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                 emails = plc.getTechEmails(loginbase)
 328                 m.send(emails)
 329
 330                 print "\tDisabling %s due to out-of-date BOOTCD" % hostname
 331                 api.UpdateNode(hostname, {'boot_state' : 'disable'})
 332                 return True
 333
 334         node = hostname
 335         print "Creating session for %s" % node
 336         # update known_hosts file (in case the node has rebooted since last run)
 337         if config and not config.quiet: print "...updating known_hosts ssh-rsa key for %s" % node
 338         try:
 339                 k = SSHKnownHosts(); k.update(node); k.write(); del k
 340         except:
 341                 print traceback.print_exc()
 342                 return False
 343
 344         try:
 345                 if config == None:
 346                         session = PlanetLabSession(node, False, True)
 347                 else:
 348                         session = PlanetLabSession(node, config.nosetup, config.verbose)
 349         except Exception, e:
 350                 print "ERROR setting up session for %s" % hostname
 351                 print traceback.print_exc()
 352                 print e
 353                 return False
 354
 355         try:
 356                 conn = session.get_connection(config)
 357         except EOFError:
 358                 # NOTE: sometimes the wait in setup_host() is not long enough.
 359                 # So, here we try to wait a little longer before giving up entirely.
 360                 try:
 361                         time.sleep(session.timeout*4)
 362                         conn = session.get_connection(config)
 363                 except:
 364                         print traceback.print_exc()
 365                         return False
 366
 367         if forced_action == "reboot":
 368                 conn.restart_node('rins')
 369                 return True
 370
 371         boot_state = conn.get_boot_state()
 372         if boot_state == "boot":
 373                 print "...Boot state of %s already completed : skipping..." % node
 374                 return True
 375         elif boot_state == "unknown":
 376                 print "...Unknown bootstate for %s : skipping..."% node
 377                 return False
 378         else:
 379                 pass
 380
 381         if conn.bootmanager_running():
 382                 print "...BootManager is currently running.  Skipping host %s" % node
 383                 return True
 384
 385         #if config != None:
 386         #       if config.force:
 387         #               conn.restart_bootmanager(config.force)
 388         #               return True
 389
 390         # Read persistent flags, tagged on one week intervals.
 391         pflags = PersistFlags(hostname, 3*60*60*24, db='debug_persistflags')
 392
 393
 394         if config and not config.quiet: print "...downloading dmesg from %s" % node
 395         dmesg = conn.get_dmesg()
 396         child = fdpexpect.fdspawn(dmesg)
 397
 398         sequence = []
 399         while True:
 400                 steps = [
 401                         ('scsierror'  , 'SCSI error : <\d+ \d+ \d+ \d+> return code = 0x\d+'),
 402                         ('ioerror'    , 'end_request: I/O error, dev sd\w+, sector \d+'),
 403                         ('ccisserror' , 'cciss: cmd \w+ has CHECK CONDITION  byte \w+ = \w+'),
 404
 405                         ('buffererror', 'Buffer I/O error on device dm-\d, logical block \d+'),
 406
 407                         ('hdaseekerror', 'hda: dma_intr: status=0x\d+ { DriveReady SeekComplete Error }'),
 408                         ('hdacorrecterror', 'hda: dma_intr: error=0x\d+ { UncorrectableError }, LBAsect=\d+, sector=\d+'),
 409
 410                         ('atareadyerror'   , 'ata\d+: status=0x\d+ { DriveReady SeekComplete Error }'),
 411                         ('atacorrecterror' , 'ata\d+: error=0x\d+ { UncorrectableError }'),
 412
 413                         ('sdXerror'   , 'sd\w: Current: sense key: Medium Error'),
 414                         ('ext3error'   , 'EXT3-fs error (device dm-\d+): ext3_find_entry: reading directory #\d+ offset \d+'),
 415
 416                         ('floppytimeout','floppy0: floppy timeout called'),
 417                         ('floppyerror',  'end_request: I/O error, dev fd\w+, sector \d+'),
 418
 419                         # hda: dma_intr: status=0x51 { DriveReady SeekComplete Error }
 420                         # hda: dma_intr: error=0x40 { UncorrectableError }, LBAsect=23331263, sector=23331263
 421
 422                         # floppy0: floppy timeout called
 423                         # end_request: I/O error, dev fd0, sector 0
 424
 425                         # Buffer I/O error on device dm-2, logical block 8888896
 426                         # ata1: status=0x51 { DriveReady SeekComplete Error }
 427                         # ata1: error=0x40 { UncorrectableError }
 428                         # SCSI error : <0 0 0 0> return code = 0x8000002
 429                         # sda: Current: sense key: Medium Error
 430                         #       Additional sense: Unrecovered read error - auto reallocate failed
 431
 432                         # SCSI error : <0 2 0 0> return code = 0x40001
 433                         # end_request: I/O error, dev sda, sector 572489600
 434                 ]
 435                 id = index_to_id(steps, child.expect( steps_to_list(steps) + [ pexpect.EOF ]))
 436                 sequence.append(id)
 437
 438                 if id == "done":
 439                         break
 440
 441         s = Set(sequence)
 442         if config and not config.quiet: print "\tSET: ", s
 443
 444         if len(s) > 1:
 445                 print "...Potential drive errors on %s" % node
 446                 if len(s) == 2 and 'floppyerror' in s:
 447                         print "...Should investigate.  Continuing with node."
 448                 else:
 449                         print "...Should investigate.  Skipping node."
 450                         # TODO: send message related to these errors.
 451                         args = {}
 452                         args['hostname'] = hostname
 453                         args['log'] = conn.get_dmesg().read()
 454
 455                         m = PersistMessage(hostname, mailtxt.baddisk[0] % args,
 456                                                                                  mailtxt.baddisk[1] % args, True, db='hardware_persistmessages')
 457
 458                         loginbase = plc.siteId(hostname)
 459                         emails = plc.getTechEmails(loginbase)
 460                         m.send(emails)
 461                         conn.set_nodestate('disable')
 462                         return False
 463
 464         print "...Downloading bm.log from %s" % node
 465         log = conn.get_bootmanager_log()
 466         child = fdpexpect.fdspawn(log)
 467
 468         try:
 469                 if config.collect: return True
 470         except:
 471                 pass
 472
 473         time.sleep(1)
 474
 475         if config and not config.quiet: print "...Scanning bm.log for errors"
 476         action_id = "dbg"
 477         sequence = []
 478         while True:
 479
 480                 steps = [
 481                         ('bminit'               , 'Initializing the BootManager.'),
 482                         ('cfg'                  , 'Reading node configuration file.'),
 483                         ('auth'                 , 'Authenticating node with PLC.'),
 484                         ('getplc'               , 'Retrieving details of node from PLC.'),
 485                         ('update'               , 'Updating node boot state at PLC.'),
 486                         ('hardware'             , 'Checking if hardware requirements met.'),
 487                         ('installinit'  , 'Install: Initializing.'),
 488                         ('installdisk'  , 'Install: partitioning disks.'),
 489                         ('installbootfs', 'Install: bootstrapfs tarball.'),
 490                         ('installcfg'   , 'Install: Writing configuration files.'),
 491                         ('installstop'  , 'Install: Shutting down installer.'),
 492                         ('update2'              , 'Updating node boot state at PLC.'),
 493                         ('installinit2' , 'Install: Initializing.'),
 494                         ('validate'             , 'Validating node installation.'),
 495                         ('rebuildinitrd', 'Rebuilding initrd'),
 496                         ('netcfg'               , 'Install: Writing Network Configuration files.'),
 497                         ('update3'              , 'Updating node configuration.'),
 498                         ('disk'                 , 'Checking for unused disks to add to LVM.'),
 499                         ('update4'              , 'Sending hardware configuration to PLC.'),
 500                         ('debug'                , 'Starting debug mode'),
 501                         ('bmexceptmount', 'BootManagerException during mount'),
 502                         ('bmexceptvgscan', 'BootManagerException during vgscan/vgchange'),
 503                         ('bmexceptrmfail', 'Unable to remove directory tree: /tmp/mnt'),
 504                         ('exception'    , 'Exception'),
 505                         ('nocfg'        , 'Found configuration file planet.cnf on floppy, but was unable to parse it.'),
 506                         ('protoerror'   , 'XML RPC protocol error'),
 507                         ('nodehostname' , 'Configured node hostname does not resolve'),
 508                         ('implementerror', 'Implementation Error'),
 509                         ('readonlyfs'   , '[Errno 30] Read-only file system'),
 510                         ('noinstall'    , 'notinstalled'),
 511                         ('bziperror'    , 'bzip2: Data integrity error when decompressing.'),
 512                         ('noblockdev'   , "No block devices detected."),
 513                         ('dnserror'     , 'Name or service not known'),
 514                         ('downloadfail' , 'Unable to download main tarball /boot/bootstrapfs-planetlab-i386.tar.bz2 from server.'),
 515                         ('disktoosmall' , 'The total usable disk size of all disks is insufficient to be usable as a PlanetLab node.'),
 516                         ('hardwarerequirefail' , 'Hardware requirements not met'),
 517                         ('mkfsfail'         , 'while running: Running mkfs.ext2 -q  -m 0 -j /dev/planetlab/vservers failed'),
 518                         ('nofilereference', "No such file or directory: '/tmp/mnt/sysimg//vservers/.vref/planetlab-f8-i386/etc/hosts'"),
 519                         ('kernelcopyfail', "cp: cannot stat `/tmp/mnt/sysimg/boot/kernel-boot': No such file or directory"),
 520                         ('chrootfail'   , 'Running chroot /tmp/mnt/sysimg'),
 521                         ('modulefail'   , 'Unable to get list of system modules'),
 522                         ('writeerror'   , 'write error: No space left on device'),
 523                         ('nospace'      , "No space left on device"),
 524                         ('nonode'       , 'Failed to authenticate call: No such node'),
 525                         ('authfail'     , 'Failed to authenticate call: Call could not be authenticated'),
 526                         ('bootcheckfail'     , 'BootCheckAuthentication'),
 527                         ('bootupdatefail'   , 'BootUpdateNode'),
 528                 ]
 529                 list = steps_to_list(steps)
 530                 index = child.expect( list + [ pexpect.EOF ])
 531                 id = index_to_id(steps,index)
 532                 sequence.append(id)
 533
 534                 if id == "exception":
 535                         if config and not config.quiet: print "...Found An Exception!!!"
 536                 elif index == len(list):
 537                         #print "Reached EOF"
 538                         break
 539
 540         s = "-".join(sequence)
 541         print "   FOUND SEQUENCE: ", s
 542
 543         # NOTE: We get or set the flag based on the current sequence identifier.
 544         #  By using the sequence identifier, we guarantee that there will be no
 545         #  frequent loops.  I'm guessing there is a better way to track loops,
 546         #  though.
 547         #if not config.force and pflags.getRecentFlag(s):
 548         #       pflags.setRecentFlag(s)
 549         #       pflags.save()
 550         #       print "... flag is set or it has already run recently. Skipping %s" % node
 551         #       return True
 552
 553         sequences = {}
 554
 555
 556         # restart_bootmanager_boot
 557         for n in ["bminit-cfg-auth-getplc-update-installinit-validate-rebuildinitrd-netcfg-update3-disk-update4-done",
 558                         "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-netcfg-update3-disk-update4-update3-exception-protoerror-update-protoerror-debug-done",
 559                         "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-netcfg-disk-update4-update3-update3-implementerror-bootupdatefail-update-debug-done",
 560
 561                         "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-netcfg-disk-update4-update3-update3-exception-protoerror-update-protoerror-debug-done",
 562
 563                         "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-netcfg-update3-disk-update4-update3-exception-protoerror-update-debug-done",
 564                         "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-netcfg-disk-update4-update3-exception-chrootfail-update-debug-done",
 565                         "bminit-cfg-auth-getplc-update-debug-done",
 566                         "bminit-cfg-auth-getplc-exception-protoerror-update-protoerror-debug-done",
 567                         "bminit-cfg-auth-protoerror-exception-update-protoerror-debug-done",
 568                         "bminit-cfg-auth-protoerror-exception-update-bootupdatefail-authfail-debug-done",
 569                         "bminit-cfg-auth-protoerror-exception-update-debug-done",
 570                         "bminit-cfg-auth-getplc-exception-protoerror-update-debug-done",
 571                         "bminit-cfg-auth-getplc-implementerror-update-debug-done",
 572                         ]:
 573                 sequences.update({n : "restart_bootmanager_boot"})
 574
 575         #       conn.restart_bootmanager('rins')
 576         for n in [ "bminit-cfg-auth-getplc-installinit-validate-exception-modulefail-update-debug-done",
 577                         "bminit-cfg-auth-getplc-update-installinit-validate-exception-modulefail-update-debug-done",
 578                         "bminit-cfg-auth-getplc-installinit-validate-bmexceptmount-exception-noinstall-update-debug-done",
 579                         "bminit-cfg-auth-getplc-update-installinit-validate-bmexceptmount-exception-noinstall-update-debug-done",
 580                         "bminit-cfg-auth-getplc-installinit-validate-bmexceptvgscan-exception-noinstall-update-debug-done",
 581                         "bminit-cfg-auth-getplc-update-installinit-validate-exception-noinstall-update-debug-done",
 582                         "bminit-cfg-auth-getplc-hardware-installinit-installdisk-bziperror-exception-update-debug-done",
 583                         "bminit-cfg-auth-getplc-update-hardware-installinit-installdisk-installbootfs-exception-update-debug-done",
 584                         "bminit-cfg-auth-getplc-update-installinit-validate-bmexceptvgscan-exception-noinstall-update-debug-done",
 585                         "bminit-cfg-auth-getplc-hardware-installinit-installdisk-installbootfs-exception-update-debug-done",
 586                         "bminit-cfg-auth-getplc-update-installinit-validate-rebuildinitrd-netcfg-update3-implementerror-nofilereference-update-debug-done",
 587                         "bminit-cfg-auth-getplc-update-hardware-installinit-installdisk-exception-mkfsfail-update-debug-done",
 588                         "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-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                         # actual solution appears to involve removing the bad files, and
 593                         # continually trying to boot the node.
 594                         "bminit-cfg-auth-getplc-update-installinit-validate-rebuildinitrd-netcfg-disk-update4-update3-update3-implementerror-update-debug-done",
 595                         ]:
 596                 sequences.update({n : "restart_bootmanager_rins"})
 597
 598         # repair_node_keys
 599         sequences.update({"bminit-cfg-auth-bootcheckfail-authfail-exception-update-bootupdatefail-authfail-debug-done": "repair_node_keys"})
 600
 601         #   conn.restart_node('rins')
 602         for n in ["bminit-cfg-auth-getplc-update-installinit-validate-rebuildinitrd-exception-chrootfail-update-debug-done",
 603                         "bminit-cfg-auth-getplc-update-installinit-validate-rebuildinitrd-netcfg-update3-disk-update4-exception-chrootfail-update-debug-done",
 604                         "bminit-cfg-auth-getplc-hardware-installinit-installdisk-installbootfs-installcfg-exception-chrootfail-update-debug-done",
 605                         "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-writeerror-exception-chrootfail-update-debug-done",
 606                         "bminit-cfg-auth-getplc-update-hardware-installinit-exception-bmexceptrmfail-update-debug-done",
 607                         "bminit-cfg-auth-getplc-hardware-installinit-exception-bmexceptrmfail-update-debug-done",
 608                         "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-netcfg-update3-disk-update4-update3-implementerror-bootupdatefail-update-debug-done",
 609                         "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-netcfg-update3-implementerror-readonlyfs-update-debug-done",
 610                         "bminit-cfg-auth-getplc-update-installinit-validate-rebuildinitrd-netcfg-update3-nospace-exception-update-debug-done",
 611                         "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-netcfg-update3-implementerror-nospace-update-debug-done",
 612                         "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-netcfg-update3-implementerror-update-debug-done",
 613                         "bminit-cfg-auth-getplc-update-hardware-installinit-installdisk-installbootfs-exception-downloadfail-update-debug-done",
 614                         ]:
 615                 sequences.update({n : "restart_node_rins"})
 616
 617         #       restart_node_boot
 618         for n in ["bminit-cfg-auth-getplc-implementerror-bootupdatefail-update-debug-done",
 619                          "bminit-cfg-auth-implementerror-bootcheckfail-update-debug-done",
 620                          "bminit-cfg-auth-implementerror-bootcheckfail-update-implementerror-bootupdatefail-done",
 621                          "bminit-cfg-auth-getplc-update-installinit-validate-rebuildinitrd-netcfg-update3-implementerror-nospace-update-debug-done",
 622                          "bminit-cfg-auth-getplc-hardware-installinit-installdisk-installbootfs-exception-downloadfail-update-debug-done",
 623                          "bminit-cfg-auth-getplc-update-installinit-validate-implementerror-update-debug-done",
 624                          ]:
 625                 sequences.update({n: "restart_node_boot"})
 626
 627         # update_node_config_email
 628         for n in ["bminit-cfg-exception-nocfg-update-bootupdatefail-nonode-debug-done",
 629                           "bminit-cfg-exception-update-bootupdatefail-nonode-debug-done",
 630                           "bminit-cfg-auth-bootcheckfail-nonode-exception-update-bootupdatefail-nonode-debug-done",
 631                         ]:
 632                 sequences.update({n : "update_node_config_email"})
 633
 634         for n in [ "bminit-cfg-exception-nodehostname-update-debug-done",
 635                            "bminit-cfg-update-exception-nodehostname-update-debug-done",
 636                         ]:
 637                 sequences.update({n : "nodenetwork_email"})
 638
 639         # update_bootcd_email
 640         for n in ["bminit-cfg-auth-getplc-update-hardware-exception-noblockdev-hardwarerequirefail-update-debug-done",
 641                         "bminit-cfg-auth-getplc-hardware-exception-noblockdev-hardwarerequirefail-update-debug-done",
 642                         "bminit-cfg-auth-getplc-update-hardware-noblockdev-exception-hardwarerequirefail-update-debug-done",
 643                         "bminit-cfg-auth-getplc-hardware-noblockdev-exception-hardwarerequirefail-update-debug-done",
 644                         "bminit-cfg-auth-getplc-hardware-exception-hardwarerequirefail-update-debug-done",
 645                         ]:
 646                 sequences.update({n : "update_bootcd_email"})
 647
 648         for n in [ "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-netcfg-update3-implementerror-nofilereference-update-debug-done",
 649                         ]:
 650                 sequences.update({n: "suspect_error_email"})
 651
 652         # update_hardware_email
 653         sequences.update({"bminit-cfg-auth-getplc-hardware-exception-disktoosmall-hardwarerequirefail-update-debug-done" : "update_hardware_email"})
 654         sequences.update({"bminit-cfg-auth-getplc-hardware-disktoosmall-exception-hardwarerequirefail-update-debug-done" : "update_hardware_email"})
 655
 656         # broken_hardware_email
 657         sequences.update({"bminit-cfg-auth-getplc-update-hardware-exception-hardwarerequirefail-update-debug-done" : "broken_hardware_email"})
 658
 659         # bad_dns_email
 660         for n in [
 661          "bminit-cfg-update-implementerror-bootupdatefail-dnserror-update-implementerror-bootupdatefail-dnserror-done",
 662                 "bminit-cfg-auth-implementerror-bootcheckfail-dnserror-update-implementerror-bootupdatefail-dnserror-done",
 663                 ]:
 664                 sequences.update( { n : "bad_dns_email"})
 665
 666         flag_set = True
 667
 668
 669         if s not in sequences:
 670                 print "   HOST %s" % hostname
 671                 print "   UNKNOWN SEQUENCE: %s" % s
 672
 673                 args = {}
 674                 args['hostname'] = hostname
 675                 args['sequence'] = s
 676                 args['bmlog'] = conn.get_bootmanager_log().read()
 677                 m = PersistMessage(hostname, mailtxt.unknownsequence[0] % args,
 678                                                                          mailtxt.unknownsequence[1] % args, False, db='unknown_persistmessages')
 679                 m.reset()
 680                 m.send([config.cc_email])
 681
 682                 conn.restart_bootmanager('boot')
 683
 684                 # NOTE: Do not set the pflags value for this sequence if it's unknown.
 685                 # This way, we can check it again after we've fixed it.
 686                 flag_set = False
 687
 688         else:
 689
 690                 if   sequences[s] == "restart_bootmanager_boot":
 691                         if config and not config.quiet: print "...Restarting BootManager.py on %s "% node
 692                         conn.restart_bootmanager('boot')
 693                 elif sequences[s] == "restart_bootmanager_rins":
 694                         if config and not config.quiet: print "...Restarting BootManager.py on %s "% node
 695                         conn.restart_bootmanager('rins')
 696                 elif sequences[s] == "restart_node_rins":
 697                         conn.restart_node('rins')
 698                 elif sequences[s] == "restart_node_boot":
 699                         conn.restart_node('boot')
 700                 elif sequences[s] == "repair_node_keys":
 701                         if conn.compare_and_repair_nodekeys():
 702                                 # the keys either are in sync or were forced in sync.
 703                                 # so try to reboot the node again.
 704                                 conn.restart_bootmanager('rins')
 705                                 pass
 706                         else:
 707                                 # there was some failure to synchronize the keys.
 708                                 print "...Unable to repair node keys on %s" % node
 709
 710                 elif sequences[s] == "suspect_error_email":
 711                         args = {}
 712                         args['hostname'] = hostname
 713                         args['sequence'] = s
 714                         args['bmlog'] = conn.get_bootmanager_log().read()
 715                         m = PersistMessage(hostname, "Suspicous error from BootManager on %s" % args,
 716                                                                                  mailtxt.unknownsequence[1] % args, False, db='suspect_persistmessages')
 717                         m.reset()
 718                         m.send([config.cc_email])
 719
 720                         conn.restart_bootmanager('boot')
 721
 722                 elif sequences[s] == "update_node_config_email":
 723                         print "...Sending message to UPDATE NODE CONFIG"
 724                         args = {}
 725                         args['hostname'] = hostname
 726                         m = PersistMessage(hostname,  mailtxt.plnode_cfg[0] % args,  mailtxt.plnode_cfg[1] % args,
 727                                                                 True, db='nodeid_persistmessages')
 728                         loginbase = plc.siteId(hostname)
 729                         emails = plc.getTechEmails(loginbase)
 730                         m.send(emails)
 731                         conn.dump_plconf_file()
 732                         conn.set_nodestate('disable')
 733
 734                 elif sequences[s] == "nodenetwork_email":
 735                         print "...Sending message to LOOK AT NODE NETWORK"
 736                         args = {}
 737                         args['hostname'] = hostname
 738                         args['bmlog'] = conn.get_bootmanager_log().read()
 739                         m = PersistMessage(hostname,  mailtxt.plnode_network[0] % args,  mailtxt.plnode_cfg[1] % args,
 740                                                                 True, db='nodenet_persistmessages')
 741                         loginbase = plc.siteId(hostname)
 742                         emails = plc.getTechEmails(loginbase)
 743                         m.send(emails)
 744                         conn.dump_plconf_file()
 745                         conn.set_nodestate('disable')
 746
 747                 elif sequences[s] == "update_bootcd_email":
 748                         print "...NOTIFY OWNER TO UPDATE BOOTCD!!!"
 749                         import getconf
 750                         args = {}
 751                         args.update(getconf.getconf(hostname)) # NOTE: Generates boot images for the user:
 752                         args['hostname_list'] = "%s" % hostname
 753
 754                         m = PersistMessage(hostname, "Please Update Boot Image for %s" % hostname,
 755                                                                 mailtxt.newalphacd_one[1] % args, True, db='bootcd_persistmessages')
 756
 757                         loginbase = plc.siteId(hostname)
 758                         emails = plc.getTechEmails(loginbase)
 759                         m.send(emails)
 760
 761                         print "\tDisabling %s due to out-of-date BOOTCD" % hostname
 762                         conn.set_nodestate('disable')
 763
 764                 elif sequences[s] == "broken_hardware_email":
 765                         # MAKE An ACTION record that this host has failed hardware.  May
 766                         # require either an exception "/minhw" or other manual intervention.
 767                         # Definitely need to send out some more EMAIL.
 768                         print "...NOTIFYING OWNERS OF BROKEN HARDWARE on %s!!!" % hostname
 769                         # TODO: email notice of broken hardware
 770                         args = {}
 771                         args['hostname'] = hostname
 772                         args['log'] = conn.get_dmesg().read()
 773                         m = PersistMessage(hostname, mailtxt.baddisk[0] % args,
 774                                                                                  mailtxt.baddisk[1] % args, True, db='hardware_persistmessages')
 775
 776                         loginbase = plc.siteId(hostname)
 777                         emails = plc.getTechEmails(loginbase)
 778                         m.send(emails)
 779                         conn.set_nodestate('disable')
 780
 781                 elif sequences[s] == "update_hardware_email":
 782                         print "...NOTIFYING OWNERS OF MINIMAL HARDWARE FAILURE on %s!!!" % hostname
 783                         args = {}
 784                         args['hostname'] = hostname
 785                         args['bmlog'] = conn.get_bootmanager_log().read()
 786                         m = PersistMessage(hostname, mailtxt.minimalhardware[0] % args,
 787                                                                                  mailtxt.minimalhardware[1] % args, True, db='minhardware_persistmessages')
 788
 789                         loginbase = plc.siteId(hostname)
 790                         emails = plc.getTechEmails(loginbase)
 791                         m.send(emails)
 792                         conn.set_nodestate('disable')
 793
 794                 elif sequences[s] == "bad_dns_email":
 795                         print "...NOTIFYING OWNERS OF DNS FAILURE on %s!!!" % hostname
 796                         args = {}
 797                         try:
 798                                 node = api.GetNodes(hostname)[0]
 799                                 net = api.GetNodeNetworks(node['nodenetwork_ids'])[0]
 800                         except:
 801                                 print traceback.print_exc()
 802                                 # TODO: api error. skip email, b/c all info is not available,
 803                                 # flag_set will not be recorded.
 804                                 return False
 805                         nodenet_str = network_config_to_str(net)
 806
 807                         args['hostname'] = hostname
 808                         args['network_config'] = nodenet_str
 809                         args['nodenetwork_id'] = net['nodenetwork_id']
 810                         m = PersistMessage(hostname, mailtxt.baddns[0] % args,
 811                                                                                  mailtxt.baddns[1] % args, True, db='baddns_persistmessages')
 812
 813                         loginbase = plc.siteId(hostname)
 814                         emails = plc.getTechEmails(loginbase)
 815                         m.send(emails)
 816                         conn.set_nodestate('disable')
 817
 818         if flag_set:
 819                 pflags.setRecentFlag(s)
 820                 pflags.save()
 821
 822         return True
 823
 824
 825 # MAIN -------------------------------------------------------------------
 826
 827 def main():
 828         import parser as parsermodule
 829         parser = parsermodule.getParser()
 830
 831         parser.set_defaults(child=False, collect=False, nosetup=False, verbose=False,
 832                                                 force=None, quiet=False)
 833         parser.add_option("", "--child", dest="child", action="store_true",
 834                                                 help="This is the child mode of this process.")
 835         parser.add_option("", "--force", dest="force", metavar="boot_state",
 836                                                 help="Force a boot state passed to BootManager.py.")
 837         parser.add_option("", "--quiet", dest="quiet", action="store_true",
 838                                                 help="Extra quiet output messages.")
 839         parser.add_option("", "--verbose", dest="verbose", action="store_true",
 840                                                 help="Extra debug output messages.")
 841         parser.add_option("", "--nonet", dest="nonet", action="store_true",
 842                                                 help="Do not setup the network, use existing log files to re-run a test pass.")
 843         parser.add_option("", "--collect", dest="collect", action="store_true",
 844                                                 help="No action, just collect dmesg, and bm.log")
 845         parser.add_option("", "--nosetup", dest="nosetup", action="store_true",
 846                                                 help="Do not perform the orginary setup phase.")
 847
 848         parser = parsermodule.getParser(['nodesets', 'defaults'], parser)
 849         config = parsermodule.parse_args(parser)
 850
 851         if config.nodelist:
 852                 nodes = config.getListFromFile(config.nodelist)
 853         elif config.node:
 854                 nodes = [ config.node ]
 855         else:
 856                 parser.print_help()
 857                 sys.exit(1)
 858
 859         for node in nodes:
 860                 reboot(node, config)
 861
 862 if __name__ == "__main__":
 863         main()