bootman.py

   1 #!/usr/bin/python
   2
   3 # Attempt to reboot a node in debug state.
   4
   5 import plc
   6 api = plc.getAuthAPI()
   7
   8 import sys
   9 import os
  10 import const
  11
  12 from getsshkeys import SSHKnownHosts
  13
  14 import subprocess
  15 import time
  16 import database
  17 import moncommands
  18 from sets import Set
  19
  20 import ssh.pxssh as pxssh
  21 import ssh.fdpexpect as fdpexpect
  22 import ssh.pexpect as pexpect
  23 from unified_model import *
  24 from emailTxt import mailtxt
  25 from nodeconfig import network_config_to_str
  26 import traceback
  27 import config
  28
  29 import signal
  30 class Sopen(subprocess.Popen):
  31         def kill(self, signal = signal.SIGTERM):
  32                 os.kill(self.pid, signal)
  33
  34 #from Rpyc import SocketConnection, Async
  35 from Rpyc import SocketConnection, Async
  36 from Rpyc.Utils import *
  37 fb = None
  38
  39 def get_fbnode(node):
  40         global fb
  41         if fb is None:
  42                 fb = database.dbLoad("findbad")
  43         fbnode = fb['nodes'][node]['values']
  44         return fbnode
  45
  46 class NodeConnection:
  47         def __init__(self, connection, node, config):
  48                 self.node = node
  49                 self.c = connection
  50                 self.config = config
  51
  52         def get_boot_state(self):
  53                 if self.c.modules.os.path.exists('/tmp/source'):
  54                         return "dbg"
  55                 elif self.c.modules.os.path.exists('/vservers'):
  56                         return "boot"
  57                 else:
  58                         return "unknown"
  59
  60         def get_dmesg(self):
  61                 self.c.modules.os.system("dmesg > /var/log/dmesg.bm.log")
  62                 download(self.c, "/var/log/dmesg.bm.log", "log/dmesg.%s.log" % self.node)
  63                 log = open("log/dmesg.%s.log" % self.node, 'r')
  64                 return log
  65
  66         def get_bootmanager_log(self):
  67                 download(self.c, "/tmp/bm.log", "log/bm.%s.log.gz" % self.node)
  68                 os.system("zcat log/bm.%s.log.gz > log/bm.%s.log" % (self.node, self.node))
  69                 log = open("log/bm.%s.log" % self.node, 'r')
  70                 return log
  71
  72         def dump_plconf_file(self):
  73                 c = self.c
  74                 self.c.modules.sys.path.append("/tmp/source/")
  75                 self.c.modules.os.chdir('/tmp/source')
  76
  77                 log = c.modules.BootManager.log('/tmp/new.log')
  78                 bm = c.modules.BootManager.BootManager(log,'boot')
  79
  80                 BootManagerException = c.modules.Exceptions.BootManagerException
  81                 InitializeBootManager = c.modules.BootManager.InitializeBootManager
  82                 ReadNodeConfiguration = c.modules.BootManager.ReadNodeConfiguration
  83                 bm_continue = True
  84
  85                 InitializeBootManager.Run(bm.VARS, bm.LOG)
  86                 try: ReadNodeConfiguration.Run(bm.VARS, bm.LOG)
  87                 except Exception, x:
  88                         bm_continue = False
  89                         print "   ERROR:", x
  90                         print "   Possibly, unable to find valid configuration file"
  91
  92                 if bm_continue and self.config and not self.config.quiet:
  93                         for key in bm.VARS.keys():
  94                                 print key, " == ", bm.VARS[key]
  95                 else:
  96                         if self.config and not self.config.quiet: print "   Unable to read Node Configuration"
  97
  98
  99         def compare_and_repair_nodekeys(self):
 100                 c = self.c
 101                 self.c.modules.sys.path.append("/tmp/source/")
 102                 self.c.modules.os.chdir('/tmp/source')
 103
 104                 log = c.modules.BootManager.log('/tmp/new.log')
 105                 bm = c.modules.BootManager.BootManager(log,'boot')
 106
 107                 BootManagerException = c.modules.Exceptions.BootManagerException
 108                 InitializeBootManager = c.modules.BootManager.InitializeBootManager
 109                 ReadNodeConfiguration = c.modules.BootManager.ReadNodeConfiguration
 110                 bm_continue = True
 111
 112                 plcnode = api.GetNodes({'hostname': self.node}, None)[0]
 113
 114                 InitializeBootManager.Run(bm.VARS, bm.LOG)
 115                 try: ReadNodeConfiguration.Run(bm.VARS, bm.LOG)
 116                 except Exception, x:
 117                         bm_continue = False
 118                         print "exception"
 119                         print x
 120                         print "   Possibly, unable to find valid configuration file"
 121
 122                 if bm_continue:
 123                         print "   NODE: %s" % bm.VARS['NODE_KEY']
 124                         print "   PLC : %s" % plcnode['key']
 125
 126                         if bm.VARS['NODE_KEY'] == plcnode['key']:
 127                                 return True
 128                         else:
 129                                 if api.UpdateNode(self.node, {'key': bm.VARS['NODE_KEY']}):
 130                                         print "   Successfully updated NODE_KEY with PLC"
 131                                         return True
 132                                 else:
 133                                         return False
 134
 135                         #for key in bm.VARS.keys():
 136                         #       print key, " == ", bm.VARS[key]
 137                 else:
 138                         print "   Unable to retrieve NODE_KEY"
 139
 140         def bootmanager_running(self):
 141                 if self.c.modules.os.path.exists('/tmp/BM_RUNNING'):
 142                         return True
 143                 else:
 144                         return False
 145
 146         def set_nodestate(self, state='boot'):
 147                 return api.UpdateNode(self.node, {'boot_state' : state})
 148
 149         def restart_node(self, state='boot'):
 150                 api.UpdateNode(self.node, {'boot_state' : state})
 151
 152                 pflags = PersistFlags(self.node, 1*60*60*24, db='restart_persistflags')
 153                 if not pflags.getRecentFlag('gentlekill'):
 154                         print "   Killing all slice processes... : %s" %  self.node
 155                         cmd_slicekill = "ls -d /proc/virtual/[0-9]* | awk -F '/' '{print $4}' | xargs -I{} /usr/sbin/vkill -s 9 --xid {} -- 0"
 156                         self.c.modules.os.system(cmd_slicekill)
 157                         cmd = """ shutdown -r +1 & """
 158                         print "   Restarting %s : %s" % ( self.node, cmd)
 159                         self.c.modules.os.system(cmd)
 160
 161                         pflags.setRecentFlag('gentlekill')
 162                         pflags.save()
 163                 else:
 164                         print "   Restarting with sysrq 'sub' %s" % self.node
 165                         cmd = """ (sleep 5; echo 's' > /proc/sysrq-trigger; echo 'u' > /proc/sysrq-trigger; echo 'b' > /proc/sysrq-trigger ) & """
 166                         self.c.modules.os.system(cmd)
 167
 168                 return
 169
 170         def restart_bootmanager(self, forceState):
 171
 172                 self.c.modules.os.chdir('/tmp/source')
 173                 if self.c.modules.os.path.exists('/tmp/BM_RUNNING'):
 174                         print "   BootManager is already running: try again soon..."
 175                 else:
 176                         print "   Starting 'BootManager.py %s' on %s " % (forceState, self.node)
 177                         cmd = "( touch /tmp/BM_RUNNING ;  " + \
 178                               "  python ./BootManager.py %s &> server.log < /dev/null ; " + \
 179                                   "  rm -f /tmp/BM_RUNNING " + \
 180                                   ") &"
 181                         cmd = cmd % forceState
 182                         self.c.modules.os.system(cmd)
 183
 184                 return
 185
 186
 187 import random
 188 class PlanetLabSession:
 189         globalport = 22000 + int(random.random()*1000)
 190
 191         def __init__(self, node, nosetup, verbose):
 192                 self.verbose = verbose
 193                 self.node = node
 194                 self.port = None
 195                 self.nosetup = nosetup
 196                 self.command = None
 197                 self.setup_host()
 198
 199         def get_connection(self, config):
 200                 return NodeConnection(SocketConnection("localhost", self.port), self.node, config)
 201
 202         def setup_host(self):
 203                 self.port = PlanetLabSession.globalport
 204                 PlanetLabSession.globalport = PlanetLabSession.globalport + 1
 205
 206                 args = {}
 207                 args['port'] = self.port
 208                 args['user'] = 'root'
 209                 args['hostname'] = self.node
 210                 args['monitordir'] = config.MONITOR_SCRIPT_ROOT
 211                 ssh_port = 22
 212
 213                 if self.nosetup:
 214                         print "Skipping setup"
 215                         return
 216
 217                 # COPY Rpyc files to host
 218                 cmd = "rsync -qv -az -e ssh %(monitordir)s/Rpyc/ %(user)s@%(hostname)s:Rpyc 2> /dev/null" % args
 219                 if self.verbose: print cmd
 220                 # TODO: Add timeout
 221                 timeout = 120
 222                 localos = moncommands.CMD()
 223
 224                 ret = localos.system(cmd, timeout)
 225                 print ret
 226                 if ret != 0:
 227                         print "\tUNKNOWN SSH KEY FOR %s; making an exception" % self.node
 228                         #print "MAKE EXPLICIT EXCEPTION FOR %s" % self.node
 229                         k = SSHKnownHosts(); k.updateDirect(self.node); k.write(); del k
 230                         ret = localos.system(cmd, timeout)
 231                         print ret
 232                         if ret != 0:
 233                                 print "\tFAILED TWICE"
 234                                 #sys.exit(1)
 235                                 raise Exception("Failed twice trying to login with updated ssh host key")
 236
 237                 t1 = time.time()
 238                 # KILL any already running servers.
 239                 ssh = moncommands.SSH(args['user'], args['hostname'], ssh_port)
 240                 (ov,ev) = ssh.run_noexcept2("""<<\EOF
 241             rm -f out.log
 242             echo "kill server" >> out.log
 243             ps ax | grep Rpyc | grep -v grep | awk '{print $1}' | xargs kill 2> /dev/null ;
 244             echo "export" >> out.log
 245             export PYTHONPATH=$HOME  ;
 246             echo "start server" >> out.log
 247             python Rpyc/Servers/forking_server.py &> server.log &
 248             echo "done" >> out.log
 249 EOF""")
 250                 #cmd = """ssh %(user)s@%(hostname)s """ + \
 251                 #        """'ps ax | grep Rpyc | grep -v grep | awk "{print \$1}" | xargs kill 2> /dev/null' """
 252                 #cmd = cmd % args
 253                 #if self.verbose: print cmd
 254                 ## TODO: Add timeout
 255                 #print localos.system(cmd,timeout)
 256
 257                 ## START a new rpyc server.
 258                 #cmd = """ssh -n %(user)s@%(hostname)s "export PYTHONPATH=\$HOME; """ + \
 259                 #        """python Rpyc/Servers/forking_server.py &> server.log < /dev/null &" """
 260                 #cmd = cmd % args
 261                 #if self.verbose: print cmd
 262                 #print localos.system(cmd,timeout)
 263                 print ssh.ret
 264
 265                 # TODO: Add timeout
 266                 # This was tricky to make synchronous.  The combination of ssh-clients-4.7p1,
 267                 # and the following options seems to work well.
 268                 cmd = """ssh -o ExitOnForwardFailure=yes -o BatchMode=yes """ + \
 269                           """-o PermitLocalCommand=yes -o LocalCommand='echo "READY"' """ + \
 270                           """-o ConnectTimeout=120 """ + \
 271                           """-n -N -L %(port)s:localhost:18812 """ + \
 272                           """%(user)s@%(hostname)s"""
 273                 cmd = cmd % args
 274                 if self.verbose: print cmd
 275                 self.command = Sopen(cmd, shell=True, stdout=subprocess.PIPE)
 276                 # TODO: the read() here may block indefinitely.  Need a better
 277                 # approach therefore, that includes a timeout.
 278                 #ret = self.command.stdout.read(5)
 279                 ret = moncommands.read_t(self.command.stdout, 5)
 280
 281                 t2 = time.time()
 282                 if 'READY' in ret:
 283                         # NOTE: There is still a slight race for machines that are slow...
 284                         self.timeout = 2*(t2-t1)
 285                         print "Sleeping for %s sec" % self.timeout
 286                         time.sleep(self.timeout)
 287                         return
 288
 289                 if self.command.returncode is not None:
 290                         print "Failed to establish tunnel!"
 291                         raise Exception("SSH Tunnel exception : %s %s" % (self.node, self.command.returncode))
 292
 293                 raise Exception("Unknown SSH Tunnel Exception: still running, but did not report 'READY'")
 294
 295         def __del__(self):
 296                 if self.command:
 297                         if self.verbose: print "Killing SSH session %s" % self.port
 298                         self.command.kill()
 299
 300
 301 def steps_to_list(steps):
 302         ret_list = []
 303         for (id,label) in steps:
 304                 ret_list.append(label)
 305         return ret_list
 306
 307 def index_to_id(steps,index):
 308         if index < len(steps):
 309                 return steps[index][0]
 310         else:
 311                 return "done"
 312
 313 def reboot(hostname, config=None, forced_action=None):
 314
 315         # NOTE: Nothing works if the bootcd is REALLY old.
 316         #       So, this is the first step.
 317         fbnode = get_fbnode(hostname)
 318         if fbnode['category'] == "OLDBOOTCD":
 319                 print "...NOTIFY OWNER TO UPDATE BOOTCD!!!"
 320                 args = {}
 321                 args['hostname_list'] = "    %s" % hostname
 322
 323                 m = PersistMessage(hostname, "Please Update Boot Image for %s" % hostname,
 324                                                         mailtxt.newbootcd_one[1] % args, True, db='bootcd_persistmessages')
 325
 326                 loginbase = plc.siteId(hostname)
 327                 m.send([const.PIEMAIL % loginbase, const.TECHEMAIL % loginbase])
 328
 329                 print "\tDisabling %s due to out-of-date BOOTCD" % hostname
 330                 api.UpdateNode(hostname, {'boot_state' : 'disable'})
 331                 return True
 332
 333         node = hostname
 334         print "Creating session for %s" % node
 335         # update known_hosts file (in case the node has rebooted since last run)
 336         if config and not config.quiet: print "...updating known_hosts ssh-rsa key for %s" % node
 337         try:
 338                 k = SSHKnownHosts(); k.update(node); k.write(); del k
 339         except:
 340                 print traceback.print_exc()
 341                 return False
 342
 343         try:
 344                 if config == None:
 345                         session = PlanetLabSession(node, False, True)
 346                 else:
 347                         session = PlanetLabSession(node, config.nosetup, config.verbose)
 348         except Exception, e:
 349                 print "ERROR setting up session for %s" % hostname
 350                 print traceback.print_exc()
 351                 print e
 352                 return False
 353
 354         try:
 355                 conn = session.get_connection(config)
 356         except EOFError:
 357                 # NOTE: sometimes the wait in setup_host() is not long enough.
 358                 # So, here we try to wait a little longer before giving up entirely.
 359                 try:
 360                         time.sleep(session.timeout*4)
 361                         conn = session.get_connection(config)
 362                 except:
 363                         print traceback.print_exc()
 364                         return False
 365
 366         if forced_action == "reboot":
 367                 conn.restart_node('rins')
 368                 return True
 369
 370         boot_state = conn.get_boot_state()
 371         if boot_state == "boot":
 372                 print "...Boot state of %s already completed : skipping..." % node
 373                 return True
 374         elif boot_state == "unknown":
 375                 print "...Unknown bootstate for %s : skipping..."% node
 376                 return False
 377         else:
 378                 pass
 379
 380         if conn.bootmanager_running():
 381                 print "...BootManager is currently running.  Skipping host %s" % node
 382                 return True
 383
 384         #if config != None:
 385         #       if config.force:
 386         #               conn.restart_bootmanager(config.force)
 387         #               return True
 388
 389         # Read persistent flags, tagged on one week intervals.
 390         pflags = PersistFlags(hostname, 3*60*60*24, db='debug_persistflags')
 391
 392
 393         if config and not config.quiet: print "...downloading dmesg from %s" % node
 394         dmesg = conn.get_dmesg()
 395         child = fdpexpect.fdspawn(dmesg)
 396
 397         sequence = []
 398         while True:
 399                 steps = [
 400                         ('scsierror'  , 'SCSI error : <\d+ \d+ \d+ \d+> return code = 0x\d+'),
 401                         ('ioerror'    , 'end_request: I/O error, dev sd\w+, sector \d+'),
 402                         ('ccisserror' , 'cciss: cmd \w+ has CHECK CONDITION  byte \w+ = \w+'),
 403
 404                         ('buffererror', 'Buffer I/O error on device dm-\d, logical block \d+'),
 405
 406                         ('hdaseekerror', 'hda: dma_intr: status=0x\d+ { DriveReady SeekComplete Error }'),
 407                         ('hdacorrecterror', 'hda: dma_intr: error=0x\d+ { UncorrectableError }, LBAsect=\d+, sector=\d+'),
 408
 409                         ('atareadyerror'   , 'ata\d+: status=0x\d+ { DriveReady SeekComplete Error }'),
 410                         ('atacorrecterror' , 'ata\d+: error=0x\d+ { UncorrectableError }'),
 411
 412                         ('sdXerror'   , 'sd\w: Current: sense key: Medium Error'),
 413                         ('ext3error'   , 'EXT3-fs error (device dm-\d+): ext3_find_entry: reading directory #\d+ offset \d+'),
 414
 415                         ('floppytimeout','floppy0: floppy timeout called'),
 416                         ('floppyerror',  'end_request: I/O error, dev fd\w+, sector \d+'),
 417
 418                         # hda: dma_intr: status=0x51 { DriveReady SeekComplete Error }
 419                         # hda: dma_intr: error=0x40 { UncorrectableError }, LBAsect=23331263, sector=23331263
 420
 421                         # floppy0: floppy timeout called
 422                         # end_request: I/O error, dev fd0, sector 0
 423
 424                         # Buffer I/O error on device dm-2, logical block 8888896
 425                         # ata1: status=0x51 { DriveReady SeekComplete Error }
 426                         # ata1: error=0x40 { UncorrectableError }
 427                         # SCSI error : <0 0 0 0> return code = 0x8000002
 428                         # sda: Current: sense key: Medium Error
 429                         #       Additional sense: Unrecovered read error - auto reallocate failed
 430
 431                         # SCSI error : <0 2 0 0> return code = 0x40001
 432                         # end_request: I/O error, dev sda, sector 572489600
 433                 ]
 434                 id = index_to_id(steps, child.expect( steps_to_list(steps) + [ pexpect.EOF ]))
 435                 sequence.append(id)
 436
 437                 if id == "done":
 438                         break
 439
 440         s = Set(sequence)
 441         if config and not config.quiet: print "\tSET: ", s
 442
 443         if len(s) > 1:
 444                 print "...Potential drive errors on %s" % node
 445                 if len(s) == 2 and 'floppyerror' in s:
 446                         print "...Should investigate.  Continuing with node."
 447                 else:
 448                         print "...Should investigate.  Skipping node."
 449                         # TODO: send message related to these errors.
 450                         args = {}
 451                         args['hostname'] = hostname
 452                         args['log'] = conn.get_dmesg().read()
 453
 454                         m = PersistMessage(hostname, mailtxt.baddisk[0] % args,
 455                                                                                  mailtxt.baddisk[1] % args, True, db='hardware_persistmessages')
 456
 457                         loginbase = plc.siteId(hostname)
 458                         m.send([const.PIEMAIL % loginbase, const.TECHEMAIL % loginbase])
 459                         conn.set_nodestate('disable')
 460                         return False
 461
 462         print "...Downloading bm.log from %s" % node
 463         log = conn.get_bootmanager_log()
 464         child = fdpexpect.fdspawn(log)
 465
 466         try:
 467                 if config.collect: return True
 468         except:
 469                 pass
 470
 471         time.sleep(1)
 472
 473         if config and not config.quiet: print "...Scanning bm.log for errors"
 474         action_id = "dbg"
 475         sequence = []
 476         while True:
 477
 478                 steps = [
 479                         ('bminit'               , 'Initializing the BootManager.'),
 480                         ('cfg'                  , 'Reading node configuration file.'),
 481                         ('auth'                 , 'Authenticating node with PLC.'),
 482                         ('getplc'               , 'Retrieving details of node from PLC.'),
 483                         ('update'               , 'Updating node boot state at PLC.'),
 484                         ('hardware'             , 'Checking if hardware requirements met.'),
 485                         ('installinit'  , 'Install: Initializing.'),
 486                         ('installdisk'  , 'Install: partitioning disks.'),
 487                         ('installbootfs', 'Install: bootstrapfs tarball.'),
 488                         ('installcfg'   , 'Install: Writing configuration files.'),
 489                         ('installstop'  , 'Install: Shutting down installer.'),
 490                         ('update2'              , 'Updating node boot state at PLC.'),
 491                         ('installinit2' , 'Install: Initializing.'),
 492                         ('validate'             , 'Validating node installation.'),
 493                         ('rebuildinitrd', 'Rebuilding initrd'),
 494                         ('netcfg'               , 'Install: Writing Network Configuration files.'),
 495                         ('update3'              , 'Updating node configuration.'),
 496                         ('disk'                 , 'Checking for unused disks to add to LVM.'),
 497                         ('update4'              , 'Sending hardware configuration to PLC.'),
 498                         ('debug'                , 'Starting debug mode'),
 499                         ('bmexceptmount', 'BootManagerException during mount'),
 500                         ('bmexceptvgscan', 'BootManagerException during vgscan/vgchange'),
 501                         ('bmexceptrmfail', 'Unable to remove directory tree: /tmp/mnt'),
 502                         ('exception'    , 'Exception'),
 503                         ('nocfg'        , 'Found configuration file planet.cnf on floppy, but was unable to parse it.'),
 504                         ('protoerror'   , 'XML RPC protocol error'),
 505                         ('nodehostname' , 'Configured node hostname does not resolve'),
 506                         ('implementerror', 'Implementation Error'),
 507                         ('readonlyfs'   , '[Errno 30] Read-only file system'),
 508                         ('baddisk'      , "IOError: [Errno 13] Permission denied: '/tmp/mnt/sysimg//vservers/\w+/etc/hosts'"),
 509                         ('noinstall'    , 'notinstalled'),
 510                         ('bziperror'    , 'bzip2: Data integrity error when decompressing.'),
 511                         ('noblockdev'   , "No block devices detected."),
 512                         ('dnserror'     , 'Name or service not known'),
 513                         ('downloadfail' , 'Unable to download main tarball /boot/bootstrapfs-planetlab-i386.tar.bz2 from server.'),
 514                         ('disktoosmall' , 'The total usable disk size of all disks is insufficient to be usable as a PlanetLab node.'),
 515                         ('hardwarerequirefail' , 'Hardware requirements not met'),
 516                         ('mkfsfail'         , 'while running: Running mkfs.ext2 -q  -m 0 -j /dev/planetlab/vservers failed'),
 517                         ('nofilereference', "No such file or directory: '/tmp/mnt/sysimg//vservers/.vref/planetlab-f8-i386/etc/hosts'"),
 518                         ('kernelcopyfail', "cp: cannot stat `/tmp/mnt/sysimg/boot/kernel-boot': No such file or directory"),
 519                         ('chrootfail'   , 'Running chroot /tmp/mnt/sysimg'),
 520                         ('modulefail'   , 'Unable to get list of system modules'),
 521                         ('writeerror'   , 'write error: No space left on device'),
 522                         ('nospace'      , "No space left on device"),
 523                         ('nonode'       , 'Failed to authenticate call: No such node'),
 524                         ('authfail'     , 'Failed to authenticate call: Call could not be authenticated'),
 525                         ('bootcheckfail'     , 'BootCheckAuthentication'),
 526                         ('bootupdatefail'   , 'BootUpdateNode'),
 527                 ]
 528                 list = steps_to_list(steps)
 529                 index = child.expect( list + [ pexpect.EOF ])
 530                 id = index_to_id(steps,index)
 531                 sequence.append(id)
 532
 533                 if id == "exception":
 534                         if config and not config.quiet: print "...Found An Exception!!!"
 535                 elif index == len(list):
 536                         #print "Reached EOF"
 537                         break
 538
 539         s = "-".join(sequence)
 540         print "   FOUND SEQUENCE: ", s
 541
 542         # NOTE: We get or set the flag based on the current sequence identifier.
 543         #  By using the sequence identifier, we guarantee that there will be no
 544         #  frequent loops.  I'm guessing there is a better way to track loops,
 545         #  though.
 546         #if not config.force and pflags.getRecentFlag(s):
 547         #       pflags.setRecentFlag(s)
 548         #       pflags.save()
 549         #       print "... flag is set or it has already run recently. Skipping %s" % node
 550         #       return True
 551
 552         sequences = {}
 553
 554
 555         # restart_bootmanager_boot
 556         for n in ["bminit-cfg-auth-getplc-update-installinit-validate-rebuildinitrd-netcfg-update3-disk-update4-done",
 557                         "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-netcfg-update3-disk-update4-update3-exception-protoerror-update-protoerror-debug-done",
 558                         "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-netcfg-disk-update4-update3-update3-implementerror-bootupdatefail-update-debug-done",
 559
 560                         "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-netcfg-disk-update4-update3-update3-exception-protoerror-update-protoerror-debug-done",
 561
 562                         "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-netcfg-update3-disk-update4-update3-exception-protoerror-update-debug-done",
 563                         "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-netcfg-disk-update4-update3-exception-chrootfail-update-debug-done",
 564                         "bminit-cfg-auth-getplc-update-debug-done",
 565                         "bminit-cfg-auth-getplc-exception-protoerror-update-protoerror-debug-done",
 566                         "bminit-cfg-auth-protoerror-exception-update-protoerror-debug-done",
 567                         "bminit-cfg-auth-protoerror-exception-update-bootupdatefail-authfail-debug-done",
 568                         "bminit-cfg-auth-protoerror-exception-update-debug-done",
 569                         "bminit-cfg-auth-getplc-exception-protoerror-update-debug-done",
 570                         "bminit-cfg-auth-getplc-implementerror-update-debug-done",
 571                         ]:
 572                 sequences.update({n : "restart_bootmanager_boot"})
 573
 574         #       conn.restart_bootmanager('rins')
 575         for n in [ "bminit-cfg-auth-getplc-installinit-validate-exception-modulefail-update-debug-done",
 576                         "bminit-cfg-auth-getplc-update-installinit-validate-exception-modulefail-update-debug-done",
 577                         "bminit-cfg-auth-getplc-installinit-validate-bmexceptmount-exception-noinstall-update-debug-done",
 578                         "bminit-cfg-auth-getplc-update-installinit-validate-bmexceptmount-exception-noinstall-update-debug-done",
 579                         "bminit-cfg-auth-getplc-installinit-validate-bmexceptvgscan-exception-noinstall-update-debug-done",
 580                         "bminit-cfg-auth-getplc-update-installinit-validate-exception-noinstall-update-debug-done",
 581                         "bminit-cfg-auth-getplc-hardware-installinit-installdisk-bziperror-exception-update-debug-done",
 582                         "bminit-cfg-auth-getplc-update-hardware-installinit-installdisk-installbootfs-exception-update-debug-done",
 583                         "bminit-cfg-auth-getplc-update-installinit-validate-bmexceptvgscan-exception-noinstall-update-debug-done",
 584                         "bminit-cfg-auth-getplc-hardware-installinit-installdisk-installbootfs-exception-update-debug-done",
 585                         "bminit-cfg-auth-getplc-update-installinit-validate-rebuildinitrd-netcfg-update3-implementerror-nofilereference-update-debug-done",
 586                         "bminit-cfg-auth-getplc-update-hardware-installinit-installdisk-exception-mkfsfail-update-debug-done",
 587                         "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-exception-chrootfail-update-debug-done",
 588                         "bminit-cfg-auth-getplc-update-installinit-validate-rebuildinitrd-netcfg-disk-update4-exception-chrootfail-update-debug-done",
 589                         "bminit-cfg-auth-getplc-update-hardware-installinit-installdisk-installbootfs-installcfg-installstop-update-installinit-validate-rebuildinitrd-netcfg-disk-update4-update3-update3-kernelcopyfail-exception-update-debug-done",
 590                         "bminit-cfg-auth-getplc-hardware-installinit-installdisk-installbootfs-installcfg-installstop-update-installinit-validate-rebuildinitrd-netcfg-disk-update4-update3-update3-kernelcopyfail-exception-update-debug-done",
 591                         "bminit-cfg-auth-getplc-installinit-validate-exception-noinstall-update-debug-done",
 592                         ]:
 593                 sequences.update({n : "restart_bootmanager_rins"})
 594
 595         # repair_node_keys
 596         sequences.update({"bminit-cfg-auth-bootcheckfail-authfail-exception-update-bootupdatefail-authfail-debug-done": "repair_node_keys"})
 597
 598         #   conn.restart_node('rins')
 599         for n in ["bminit-cfg-auth-getplc-update-installinit-validate-rebuildinitrd-exception-chrootfail-update-debug-done",
 600                         "bminit-cfg-auth-getplc-update-installinit-validate-rebuildinitrd-netcfg-update3-disk-update4-exception-chrootfail-update-debug-done",
 601                         "bminit-cfg-auth-getplc-hardware-installinit-installdisk-installbootfs-installcfg-exception-chrootfail-update-debug-done",
 602                         "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-writeerror-exception-chrootfail-update-debug-done",
 603                         "bminit-cfg-auth-getplc-update-hardware-installinit-exception-bmexceptrmfail-update-debug-done",
 604                         "bminit-cfg-auth-getplc-hardware-installinit-exception-bmexceptrmfail-update-debug-done",
 605                         "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-netcfg-update3-disk-update4-update3-implementerror-bootupdatefail-update-debug-done",
 606                         "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-netcfg-update3-implementerror-readonlyfs-update-debug-done",
 607                         "bminit-cfg-auth-getplc-update-installinit-validate-rebuildinitrd-netcfg-update3-nospace-exception-update-debug-done",
 608                         "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-netcfg-update3-implementerror-nospace-update-debug-done",
 609                         "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-netcfg-update3-implementerror-update-debug-done",
 610                         "bminit-cfg-auth-getplc-update-hardware-installinit-installdisk-installbootfs-exception-downloadfail-update-debug-done",
 611                         ]:
 612                 sequences.update({n : "restart_node_rins"})
 613
 614         #       restart_node_boot
 615         for n in ["bminit-cfg-auth-getplc-implementerror-bootupdatefail-update-debug-done",
 616                          "bminit-cfg-auth-implementerror-bootcheckfail-update-debug-done",
 617                          "bminit-cfg-auth-implementerror-bootcheckfail-update-implementerror-bootupdatefail-done",
 618                          "bminit-cfg-auth-getplc-update-installinit-validate-rebuildinitrd-netcfg-update3-implementerror-nospace-update-debug-done",
 619                          "bminit-cfg-auth-getplc-hardware-installinit-installdisk-installbootfs-exception-downloadfail-update-debug-done",
 620                          ]:
 621                 sequences.update({n: "restart_node_boot"})
 622
 623         # update_node_config_email
 624         for n in ["bminit-cfg-exception-nocfg-update-bootupdatefail-nonode-debug-done",
 625                           "bminit-cfg-exception-update-bootupdatefail-nonode-debug-done",
 626                           "bminit-cfg-auth-bootcheckfail-nonode-exception-update-bootupdatefail-nonode-debug-done",
 627                         ]:
 628                 sequences.update({n : "update_node_config_email"})
 629
 630         for n in [ "bminit-cfg-exception-nodehostname-update-debug-done",
 631                            "bminit-cfg-update-exception-nodehostname-update-debug-done",
 632                         ]:
 633                 sequences.update({n : "nodenetwork_email"})
 634
 635         # update_bootcd_email
 636         for n in ["bminit-cfg-auth-getplc-update-hardware-exception-noblockdev-hardwarerequirefail-update-debug-done",
 637                         "bminit-cfg-auth-getplc-hardware-exception-noblockdev-hardwarerequirefail-update-debug-done",
 638                         "bminit-cfg-auth-getplc-update-hardware-noblockdev-exception-hardwarerequirefail-update-debug-done",
 639                         "bminit-cfg-auth-getplc-hardware-noblockdev-exception-hardwarerequirefail-update-debug-done",
 640                         "bminit-cfg-auth-getplc-hardware-exception-hardwarerequirefail-update-debug-done",
 641                         ]:
 642                 sequences.update({n : "update_bootcd_email"})
 643
 644         for n in [ "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-netcfg-update3-implementerror-nofilereference-update-debug-done",
 645                         ]:
 646                 sequences.update({n: "suspect_error_email"})
 647
 648         # update_hardware_email
 649         sequences.update({"bminit-cfg-auth-getplc-hardware-exception-disktoosmall-hardwarerequirefail-update-debug-done" : "update_hardware_email"})
 650         sequences.update({"bminit-cfg-auth-getplc-hardware-disktoosmall-exception-hardwarerequirefail-update-debug-done" : "update_hardware_email"})
 651
 652         # broken_hardware_email
 653         sequences.update({"bminit-cfg-auth-getplc-update-hardware-exception-hardwarerequirefail-update-debug-done" : "broken_hardware_email"})
 654
 655         # bad_dns_email
 656         for n in [
 657          "bminit-cfg-update-implementerror-bootupdatefail-dnserror-update-implementerror-bootupdatefail-dnserror-done",
 658                 "bminit-cfg-auth-implementerror-bootcheckfail-dnserror-update-implementerror-bootupdatefail-dnserror-done",
 659                 ]:
 660                 sequences.update( { n : "bad_dns_email"})
 661
 662         flag_set = True
 663
 664
 665         if s not in sequences:
 666                 print "   HOST %s" % hostname
 667                 print "   UNKNOWN SEQUENCE: %s" % s
 668
 669                 args = {}
 670                 args['hostname'] = hostname
 671                 args['sequence'] = s
 672                 args['bmlog'] = conn.get_bootmanager_log().read()
 673                 m = PersistMessage(hostname, mailtxt.unknownsequence[0] % args,
 674                                                                          mailtxt.unknownsequence[1] % args, False, db='unknown_persistmessages')
 675                 m.reset()
 676                 m.send(['monitor-list@lists.planet-lab.org'])
 677
 678                 conn.restart_bootmanager('boot')
 679
 680                 # NOTE: Do not set the pflags value for this sequence if it's unknown.
 681                 # This way, we can check it again after we've fixed it.
 682                 flag_set = False
 683
 684         else:
 685
 686                 if   sequences[s] == "restart_bootmanager_boot":
 687                         if config and not config.quiet: print "...Restarting BootManager.py on %s "% node
 688                         conn.restart_bootmanager('boot')
 689                 elif sequences[s] == "restart_bootmanager_rins":
 690                         if config and not config.quiet: print "...Restarting BootManager.py on %s "% node
 691                         conn.restart_bootmanager('rins')
 692                 elif sequences[s] == "restart_node_rins":
 693                         conn.restart_node('rins')
 694                 elif sequences[s] == "restart_node_boot":
 695                         conn.restart_node('boot')
 696                 elif sequences[s] == "repair_node_keys":
 697                         if conn.compare_and_repair_nodekeys():
 698                                 # the keys either are in sync or were forced in sync.
 699                                 # so try to reboot the node again.
 700                                 conn.restart_bootmanager('rins')
 701                                 pass
 702                         else:
 703                                 # there was some failure to synchronize the keys.
 704                                 print "...Unable to repair node keys on %s" % node
 705
 706                 elif sequences[s] == "suspect_error_email":
 707                         args = {}
 708                         args['hostname'] = hostname
 709                         args['sequence'] = s
 710                         args['bmlog'] = conn.get_bootmanager_log().read()
 711                         m = PersistMessage(hostname, "Suspicous error from BootManager on %s" % args,
 712                                                                                  mailtxt.unknownsequence[1] % args, False, db='suspect_persistmessages')
 713                         m.reset()
 714                         m.send(['monitor-list@lists.planet-lab.org'])
 715
 716                         conn.restart_bootmanager('boot')
 717
 718                 elif sequences[s] == "update_node_config_email":
 719                         print "...Sending message to UPDATE NODE CONFIG"
 720                         args = {}
 721                         args['hostname'] = hostname
 722                         m = PersistMessage(hostname,  mailtxt.plnode_cfg[0] % args,  mailtxt.plnode_cfg[1] % args,
 723                                                                 True, db='nodeid_persistmessages')
 724                         loginbase = plc.siteId(hostname)
 725                         m.send([const.PIEMAIL % loginbase, const.TECHEMAIL % loginbase])
 726                         conn.dump_plconf_file()
 727                         conn.set_nodestate('disable')
 728
 729                 elif sequences[s] == "nodenetwork_email":
 730                         print "...Sending message to LOOK AT NODE NETWORK"
 731                         args = {}
 732                         args['hostname'] = hostname
 733                         args['bmlog'] = conn.get_bootmanager_log().read()
 734                         m = PersistMessage(hostname,  mailtxt.plnode_network[0] % args,  mailtxt.plnode_cfg[1] % args,
 735                                                                 True, db='nodenet_persistmessages')
 736                         loginbase = plc.siteId(hostname)
 737                         m.send([const.PIEMAIL % loginbase, const.TECHEMAIL % loginbase])
 738                         conn.dump_plconf_file()
 739                         conn.set_nodestate('disable')
 740
 741                 elif sequences[s] == "update_bootcd_email":
 742                         print "...NOTIFY OWNER TO UPDATE BOOTCD!!!"
 743                         import getconf
 744                         args = {}
 745                         args.update(getconf.getconf(hostname)) # NOTE: Generates boot images for the user:
 746                         args['hostname_list'] = "%s" % hostname
 747
 748                         m = PersistMessage(hostname, "Please Update Boot Image for %s" % hostname,
 749                                                                 mailtxt.newalphacd_one[1] % args, True, db='bootcd_persistmessages')
 750
 751                         loginbase = plc.siteId(hostname)
 752                         m.send([const.PIEMAIL % loginbase, const.TECHEMAIL % loginbase])
 753
 754                         print "\tDisabling %s due to out-of-date BOOTCD" % hostname
 755                         conn.set_nodestate('disable')
 756
 757                 elif sequences[s] == "broken_hardware_email":
 758                         # MAKE An ACTION record that this host has failed hardware.  May
 759                         # require either an exception "/minhw" or other manual intervention.
 760                         # Definitely need to send out some more EMAIL.
 761                         print "...NOTIFYING OWNERS OF BROKEN HARDWARE on %s!!!" % hostname
 762                         # TODO: email notice of broken hardware
 763                         args = {}
 764                         args['hostname'] = hostname
 765                         args['log'] = conn.get_dmesg().read()
 766                         m = PersistMessage(hostname, mailtxt.baddisk[0] % args,
 767                                                                                  mailtxt.baddisk[1] % args, True, db='hardware_persistmessages')
 768
 769                         loginbase = plc.siteId(hostname)
 770                         m.send([const.PIEMAIL % loginbase, const.TECHEMAIL % loginbase])
 771                         conn.set_nodestate('disable')
 772
 773                 elif sequences[s] == "update_hardware_email":
 774                         print "...NOTIFYING OWNERS OF MINIMAL HARDWARE FAILURE on %s!!!" % hostname
 775                         args = {}
 776                         args['hostname'] = hostname
 777                         args['bmlog'] = conn.get_bootmanager_log().read()
 778                         m = PersistMessage(hostname, mailtxt.minimalhardware[0] % args,
 779                                                                                  mailtxt.minimalhardware[1] % args, True, db='minhardware_persistmessages')
 780
 781                         loginbase = plc.siteId(hostname)
 782                         m.send([const.PIEMAIL % loginbase, const.TECHEMAIL % loginbase])
 783                         conn.set_nodestate('disable')
 784
 785                 elif sequences[s] == "bad_dns_email":
 786                         print "...NOTIFYING OWNERS OF DNS FAILURE on %s!!!" % hostname
 787                         args = {}
 788                         try:
 789                                 node = api.GetNodes(hostname)[0]
 790                                 net = api.GetNodeNetworks(node['nodenetwork_ids'])[0]
 791                         except:
 792                                 print traceback.print_exc()
 793                                 # TODO: api error. skip email, b/c all info is not available,
 794                                 # flag_set will not be recorded.
 795                                 return False
 796                         nodenet_str = network_config_to_str(net)
 797
 798                         args['hostname'] = hostname
 799                         args['network_config'] = nodenet_str
 800                         args['nodenetwork_id'] = net['nodenetwork_id']
 801                         m = PersistMessage(hostname, mailtxt.baddns[0] % args,
 802                                                                                  mailtxt.baddns[1] % args, True, db='baddns_persistmessages')
 803
 804                         loginbase = plc.siteId(hostname)
 805                         m.send([const.PIEMAIL % loginbase, const.TECHEMAIL % loginbase])
 806                         conn.set_nodestate('disable')
 807
 808         if flag_set:
 809                 pflags.setRecentFlag(s)
 810                 pflags.save()
 811
 812         return True
 813
 814
 815 # MAIN -------------------------------------------------------------------
 816
 817 def main():
 818         import parser as parsermodule
 819         parser = parsermodule.getParser()
 820
 821         parser.set_defaults(child=False, collect=False, nosetup=False, verbose=False,
 822                                                 force=None, quiet=False)
 823         parser.add_option("", "--child", dest="child", action="store_true",
 824                                                 help="This is the child mode of this process.")
 825         parser.add_option("", "--force", dest="force", metavar="boot_state",
 826                                                 help="Force a boot state passed to BootManager.py.")
 827         parser.add_option("", "--quiet", dest="quiet", action="store_true",
 828                                                 help="Extra quiet output messages.")
 829         parser.add_option("", "--verbose", dest="verbose", action="store_true",
 830                                                 help="Extra debug output messages.")
 831         parser.add_option("", "--nonet", dest="nonet", action="store_true",
 832                                                 help="Do not setup the network, use existing log files to re-run a test pass.")
 833         parser.add_option("", "--collect", dest="collect", action="store_true",
 834                                                 help="No action, just collect dmesg, and bm.log")
 835         parser.add_option("", "--nosetup", dest="nosetup", action="store_true",
 836                                                 help="Do not perform the orginary setup phase.")
 837
 838         parser = parsermodule.getParser(['nodesets', 'defaults'], parser)
 839         config = parsermodule.parse_args(parser)
 840
 841         if config.nodelist:
 842                 nodes = config.getListFromFile(config.nodelist)
 843         elif config.node:
 844                 nodes = [ config.node ]
 845         else:
 846                 parser.print_help()
 847                 sys.exit(1)
 848
 849         for node in nodes:
 850                 reboot(node, config)
 851
 852 if __name__ == "__main__":
 853         main()