monitor/bootman.py

   1 #!/usr/bin/python
   2
   3 # Attempt to reboot a node in debug state.
   4
   5
   6
   7 import os
   8 import sys
   9 import time
  10 import random
  11 import signal
  12 import traceback
  13 import subprocess
  14 from sets import Set
  15
  16 from monitor.getsshkeys import SSHKnownHosts
  17
  18 from monitor.Rpyc import SocketConnection, Async
  19 from monitor.Rpyc.Utils import *
  20
  21 from monitor import getconf
  22 from monitor import config
  23 from monitor import const
  24 from monitor.model import *
  25 from monitor.common import email_exception, found_within
  26 from monitor.database.info.model import *
  27 from monitor.database.info.interface import *
  28 from monitor.wrapper import plc
  29 from monitor.wrapper import plccache
  30 from monitor.wrapper.emailTxt import mailtxt
  31 from monitor.nodeconfig import network_config_to_str
  32
  33 from pcucontrol.util import command as moncommands
  34 from pcucontrol.util.command import Sopen
  35 from pcucontrol.transports.ssh import pxssh as pxssh
  36 from pcucontrol.transports.ssh import fdpexpect as fdpexpect
  37 from pcucontrol.transports.ssh import pexpect as pexpect
  38
  39
  40
  41 api = plc.getAuthAPI()
  42 fb = None
  43
  44
  45 class ExceptionDoubleSSHError(Exception): pass
  46
  47 class NodeConnection:
  48         def __init__(self, connection, node, config):
  49                 self.node = node
  50                 self.c = connection
  51                 self.config = config
  52
  53         def get_boot_state(self):
  54                 try:
  55                         if self.c.modules.os.path.exists('/tmp/source'):
  56                                 return "debug"
  57                         elif self.c.modules.os.path.exists('/vservers'):
  58                                 return "boot"
  59                         else:
  60                                 return "unknown"
  61                 except EOFError:
  62                         traceback.print_exc()
  63                         print self.c.modules.sys.path
  64                 except:
  65                         email_exception()
  66                         traceback.print_exc()
  67
  68                 return "unknown"
  69
  70         def get_dmesg(self):
  71                 self.c.modules.os.system("dmesg > /var/log/dmesg.bm.log")
  72                 download(self.c, "/var/log/dmesg.bm.log", "log/dmesg.%s.log" % self.node)
  73                 log = open("log/dmesg.%s.log" % self.node, 'r')
  74                 return log
  75
  76         def get_bootmanager_log(self):
  77                 download(self.c, "/tmp/bm.log", "log/bm.%s.log.gz" % self.node)
  78                 #os.system("zcat log/bm.%s.log.gz > log/bm.%s.log" % (self.node, self.node))
  79                 os.system("cp log/bm.%s.log.gz log/bm.%s.log" % (self.node, self.node))
  80                 log = open("log/bm.%s.log" % self.node, 'r')
  81                 return log
  82
  83         def dump_plconf_file(self):
  84                 c = self.c
  85                 self.c.modules.sys.path.append("/tmp/source/")
  86                 self.c.modules.os.chdir('/tmp/source')
  87
  88                 log = c.modules.BootManager.log('/tmp/new.log')
  89                 bm = c.modules.BootManager.BootManager(log,'boot')
  90
  91                 BootManagerException = c.modules.Exceptions.BootManagerException
  92                 InitializeBootManager = c.modules.BootManager.InitializeBootManager
  93                 ReadNodeConfiguration = c.modules.BootManager.ReadNodeConfiguration
  94                 bm_continue = True
  95
  96                 InitializeBootManager.Run(bm.VARS, bm.LOG)
  97                 try: ReadNodeConfiguration.Run(bm.VARS, bm.LOG)
  98                 except Exception, x:
  99                         bm_continue = False
 100                         print "   ERROR:", x
 101                         print "   Possibly, unable to find valid configuration file"
 102
 103                 if bm_continue:
 104                         for key in bm.VARS.keys():
 105                                 print key, " == ", bm.VARS[key]
 106                 else:
 107                         print "   Unable to read Node Configuration"
 108
 109
 110         def compare_and_repair_nodekeys(self):
 111                 c = self.c
 112                 self.c.modules.sys.path.append("/tmp/source/")
 113                 self.c.modules.os.chdir('/tmp/source')
 114
 115                 log = c.modules.BootManager.log('/tmp/new.log')
 116                 bm = c.modules.BootManager.BootManager(log,'boot')
 117
 118                 BootManagerException = c.modules.Exceptions.BootManagerException
 119                 InitializeBootManager = c.modules.BootManager.InitializeBootManager
 120                 ReadNodeConfiguration = c.modules.BootManager.ReadNodeConfiguration
 121                 bm_continue = True
 122
 123                 plcnode = plccache.GetNodeByName(self.node)
 124
 125                 InitializeBootManager.Run(bm.VARS, bm.LOG)
 126                 try: ReadNodeConfiguration.Run(bm.VARS, bm.LOG)
 127                 except Exception, x:
 128                         bm_continue = False
 129                         print "exception"
 130                         print x
 131                         print "   Possibly, unable to find valid configuration file"
 132
 133                 if bm_continue:
 134                         print "   NODE: %s" % bm.VARS['NODE_KEY']
 135                         print "   PLC : %s" % plcnode['key']
 136
 137                         if bm.VARS['NODE_KEY'] == plcnode['key']:
 138                                 return True
 139                         else:
 140                                 if api.UpdateNode(self.node, {'key': bm.VARS['NODE_KEY']}):
 141                                         print "   Successfully updated NODE_KEY with PLC"
 142                                         return True
 143                                 else:
 144                                         return False
 145
 146                         #for key in bm.VARS.keys():
 147                         #       print key, " == ", bm.VARS[key]
 148                 else:
 149                         print "   Unable to retrieve NODE_KEY"
 150
 151         def bootmanager_running(self):
 152                 if self.c.modules.os.path.exists('/tmp/BM_RUNNING'):
 153                         return True
 154                 else:
 155                         return False
 156
 157         def set_nodestate(self, state='boot'):
 158                 return api.UpdateNode(self.node, {'boot_state' : state})
 159
 160         def restart_node(self, state='boot'):
 161                 api.UpdateNode(self.node, {'boot_state' : state})
 162
 163                 pflags = PersistFlags(self.node, 1*60*60*24, db='restart_persistflags')
 164                 if not pflags.getRecentFlag('gentlekill'):
 165                         print "   Killing all slice processes... : %s" %  self.node
 166                         cmd_slicekill = "ls -d /proc/virtual/[0-9]* | awk -F '/' '{print $4}' | xargs -I{} /usr/sbin/vkill -s 9 --xid {} -- 0"
 167                         self.c.modules.os.system(cmd_slicekill)
 168                         cmd = """ shutdown -r +1 & """
 169                         print "   Restarting %s : %s" % ( self.node, cmd)
 170                         self.c.modules.os.system(cmd)
 171
 172                         pflags.setRecentFlag('gentlekill')
 173                         pflags.save()
 174                 else:
 175                         print "   Restarting with sysrq 'sub' %s" % self.node
 176                         cmd = """ (sleep 5; echo 's' > /proc/sysrq-trigger; echo 'u' > /proc/sysrq-trigger; echo 'b' > /proc/sysrq-trigger ) & """
 177                         self.c.modules.os.system(cmd)
 178
 179                 return
 180
 181         def restart_bootmanager(self, forceState):
 182
 183                 self.c.modules.os.chdir('/tmp/source')
 184                 if self.c.modules.os.path.exists('/tmp/BM_RUNNING'):
 185                         print "   BootManager is already running: try again soon..."
 186                 else:
 187                         print "   Starting 'BootManager.py %s' on %s " % (forceState, self.node)
 188                         cmd = "( touch /tmp/BM_RUNNING ;  " + \
 189                               "  python ./BootManager.py %s &> server.log < /dev/null ; " + \
 190                                   "  rm -f /tmp/BM_RUNNING " + \
 191                                   ") &"
 192                         cmd = cmd % forceState
 193                         self.c.modules.os.system(cmd)
 194
 195                 return
 196
 197
 198 class PlanetLabSession:
 199         globalport = 22000 + int(random.random()*1000)
 200
 201         def __init__(self, node, nosetup, verbose):
 202                 self.verbose = verbose
 203                 self.node = node
 204                 self.port = None
 205                 self.nosetup = nosetup
 206                 self.command = None
 207                 self.setup_host()
 208
 209         def get_connection(self, config):
 210                 conn = NodeConnection(SocketConnection("localhost", self.port), self.node, config)
 211                 #i = 0
 212                 #while i < 3:
 213                 #       print i, conn.c.modules.sys.path
 214                 #       print conn.c.modules.os.path.exists('/tmp/source')
 215                 #       i+=1
 216                 #       time.sleep(1)
 217                 return conn
 218
 219         def setup_host(self):
 220                 self.port = PlanetLabSession.globalport
 221                 PlanetLabSession.globalport = PlanetLabSession.globalport + 1
 222
 223                 args = {}
 224                 args['port'] = self.port
 225                 args['user'] = 'root'
 226                 args['hostname'] = self.node
 227                 args['monitordir'] = config.MONITOR_SCRIPT_ROOT
 228                 ssh_port = 22
 229
 230                 if self.nosetup:
 231                         print "Skipping setup"
 232                         return
 233
 234                 # COPY Rpyc files to host
 235                 cmd = "rsync -qv -az -e ssh %(monitordir)s/Rpyc/ %(user)s@%(hostname)s:Rpyc 2> /dev/null" % args
 236                 if self.verbose: print cmd
 237                 print cmd
 238                 # TODO: Add timeout
 239                 timeout = 120
 240                 localos = moncommands.CMD()
 241
 242                 ret = localos.system(cmd, timeout)
 243                 print ret
 244                 if ret != 0:
 245                         print "\tUNKNOWN SSH KEY FOR %s; making an exception" % self.node
 246                         #print "MAKE EXPLICIT EXCEPTION FOR %s" % self.node
 247                         k = SSHKnownHosts(); k.updateDirect(self.node); k.write(); del k
 248                         ret = localos.system(cmd, timeout)
 249                         print ret
 250                         if ret != 0:
 251                                 print "\tFAILED TWICE"
 252                                 #sys.exit(1)
 253                                 raise ExceptionDoubleSSHError("Failed twice trying to login with updated ssh host key")
 254
 255                 t1 = time.time()
 256                 # KILL any already running servers.
 257                 ssh = moncommands.SSH(args['user'], args['hostname'], ssh_port)
 258                 (ov,ev) = ssh.run_noexcept2("""<<\EOF
 259             rm -f out.log
 260             echo "kill server" >> out.log
 261             ps ax | grep Rpyc | grep -v grep | awk '{print $1}' | xargs kill 2> /dev/null ;
 262             echo "export" >> out.log
 263             export PYTHONPATH=$HOME  ;
 264             echo "start server" >> out.log
 265             python Rpyc/Servers/forking_server.py &> server.log &
 266             echo "done" >> out.log
 267 EOF""")
 268                 #cmd = """ssh %(user)s@%(hostname)s """ + \
 269                 #        """'ps ax | grep Rpyc | grep -v grep | awk "{print \$1}" | xargs kill 2> /dev/null' """
 270                 #cmd = cmd % args
 271                 #if self.verbose: print cmd
 272                 ## TODO: Add timeout
 273                 #print localos.system(cmd,timeout)
 274
 275                 ## START a new rpyc server.
 276                 #cmd = """ssh -n %(user)s@%(hostname)s "export PYTHONPATH=\$HOME; """ + \
 277                 #        """python Rpyc/Servers/forking_server.py &> server.log < /dev/null &" """
 278                 #cmd = cmd % args
 279                 #if self.verbose: print cmd
 280                 #print localos.system(cmd,timeout)
 281                 print "setup rpyc server over ssh"
 282                 print ssh.ret
 283
 284                 # TODO: Add timeout
 285                 # This was tricky to make synchronous.  The combination of ssh-clients-4.7p1,
 286                 # and the following options seems to work well.
 287                 cmd = """ssh -o ExitOnForwardFailure=yes -o BatchMode=yes """ + \
 288                           """-o PermitLocalCommand=yes -o LocalCommand='echo "READY"' """ + \
 289                           """-o ConnectTimeout=120 """ + \
 290                           """-n -N -L %(port)s:localhost:18812 """ + \
 291                           """%(user)s@%(hostname)s"""
 292                 cmd = cmd % args
 293                 if self.verbose: print cmd
 294                 print cmd
 295                 self.command = Sopen(cmd, shell=True, stdout=subprocess.PIPE)
 296                 # TODO: the read() here may block indefinitely.  Need a better
 297                 # approach therefore, that includes a timeout.
 298                 #ret = self.command.stdout.read(5)
 299                 ret = moncommands.read_t(self.command.stdout, 5)
 300
 301                 t2 = time.time()
 302                 if 'READY' in ret:
 303                         # NOTE: There is still a slight race for machines that are slow...
 304                         self.timeout = 2*(t2-t1)
 305                         print "Sleeping for %s sec" % self.timeout
 306                         time.sleep(self.timeout)
 307                         return
 308
 309                 if self.command.returncode is not None:
 310                         print "Failed to establish tunnel!"
 311                         raise Exception("SSH Tunnel exception : %s %s" % (self.node, self.command.returncode))
 312
 313                 raise Exception("Unknown SSH Tunnel Exception: still running, but did not report 'READY'")
 314
 315         def __del__(self):
 316                 if self.command:
 317                         if self.verbose: print "Killing SSH session %s" % self.port
 318                         print "Killing SSH session %s" % self.port
 319                         self.command.kill()
 320
 321
 322 def steps_to_list(steps, index=1):
 323         return map(lambda x: x[index], steps)
 324
 325 def index_to_id(steps,index):
 326         if index < len(steps):
 327                 return steps[index][0]
 328         else:
 329                 return "done"
 330
 331 class DebugInterface:
 332         def __init__(self, hostname):
 333                 self.hostname = hostname
 334                 self.session = None
 335
 336         def getConnection(self):
 337                 print "Creating session for %s" % self.hostname
 338                 # update known_hosts file (in case the node has rebooted since last run)
 339                 try:
 340                         k = SSHKnownHosts(); k.update(self.hostname); k.write(); del k
 341                 except:
 342                         email_exception()
 343                         print traceback.print_exc()
 344                         return False
 345
 346                 try:
 347                         if config == None:
 348                                 self.session = PlanetLabSession(self.hostname, False, True)
 349                         else:
 350                                 self.session = PlanetLabSession(self.hostname, config.nosetup, config.verbose)
 351                 except ExceptionDoubleSSHError, e:
 352                         msg = "ERROR setting up session for %s" % self.hostname
 353                         print msg
 354                         return False
 355                 except Exception, e:
 356                         traceback.print_exc()
 357                         email_exception(msg)
 358                         return False
 359
 360                 try:
 361                         conn = self.session.get_connection(config)
 362                 except EOFError:
 363                         # NOTE: sometimes the wait in setup_host() is not long enough.
 364                         # So, here we try to wait a little longer before giving up entirely.
 365                         try:
 366                                 time.sleep(self.session.timeout*5)
 367                                 conn = self.session.get_connection(config)
 368                         except EOFError:
 369                                 # failed twice... no need to report this really, it's just in a
 370                                 # weird state...
 371                                 return False
 372                         except:
 373                                 traceback.print_exc()
 374                                 email_exception(self.hostname)
 375                                 return False
 376                 #print "trying to use conn before returning it."
 377                 #print conn.c.modules.sys.path
 378                 #print conn.c.modules.os.path.exists('/tmp/source')
 379                 #time.sleep(1)
 380
 381                 #print "conn: %s" % conn
 382                 return conn
 383
 384         def getSequences(self):
 385
 386                 # TODO: This can be replaced with a DB definition at a future time.
 387                 #               This would make it possible for an admin to introduce new
 388                 #               patterns without touching code.
 389
 390                 sequences = {}
 391                 # restart_bootmanager_boot
 392                 for n in ["bminit-cfg-auth-getplc-update-installinit-validate-rebuildinitrd-netcfg-update3-disk-update4-done",
 393                                 "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-netcfg-update3-disk-update4-update3-exception-protoerror-update-protoerror-debug-done",
 394                                 "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-netcfg-disk-update4-update3-update3-implementerror-bootupdatefail-update-debug-done",
 395
 396                                 "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-netcfg-disk-update4-update3-update3-exception-protoerror-update-protoerror-debug-done",
 397
 398                                 "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-netcfg-update3-disk-update4-update3-exception-protoerror-update-debug-done",
 399                                 "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-netcfg-disk-update4-update3-exception-chrootfail-update-debug-done",
 400                                 "bminit-cfg-auth-getplc-update-debug-done",
 401                                 "bminit-cfg-auth-getplc-exception-protoerror-update-protoerror-debug-done",
 402                                 "bminit-cfg-auth-protoerror-exception-update-protoerror-debug-done",
 403                                 "bminit-cfg-auth-protoerror-exception-update-bootupdatefail-authfail-debug-done",
 404                                 "bminit-cfg-auth-protoerror-exception-update-debug-done",
 405                                 "bminit-cfg-auth-getplc-exception-protoerror-update-debug-done",
 406                                 "bminit-cfg-auth-getplc-implementerror-update-debug-done",
 407                                 ]:
 408                         sequences.update({n : "restart_bootmanager_boot"})
 409
 410                 #       conn.restart_bootmanager('reinstall')
 411                 for n in [ "bminit-cfg-auth-getplc-installinit-validate-exception-modulefail-update-debug-done",
 412                                 "bminit-cfg-auth-getplc-update-installinit-validate-exception-modulefail-update-debug-done",
 413                                 "bminit-cfg-auth-getplc-installinit-validate-bmexceptmount-exception-noinstall-update-debug-done",
 414                                 "bminit-cfg-auth-getplc-update-installinit-validate-bmexceptmount-exception-noinstall-update-debug-done",
 415                                 "bminit-cfg-auth-getplc-installinit-validate-bmexceptvgscan-exception-noinstall-update-debug-done",
 416                                 "bminit-cfg-auth-getplc-update-installinit-validate-exception-noinstall-update-debug-done",
 417                                 "bminit-cfg-auth-getplc-hardware-installinit-installdisk-bziperror-exception-update-debug-done",
 418                                 "bminit-cfg-auth-getplc-update-hardware-installinit-installdisk-installbootfs-exception-update-debug-done",
 419                                 "bminit-cfg-auth-getplc-update-installinit-validate-bmexceptvgscan-exception-noinstall-update-debug-done",
 420                                 "bminit-cfg-auth-getplc-hardware-installinit-installdisk-installbootfs-exception-update-debug-done",
 421                                 "bminit-cfg-auth-getplc-update-installinit-validate-rebuildinitrd-netcfg-update3-implementerror-nofilereference-update-debug-done",
 422                                 "bminit-cfg-auth-getplc-update-hardware-installinit-installdisk-exception-mkfsfail-update-debug-done",
 423                                 "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-exception-chrootfail-update-debug-done",
 424                                 "bminit-cfg-auth-getplc-update-installinit-validate-rebuildinitrd-netcfg-disk-update4-exception-chrootfail-update-debug-done",
 425                                 "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",
 426                                 "bminit-cfg-auth-getplc-hardware-installinit-installdisk-installbootfs-installcfg-installstop-update-installinit-validate-rebuildinitrd-netcfg-disk-update4-update3-update3-kernelcopyfail-exception-update-debug-done",
 427                                 "bminit-cfg-auth-getplc-installinit-validate-exception-noinstall-update-debug-done",
 428                                 # actual solution appears to involve removing the bad files, and
 429                                 # continually trying to boot the node.
 430                                 "bminit-cfg-auth-getplc-update-installinit-validate-rebuildinitrd-netcfg-disk-update4-update3-update3-implementerror-update-debug-done",
 431                                 "bminit-cfg-auth-getplc-installinit-validate-exception-bmexceptmount-exception-noinstall-update-debug-done",
 432                                 "bminit-cfg-auth-getplc-update-installinit-validate-exception-bmexceptmount-exception-noinstall-update-debug-done",
 433                                 ]:
 434                         sequences.update({n : "restart_bootmanager_rins"})
 435
 436                 # repair_node_keys
 437                 sequences.update({"bminit-cfg-auth-bootcheckfail-authfail-exception-update-bootupdatefail-authfail-debug-done": "repair_node_keys"})
 438
 439                 #   conn.restart_node('reinstall')
 440                 for n in ["bminit-cfg-auth-getplc-update-installinit-validate-rebuildinitrd-exception-chrootfail-update-debug-done",
 441                                 "bminit-cfg-auth-getplc-update-installinit-validate-rebuildinitrd-netcfg-update3-disk-update4-exception-chrootfail-update-debug-done",
 442                                 "bminit-cfg-auth-getplc-hardware-installinit-installdisk-installbootfs-installcfg-exception-chrootfail-update-debug-done",
 443                                 "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-writeerror-exception-chrootfail-update-debug-done",
 444                                 "bminit-cfg-auth-getplc-update-hardware-installinit-exception-bmexceptrmfail-update-debug-done",
 445                                 "bminit-cfg-auth-getplc-hardware-installinit-exception-bmexceptrmfail-update-debug-done",
 446                                 "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-netcfg-update3-disk-update4-update3-implementerror-bootupdatefail-update-debug-done",
 447                                 "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-netcfg-update3-implementerror-readonlyfs-update-debug-done",
 448                                 "bminit-cfg-auth-getplc-update-installinit-validate-rebuildinitrd-netcfg-update3-nospace-exception-update-debug-done",
 449                                 "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-netcfg-update3-implementerror-nospace-update-debug-done",
 450                                 "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-netcfg-update3-implementerror-update-debug-done",
 451                                 "bminit-cfg-auth-getplc-update-hardware-installinit-installdisk-installbootfs-exception-downloadfail-update-debug-done",
 452                                 ]:
 453                         sequences.update({n : "restart_node_rins"})
 454
 455                 #       restart_node_boot
 456                 for n in ["bminit-cfg-auth-getplc-implementerror-bootupdatefail-update-debug-done",
 457                                  "bminit-cfg-auth-implementerror-bootcheckfail-update-debug-done",
 458                                  "bminit-cfg-auth-implementerror-bootcheckfail-update-implementerror-bootupdatefail-done",
 459                                  "bminit-cfg-auth-getplc-update-installinit-validate-rebuildinitrd-netcfg-update3-implementerror-nospace-update-debug-done",
 460                                  "bminit-cfg-auth-getplc-hardware-installinit-installdisk-installbootfs-exception-downloadfail-update-debug-done",
 461                                  "bminit-cfg-auth-getplc-update-installinit-validate-implementerror-update-debug-done",
 462                                  ]:
 463                         sequences.update({n: "restart_node_boot"})
 464
 465                 # update_node_config_email
 466                 for n in ["bminit-cfg-exception-nocfg-update-bootupdatefail-nonode-debug-done",
 467                                   "bminit-cfg-exception-update-bootupdatefail-nonode-debug-done",
 468                                   "bminit-cfg-auth-bootcheckfail-nonode-exception-update-bootupdatefail-nonode-debug-done",
 469                                 ]:
 470                         sequences.update({n : "update_node_config_email"})
 471
 472                 for n in [ "bminit-cfg-exception-nodehostname-update-debug-done",
 473                                    "bminit-cfg-update-exception-nodehostname-update-debug-done",
 474                                 ]:
 475                         sequences.update({n : "nodenetwork_email"})
 476
 477                 # update_bootcd_email
 478                 for n in ["bminit-cfg-auth-getplc-update-hardware-exception-noblockdev-hardwarerequirefail-update-debug-done",
 479                                 "bminit-cfg-auth-getplc-hardware-exception-noblockdev-hardwarerequirefail-update-debug-done",
 480                                 "bminit-cfg-auth-getplc-update-hardware-noblockdev-exception-hardwarerequirefail-update-debug-done",
 481                                 "bminit-cfg-auth-getplc-hardware-noblockdev-exception-hardwarerequirefail-update-debug-done",
 482                                 "bminit-cfg-auth-getplc-hardware-exception-hardwarerequirefail-update-debug-done",
 483                                 ]:
 484                         sequences.update({n : "update_bootcd_email"})
 485
 486                 for n in [ "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-netcfg-update3-implementerror-nofilereference-update-debug-done",
 487                                 ]:
 488                         sequences.update({n: "suspect_error_email"})
 489
 490                 # update_hardware_email
 491                 sequences.update({"bminit-cfg-auth-getplc-hardware-exception-disktoosmall-hardwarerequirefail-update-debug-done" : "update_hardware_email"})
 492                 sequences.update({"bminit-cfg-auth-getplc-hardware-disktoosmall-exception-hardwarerequirefail-update-debug-done" : "update_hardware_email"})
 493
 494                 # broken_hardware_email
 495                 sequences.update({"bminit-cfg-auth-getplc-update-hardware-exception-hardwarerequirefail-update-debug-done" : "broken_hardware_email"})
 496
 497                 # bad_dns_email
 498                 for n in [
 499                  "bminit-cfg-update-implementerror-bootupdatefail-dnserror-update-implementerror-bootupdatefail-dnserror-done",
 500                         "bminit-cfg-auth-implementerror-bootcheckfail-dnserror-update-implementerror-bootupdatefail-dnserror-done",
 501                         ]:
 502                         sequences.update( { n : "bad_dns_email"})
 503
 504                 return sequences
 505
 506         def getDiskSteps(self):
 507                 steps = [
 508                         ('scsierror'  , 'SCSI error : <\d+ \d+ \d+ \d+> return code = 0x\d+'),
 509                         ('ioerror'    , 'end_request: I/O error, dev sd\w+, sector \d+'),
 510                         ('ccisserror' , 'cciss: cmd \w+ has CHECK CONDITION  byte \w+ = \w+'),
 511
 512                         ('buffererror', 'Buffer I/O error on device dm-\d, logical block \d+'),
 513
 514                         ('hdaseekerror', 'hda: dma_intr: status=0x\d+ { DriveReady SeekComplete Error }'),
 515                         ('hdacorrecterror', 'hda: dma_intr: error=0x\d+ { UncorrectableError }, LBAsect=\d+, sector=\d+'),
 516
 517                         ('atareadyerror'   , 'ata\d+: status=0x\d+ { DriveReady SeekComplete Error }'),
 518                         ('atacorrecterror' , 'ata\d+: error=0x\d+ { UncorrectableError }'),
 519
 520                         ('sdXerror'   , 'sd\w: Current: sense key: Medium Error'),
 521                         ('ext3error'   , 'EXT3-fs error (device dm-\d+): ext3_find_entry: reading directory #\d+ offset \d+'),
 522
 523                         ('floppytimeout','floppy0: floppy timeout called'),
 524                         ('floppyerror',  'end_request: I/O error, dev fd\w+, sector \d+'),
 525
 526                         # hda: dma_intr: status=0x51 { DriveReady SeekComplete Error }
 527                         # hda: dma_intr: error=0x40 { UncorrectableError }, LBAsect=23331263, sector=23331263
 528
 529                         # floppy0: floppy timeout called
 530                         # end_request: I/O error, dev fd0, sector 0
 531
 532                         # Buffer I/O error on device dm-2, logical block 8888896
 533                         # ata1: status=0x51 { DriveReady SeekComplete Error }
 534                         # ata1: error=0x40 { UncorrectableError }
 535                         # SCSI error : <0 0 0 0> return code = 0x8000002
 536                         # sda: Current: sense key: Medium Error
 537                         #       Additional sense: Unrecovered read error - auto reallocate failed
 538
 539                         # SCSI error : <0 2 0 0> return code = 0x40001
 540                         # end_request: I/O error, dev sda, sector 572489600
 541                 ]
 542                 return steps
 543
 544         def getDiskSequence(self, steps, child):
 545                 sequence = []
 546                 while True:
 547                         id = index_to_id(steps, child.expect( steps_to_list(steps) + [ pexpect.EOF ]))
 548                         sequence.append(id)
 549
 550                         if id == "done":
 551                                 break
 552                 return sequence
 553
 554         def getBootManagerStepPatterns(self):
 555                 steps = [
 556                         ('bminit'               , 'Initializing the BootManager.'),
 557                         ('cfg'                  , 'Reading node configuration file.'),
 558                         ('auth'                 , 'Authenticating node with PLC.'),
 559                         ('getplc'               , 'Retrieving details of node from PLC.'),
 560                         ('update'               , 'Updating node boot state at PLC.'),
 561                         ('hardware'             , 'Checking if hardware requirements met.'),
 562                         ('installinit'  , 'Install: Initializing.'),
 563                         ('installdisk'  , 'Install: partitioning disks.'),
 564                         ('installbootfs', 'Install: bootstrapfs tarball.'),
 565                         ('installcfg'   , 'Install: Writing configuration files.'),
 566                         ('installstop'  , 'Install: Shutting down installer.'),
 567                         ('update2'              , 'Updating node boot state at PLC.'),
 568                         ('installinit2' , 'Install: Initializing.'),
 569                         ('validate'             , 'Validating node installation.'),
 570                         ('rebuildinitrd', 'Rebuilding initrd'),
 571                         ('netcfg'               , 'Install: Writing Network Configuration files.'),
 572                         ('update3'              , 'Updating node configuration.'),
 573                         ('disk'                 , 'Checking for unused disks to add to LVM.'),
 574                         ('update4'              , 'Sending hardware configuration to PLC.'),
 575                         ('debug'                , 'Starting debug mode'),
 576                         ('bmexceptmount', 'BootManagerException during mount'),
 577                         ('bmexceptvgscan', 'BootManagerException during vgscan/vgchange'),
 578                         ('bmexceptrmfail', 'Unable to remove directory tree: /tmp/mnt'),
 579                         ('exception'    , 'Exception'),
 580                         ('nocfg'        , 'Found configuration file planet.cnf on floppy, but was unable to parse it.'),
 581                         ('protoerror'   , 'XML RPC protocol error'),
 582                         ('nodehostname' , 'Configured node hostname does not resolve'),
 583                         ('implementerror', 'Implementation Error'),
 584                         ('readonlyfs'   , '[Errno 30] Read-only file system'),
 585                         ('baddisk'      , "IOError: [Errno 13] Permission denied: '/tmp/mnt/sysimg//vservers/\w+/etc/hosts'"),
 586                         ('noinstall'    , 'notinstalled'),
 587                         ('bziperror'    , 'bzip2: Data integrity error when decompressing.'),
 588                         ('noblockdev'   , "No block devices detected."),
 589                         ('dnserror'     , 'Name or service not known'),
 590                         ('downloadfail' , 'Unable to download main tarball /boot/bootstrapfs-planetlab-i386.tar.bz2 from server.'),
 591                         ('disktoosmall' , 'The total usable disk size of all disks is insufficient to be usable as a PlanetLab node.'),
 592                         ('hardwarerequirefail' , 'Hardware requirements not met'),
 593                         ('mkfsfail'         , 'while running: Running mkfs.ext2 -q  -m 0 -j /dev/planetlab/vservers failed'),
 594                         ('nofilereference', "No such file or directory: '/tmp/mnt/sysimg//vservers/.vref/planetlab-f8-i386/etc/hosts'"),
 595                         ('kernelcopyfail', "cp: cannot stat `/tmp/mnt/sysimg/boot/kernel-boot': No such file or directory"),
 596                         ('chrootfail'   , 'Running chroot /tmp/mnt/sysimg'),
 597                         ('modulefail'   , 'Unable to get list of system modules'),
 598                         ('writeerror'   , 'write error: No space left on device'),
 599                         ('nospace'      , "No space left on device"),
 600                         ('nonode'       , 'Failed to authenticate call: No such node'),
 601                         ('authfail'     , 'Failed to authenticate call: Call could not be authenticated'),
 602                         ('bootcheckfail'     , 'BootCheckAuthentication'),
 603                         ('bootupdatefail'   , 'BootUpdateNode'),
 604                 ]
 605                 return steps
 606
 607         def getBootManagerSequenceFromLog(self, steps, child):
 608                 sequence = []
 609                 while True:
 610
 611                         index = child.expect( steps_to_list(steps) + [ pexpect.EOF ])
 612                         id = index_to_id(steps,index)
 613                         sequence.append(id)
 614
 615                         if id == "exception":
 616                                 print "...Found An Exception!!!"
 617                         elif id == "done": #index == len(steps_to_list(steps)):
 618                                 #print "Reached EOF"
 619                                 break
 620
 621                 return sequence
 622
 623
 624 def restore(sitehist, hostname, config=None, forced_action=None):
 625
 626         # NOTE: Nothing works if the bootcd is REALLY old.
 627         #       So, this is the first step.
 628
 629         fbnode = FindbadNodeRecord.get_latest_by(hostname=hostname).to_dict()
 630         recent_actions = sitehist.getRecentActions(hostname=hostname)
 631
 632         if fbnode['observed_category'] == "OLDBOOTCD":
 633                 print "\t...Notify owner to update BootImage!!!"
 634
 635                 if not found_within(recent_actions, 'newbootcd_notice', 3):
 636                         sitehist.sendMessage('newbootcd_notice', hostname=hostname)
 637
 638                         print "\tDisabling %s due to out-of-date BootImage" % hostname
 639                         api.UpdateNode(hostname, {'boot_state' : 'disable'})
 640
 641                 # NOTE: nothing else is possible.
 642                 return True
 643
 644         debugnode = DebugInterface(hostname)
 645         conn = debugnode.getConnection()
 646         #print "conn: %s" % conn
 647         #print "trying to use conn after returning it."
 648         #print conn.c.modules.sys.path
 649         #print conn.c.modules.os.path.exists('/tmp/source')
 650         if type(conn) == type(False): return False
 651
 652         #if forced_action == "reboot":
 653         #       conn.restart_node('reinstall')
 654         #       return True
 655
 656         boot_state = conn.get_boot_state()
 657         if boot_state != "debug":
 658                 print "... %s in %s state: skipping..." % (hostname , boot_state)
 659                 return boot_state == "boot"
 660
 661         if conn.bootmanager_running():
 662                 print "...BootManager is currently running.  Skipping host %s" %hostname
 663                 return True
 664
 665         # Read persistent flags, tagged on one week intervals.
 666
 667         if config and not config.quiet: print "...downloading dmesg from %s" %hostname
 668         dmesg = conn.get_dmesg()
 669         child = fdpexpect.fdspawn(dmesg)
 670
 671         steps = debugnode.getDiskSteps()
 672         sequence = debugnode.getDiskSequence(steps, child)
 673
 674         s = Set(sequence)
 675         if config and not config.quiet: print "\tSET: ", s
 676
 677         if len(s) > 1:
 678                 print "...Potential drive errors on %s" % hostname
 679                 if len(s) == 2 and 'floppyerror' in s:
 680                         print "...Should investigate.  Continuing with node."
 681                 else:
 682                         print "...Should investigate.  Skipping node."
 683                         # TODO: send message related to these errors.
 684
 685                         if not found_within(recent_actions, 'newbootcd_notice', 3):
 686
 687                                 log=conn.get_dmesg().read()
 688                                 sitehist.sendMessage('baddisk_notice', hostname=hostname, log=log)
 689                                 conn.set_nodestate('disable')
 690
 691                         return False
 692
 693         print "...Downloading bm.log from %s" %hostname
 694         log = conn.get_bootmanager_log()
 695         child = fdpexpect.fdspawn(log)
 696
 697         if hasattr(config, 'collect') and config.collect: return True
 698
 699         if config and not config.quiet: print "...Scanning bm.log for errors"
 700
 701         time.sleep(1)
 702
 703         steps = debugnode.getBootManagerStepPatterns()
 704         sequence = debugnode.getBootManagerSequenceFromLog(steps, child)
 705
 706         s = "-".join(sequence)
 707         print "   FOUND SEQUENCE: ", s
 708
 709         # NOTE: We get or set the flag based on the current sequence identifier.
 710         #  By using the sequence identifier, we guarantee that there will be no
 711         #  frequent loops.  I'm guessing there is a better way to track loops,
 712         #  though.
 713
 714         sequences = debugnode.getSequences()
 715         flag_set = True
 716
 717         if s not in sequences:
 718                 print "   HOST %s" % hostname
 719                 print "   UNKNOWN SEQUENCE: %s" % s
 720
 721                 args = {}
 722                 args['hostname'] = hostname
 723                 args['sequence'] = s
 724                 args['bmlog'] = conn.get_bootmanager_log().read()
 725                 args['viart'] = False
 726
 727                 sitehist.sendMessage('unknownsequence_notice', **args)
 728
 729                 conn.restart_bootmanager('boot')
 730
 731                 # NOTE: Do not set the pflags value for this sequence if it's unknown.
 732                 # This way, we can check it again after we've fixed it.
 733                 flag_set = False
 734
 735         else:
 736
 737                 if   sequences[s] == "restart_bootmanager_boot":
 738                         print "...Restarting BootManager.py on %s "%hostname
 739                         conn.restart_bootmanager('boot')
 740                 elif sequences[s] == "restart_bootmanager_rins":
 741                         print "...Restarting BootManager.py on %s "%hostname
 742                         conn.restart_bootmanager('reinstall')
 743                 elif sequences[s] == "restart_node_rins":
 744                         conn.restart_node('reinstall')
 745                 elif sequences[s] == "restart_node_boot":
 746                         conn.restart_node('boot')
 747                 elif sequences[s] == "repair_node_keys":
 748                         if conn.compare_and_repair_nodekeys():
 749                                 # the keys either are in sync or were forced in sync.
 750                                 # so try to reboot the node again.
 751                                 conn.restart_bootmanager('reinstall')
 752                                 pass
 753                         else:
 754                                 # there was some failure to synchronize the keys.
 755                                 print "...Unable to repair node keys on %s" %hostname
 756
 757                 elif sequences[s] == "suspect_error_email":
 758                         args = {}
 759                         args['hostname'] = hostname
 760                         args['sequence'] = s
 761                         args['bmlog'] = conn.get_bootmanager_log().read()
 762                         args['viart'] = False
 763
 764                         sitehist.sendMessage('unknownsequence_notice', **args)
 765                         conn.restart_bootmanager('boot')
 766
 767                 # TODO: differentiate this and the 'nodenetwork_email' actions.
 768                 elif sequences[s] == "update_node_config_email":
 769
 770                         if not found_within(recent_actions, 'nodeconfig_notice', 3):
 771                                 args = {}
 772                                 args['hostname'] = hostname
 773                                 sitehist.sendMessage('nodeconfig_notice', **args)
 774                                 conn.dump_plconf_file()
 775
 776                 elif sequences[s] == "nodenetwork_email":
 777
 778                         if not found_within(recent_actions, 'nodeconfig_notice', 3):
 779                                 args = {}
 780                                 args['hostname'] = hostname
 781                                 args['bmlog'] = conn.get_bootmanager_log().read()
 782                                 sitehist.sendMessage('nodeconfig_notice', **args)
 783                                 conn.dump_plconf_file()
 784
 785                 elif sequences[s] == "update_bootcd_email":
 786
 787                         if not found_within(recent_actions, 'newalphacd_notice', 3):
 788                                 args = {}
 789                                 args.update(getconf.getconf(hostname)) # NOTE: Generates boot images for the user:
 790                                 args['hostname'] = hostname
 791
 792                                 sitehist.sendMessage('newalphacd_notice', **args)
 793
 794                                 print "\tDisabling %s due to out-of-date BOOTCD" % hostname
 795
 796                 elif sequences[s] == "broken_hardware_email":
 797                         # MAKE An ACTION record that this host has failed hardware.  May
 798                         # require either an exception "/minhw" or other manual intervention.
 799                         # Definitely need to send out some more EMAIL.
 800                         # TODO: email notice of broken hardware
 801                         if not found_within(recent_actions, 'baddisk_notice', 1):
 802                                 print "...NOTIFYING OWNERS OF BROKEN HARDWARE on %s!!!" % hostname
 803                                 args = {}
 804                                 args['hostname'] = hostname
 805                                 args['log'] = conn.get_dmesg().read()
 806
 807                                 sitehist.sendMessage('baddisk_notice', **args)
 808                                 conn.set_nodestate('disable')
 809
 810                 elif sequences[s] == "update_hardware_email":
 811                         if not found_within(recent_actions, 'minimalhardware_notice', 1):
 812                                 print "...NOTIFYING OWNERS OF MINIMAL HARDWARE FAILURE on %s!!!" % hostname
 813                                 args = {}
 814                                 args['hostname'] = hostname
 815                                 args['bmlog'] = conn.get_bootmanager_log().read()
 816                                 sitehist.sendMessage('minimalhardware_notice', **args)
 817
 818                 elif sequences[s] == "bad_dns_email":
 819                         if not found_within(recent_actions, 'baddns_notice', 1):
 820                                 print "...NOTIFYING OWNERS OF DNS FAILURE on %s!!!" % hostname
 821                                 args = {}
 822                                 try:
 823                                         node = plccache.GetNodeByName(hostname)
 824                                         net = api.GetInterfaces(node['interface_ids'])[0]
 825                                 except:
 826                                         email_exception()
 827                                         print traceback.print_exc()
 828                                         # TODO: api error. skip email, b/c all info is not available,
 829                                         # flag_set will not be recorded.
 830                                         return False
 831                                 nodenet_str = network_config_to_str(net)
 832
 833                                 args['hostname'] = hostname
 834                                 args['network_config'] = nodenet_str
 835                                 args['interface_id'] = net['interface_id']
 836
 837                                 sitehist.sendMessage('baddns_notice', **args)
 838
 839         return True
 840
 841
 842 # MAIN -------------------------------------------------------------------
 843
 844 def main():
 845         from monitor import parser as parsermodule
 846         parser = parsermodule.getParser()
 847
 848         parser.set_defaults(child=False, collect=False, nosetup=False, verbose=False,
 849                                                 force=None, quiet=False)
 850         parser.add_option("", "--child", dest="child", action="store_true",
 851                                                 help="This is the child mode of this process.")
 852         parser.add_option("", "--force", dest="force", metavar="boot_state",
 853                                                 help="Force a boot state passed to BootManager.py.")
 854         parser.add_option("", "--quiet", dest="quiet", action="store_true",
 855                                                 help="Extra quiet output messages.")
 856         parser.add_option("", "--verbose", dest="verbose", action="store_true",
 857                                                 help="Extra debug output messages.")
 858         parser.add_option("", "--nonet", dest="nonet", action="store_true",
 859                                                 help="Do not setup the network, use existing log files to re-run a test pass.")
 860         parser.add_option("", "--collect", dest="collect", action="store_true",
 861                                                 help="No action, just collect dmesg, and bm.log")
 862         parser.add_option("", "--nosetup", dest="nosetup", action="store_true",
 863                                                 help="Do not perform the orginary setup phase.")
 864
 865         parser = parsermodule.getParser(['nodesets', 'defaults'], parser)
 866         config = parsermodule.parse_args(parser)
 867
 868         if config.nodelist:
 869                 nodes = config.getListFromFile(config.nodelist)
 870         elif config.node:
 871                 nodes = [ config.node ]
 872         else:
 873                 parser.print_help()
 874                 sys.exit(1)
 875
 876         for node in nodes:
 877                 # get sitehist
 878                 lb = plccache.plcdb_hn2lb[node]
 879                 sitehist = SiteInterface.get_or_make(loginbase=lb)
 880                 #reboot(node, config)
 881                 restore(sitehist, node, config=None, forced_action=None)
 882
 883 if __name__ == "__main__":
 884         main()