result = DBQueryFatal("SELECT control_iface FROM node_types " .
181 182 183 184 185
	"WHERE type='$type'");
    if (!$query_result->num_rows()) {
	warn "Node $node_id failed: Type $type does not exist!\n";
	next NODE;
    }
186
    my ($control_iface) = $query_result->fetchrow();
187 188 189 190

    #
    # Grab the node's MACs from the new_interfaces table
    #
191
    $query_result = DBQueryFatal("SELECT card, MAC, interface_type, " .
192
	"switch_id, switch_card, switch_port, cable, len, role " .
193
	"FROM new_interfaces WHERE new_node_id='$new_node_id'");
194 195 196 197 198 199
    if (!$query_result->num_rows()) {
	warn "Node $node_id failed: Must have at least one interface!\n";
	next NODE;
    }

    my %interfaces;
200
    while (my ($card, $MAC, $iface_type, $switch_id, $switch_card,
201
	    $switch_port, $cable, $len, $ifrole) = $query_result->fetchrow()) {
202 203 204
	#
	# Get some more information about this interface type
	#
205
	my $iface_query = DBQueryFatal("SELECT max_speed, full_duplex, " .
206
	    "connector FROM interface_types WHERE type='$iface_type'");
207
	if (!$iface_query->num_rows()) {
208
	    warn "Node $node_id failed: Interface $card is of unknown type " .
209
	    	"$iface_type\n";
210 211 212
	    next NODE;
	}

213
	my ($max_speed, $full_duplex, $connector) = $iface_query->fetchrow();
214

215 216 217 218
	#
	# Do a sanity check - make sure that we have a switch recorded for all
	# experimental interfaces
	#
219
	unless ($force_unconnected_interfaces || ($role eq "ctrlnode")) {
220
	    my $iface = "eth$card";
221 222
	    if (($iface ne $control_iface) && ($connector !~ /wireless/i)
		&& !$switch_id) {
223 224 225 226 227 228
		warn "Node $node_id failed: Don't know which switch " .
		    "card $iface is connected to\n";
		next NODE;
	    }
	}

229 230 231
	#
	# Stash it away...
	#
232
	$interfaces{$card} = [$MAC, $iface_type, $max_speed, $full_duplex,
233
		$switch_id, $switch_card, $switch_port, $cable, $len, $ifrole];
234 235 236 237 238 239

    }

    #
    # Make up a priority (just used for sorting)
    #
240 241
    $node_id =~ /^(.*\D)(\d+)$/;
    my ($prefix,$nodenum) = ($1, $2);
242
    my $priority;
243 244
    if (defined $nodenum) {
	$priority = $nodenum;
245 246 247 248
    } else {
	$priority = 1;
    }

249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290
    #
    # Figure out which project and experiment this node needs to go into
    #
    my ($pid, $eid);
    if ($role eq "ctrlnode") {
	$pid = $PID_OPSNODE;
	$eid = $EID_OPSNODE;
    } else {
	$pid = $PID_HWDOWN;
	$eid = $EID_HWDOWN;
    }

    #
    # Figure out what OS this node should run
    #
    my $osid;
    if ($role eq "ctrlnode") {
	$osid = $OS_OPSNODE;
    } else {
	$osid = $MFS_INITIAL;
    }

    #
    # Figure out which opmode the OS runs in
    #
    my $result = DBQueryFatal("select op_mode from os_info where " .
	" osid='$osid'");
    if ($result->numrows() != 1) {
	die "Unable to find OS information for $osid\n";
    }
    my ($opmode) = $result->fetchrow();

    #
    # Figure out what state to start the node in
    #
    my $state;
    if ($role eq "ctrlnode") {
	$state = $STATE_OPSNODE;
    } else {
	$state = $STATE_INITIAL;
    }

291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318
    #
    # When operating inside an inner elab, we need an extra bit of information
    # from the outer boss; which the IP of the outer control network. This is
    # strictly for bootinfo which has some really gross hacks in it.
    #
    my $outer_control_ip;
    
    if ($ELABINELAB) {
        libxmlrpc::Config({"server"  => $RPCSERVER,
			   "verbose" => 0,
			   "cert"    => $RPCCERT,
			   "portnum" => $RPCPORT});

	my $rval = libxmlrpc::CallMethod("elabinelab", "newnode_info",
					 {"mac" => $interfaces{0}[0]});

	if (!defined($rval)) {
	    print "XMPRPC server did not return control iface info!\n";
	    next NODE;
	}
	else {
	    #
	    # Grab what we need by searching the interface marked as
	    # the outer control network.
	    #
	    my %iface_array = %{ $rval->{"interfaces"} };
		
	    foreach my $mac (keys(%iface_array)) {
319 320
		my $aref   = $iface_array{$mac};
		my $ifrole = $aref->{"role"};
321

322
		if ($ifrole eq TBDB_IFACEROLE_OUTER_CONTROL()) {
323 324 325 326 327 328 329 330 331 332 333
		    $outer_control_ip = $aref->{"IP"};
		    last;
		}
	    }
	}
	if (!defined($outer_control_ip)) {
	    print "Could not determine outer control interface IP!\n";
	    next NODE;
	}
    }

334 335 336 337 338
    #
    # Okay, time to actually add the node!
    #
    
    DBQueryFatal("INSERT INTO nodes SET node_id='$node_id', type='$type', " .
339 340 341
	"phys_nodeid='$node_id', role='$role', priority=$priority, " .
    	"eventstate='$state', op_mode='$opmode', " .
	"def_boot_osid='$MFS_INITIAL', " .
342 343
	"state_timestamp=unix_timestamp(NOW()), " .
	"op_mode_timestamp=unix_timestamp(NOW())");
344

345 346
    DBQueryFatal("insert into node_hostkeys (node_id) ".
		 "values ('$node_id')");
347 348 349 350 351
    DBQueryFatal("insert into node_status ".
		 "(node_id, status, status_timestamp) ".
		 "values ('$node_id', 'down', now()) ");
    DBQueryFatal("insert into node_activity ".
		 "(node_id) values ('$node_id')");
352

353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382
    #
    # Copy data into the location_info table, if they provided any
    #
    if ($floor || $building || $loc_x || $loc_y || $contact
	|| $phone || $room) {

	#
	# Handle columns that could have come to us NULL (and thus are undef)
	#
	if (defined $floor) {
	    $floor = "'$floor'";
	} else {
	    $floor = "NULL";
	}
	if (defined $building) {
	    $building = "'$building'";
	} else {
	    $building = "NULL";
	}
	if (defined $room) {
	    $room = "'$room'";
	} else {
	    $room = "NULL";
	}

	DBQueryFatal("INSERT INTO location_info SET node_id='$node_id', " .
	    "floor=$floor, building=$building, loc_x=$loc_x, loc_y=$loc_y, ".
	    "contact='$contact', phone='$phone', room=$room;");
    }

383 384 385
    #
    # Add some vnodes
    #
386
    if (defined $nodenum && ($role eq "testnode")) {
387 388 389 390
	my $vtype = $type;
	if (!($vtype =~ s/pc/pcvm/)) {
	    $vtype = "$vtype-vm";
	}
391 392 393 394
	DBQueryFatal("INSERT INTO node_auxtypes set node_id='$node_id', " .
	    "type='pcvm', count=$DEFAULT_PCVM_COUNT");
	DBQueryFatal("INSERT INTO node_auxtypes set node_id='$node_id', " .
	    "type='$vtype', count=$DEFAULT_PCVM_COUNT");
395 396
    }

397
    while (my ($card, $aref) = each %interfaces) {
398
	my ($MAC, $iface_type, $max_speed, $duplex, $switch_id, $switch_card,
399
	    $switch_port, $cable, $len, $ifrole) = @$aref;
400
	my $iface = "eth$card";
401 402 403 404 405
	my $iface_IP;
	my $wire_type;
	my $iface_role;

	# Role from new_interfaces overrides anything we decide here.
406 407
	if (defined($ifrole)) {
	    $iface_role = $ifrole;
408 409
	}
	elsif ($iface eq $control_iface) {
410
	    $iface_role = TBDB_IFACEROLE_CONTROL();
411
	}
412 413 414 415 416 417 418 419 420 421 422 423 424
	else {
	    $iface_role = TBDB_IFACEROLE_EXPERIMENT();
	}

	# Okay, now set up some stuff for the inserts.
	if ($iface_role eq TBDB_IFACEROLE_CONTROL()) {
	    $iface_IP  = $IP;
	    $wire_type = "Control";
	}
	elsif ($iface_role eq TBDB_IFACEROLE_OUTER_CONTROL()) {
	    $iface_IP  = $outer_control_ip;
	    $wire_type = "OuterControl";
	}
425 426
	else {
	    # Experimental interfaces start with no speed set. 
Kirk Webb's avatar
Kirk Webb committed
427
	    $max_speed = 0;
428 429
	    $iface_IP  = "";
	    $wire_type = "Node";
430
	}
431

432 433 434
	DBQueryFatal("INSERT INTO interfaces SET node_id='$node_id', " .
	    "card=$card, port=1, mac='$MAC', IP='$iface_IP', " .
	    "interface_type='$iface_type', iface='$iface', " .
Kirk Webb's avatar
Kirk Webb committed
435
	    "current_speed='$max_speed', duplex=$duplex, role='$iface_role'");
436

437 438 439 440
	if (!$switch_id) {
	    print "No switch found for ${node_id}:$iface - skipping\n";
	    next;
	}
441 442 443 444 445 446 447 448

	my $cable_len = "";
	if ($cable) {
	    $cable_len .= ", cable=$cable";
	}
	if ($len) {
	    $cable_len .= ", len=$len";
	}
449 450 451
	DBQueryFatal("INSERT INTO wires SET type='$wire_type', " .
	    "node_id1='$node_id', card1=$card, port1=1, " .
	    "node_id2='$switch_id', card2='$switch_card', " .
452
	    "port2='$switch_port' $cable_len");
453 454 455
    }

    #
456 457
    # Put it into hwdown for now - I would put them in reloading, but I'm
    # afriad the reload_daemon might do the wrong thing to them
458
    #
459
    system "$nalloc $pid $eid $node_id";
460 461 462 463

    #
    # Remove the node from the new_ tables
    #
464
    DBQueryFatal("DELETE FROM new_nodes WHERE new_node_id=$new_node_id");
465
    DBQueryFatal("DELETE FROM new_interfaces WHERE new_node_id=$new_node_id");
466

467
    print "$node_id succesfully added!\n";
468 469

    push @succeeded_nodes, $node_id;
Robert Ricci's avatar
Robert Ricci committed
470 471 472 473 474 475 476

    #
    # Only reboot regular nodes, not ops nodes
    #
    if ($role ne "ctrlnode") {
	push @reboot_IPs, $tempIP;
    }
477 478 479 480 481 482 483 484 485 486 487 488
}

#
# No point in restarting dhcpd, etc. if there are no nodes that succeeded
#
if (!@succeeded_nodes) {
    die "No nodes succeeded, exiting early\n";
}

#
# Re-generate dhcpd.conf
#
489
print "Re-generating dhcpd.conf\n";
490 491 492
if (system("$dhcpd_makeconf -i")) {
    die "Unable to open regenerate dhcpd config file\n";
}
493 494 495 496 497 498 499 500 501 502 503 504

print "Restarting dhcpd: $sudo $dhcpd_rc stop\n";
my $sudo_rv = system "$sudo $dhcpd_rc stop";
if ($sudo_rv) {
    die "Error stopping dhcpd - return value was $sudo_rv\n";
}
sleep 2;
print "Restarting dhcpd: $sudo $dhcpd_rc start\n";
$sudo_rv = system "$sudo $dhcpd_rc start";
if ($sudo_rv) {
    die "Error starting dhcpd - return value was $sudo_rv\n";
}
505

506 507 508 509 510
print "Setting up nameserver\n";
my $named_rv = system "$named_setup";
if ($named_rv) {
    die "Error running named_setup - return value was $named_rv\n";
}
511

512 513 514 515 516 517 518 519 520
#
# Before we boot nodes into the MFS, we have to make sure they can mount
# NFS filesystems
#
print "Running exports_setup\n";
my $exports_rv = system "$exports_setup";
if ($exports_rv) {
    warn "WARNING - exports_setup returned $exports_rv";
}
521

522 523 524 525
#
# Start rebooting nodes
#
print "Rebooting nodes...\n";
Robert Ricci's avatar
Robert Ricci committed
526
foreach my $IP (@reboot_IPs) {
527 528
    print "Rebooting $IP\n";
    system "$newnode_reboot $IP\n";
529 530
}

531 532 533 534 535
print "\n\n";
print "Finished - when you are satisifed that the nodes are working\n";
print "correctly, use nfree on boss to free them from the emulab-ops/hwdown\n";
print "experiment.\n";

536
#
537 538 539
# TODO -
#    disable interfaces
#    console setup
540
#