monitor_policy.py

   1 import config
   2 import database
   3 import time
   4 import mailer
   5 from unified_model import cmpCategoryVal
   6 import sys
   7 import emailTxt
   8 import string
   9
  10 from rt import is_host_in_rt_tickets
  11 import plc
  12
  13 def get_ticket_id(record):
  14         if 'ticket_id' in record and record['ticket_id'] is not "" and record['ticket_id'] is not None:
  15                 return record['ticket_id']
  16         elif            'found_rt_ticket' in record and \
  17                  record['found_rt_ticket'] is not "" and \
  18                  record['found_rt_ticket'] is not None:
  19                 return record['found_rt_ticket']
  20         else:
  21                 return None
  22
  23 # Time to enforce policy
  24 POLSLEEP = 7200
  25
  26 # Where to email the summary
  27 SUMTO = "soltesz@cs.princeton.edu"
  28 TECHEMAIL="tech-%s@sites.planet-lab.org"
  29 PIEMAIL="pi-%s@sites.planet-lab.org"
  30 SLICEMAIL="%s@slices.planet-lab.org"
  31 PLCEMAIL="support@planet-lab.org"
  32
  33 #Thresholds (DAYS)
  34 SPERMIN = 60
  35 SPERHOUR = 60*60
  36 SPERDAY = 86400
  37 PITHRESH = 7 * SPERDAY
  38 SLICETHRESH = 7 * SPERDAY
  39 # Days before attempting rins again
  40 RINSTHRESH = 5 * SPERDAY
  41
  42 # Days before calling the node dead.
  43 DEADTHRESH = 30 * SPERDAY
  44 # Minimum number of nodes up before squeezing
  45 MINUP = 2
  46
  47 TECH=1
  48 PI=2
  49 USER=4
  50 ADMIN=8
  51
  52 from unified_model import *
  53
  54 class Merge:
  55         def __init__(self, l_merge):
  56                 self.merge_list = l_merge
  57
  58                 # the hostname to loginbase mapping
  59                 self.plcdb_hn2lb = database.dbLoad("plcdb_hn2lb")
  60
  61                 # Previous actions taken on nodes.
  62                 self.act_all = database.if_cached_else(1, "act_all", lambda : {})
  63                 self.findbad = database.if_cached_else(1, "findbad", lambda : {})
  64
  65                 self.cache_all = database.if_cached_else(1, "act_all", lambda : {})
  66                 self.sickdb = {}
  67                 self.mergedb = {}
  68
  69         def run(self):
  70                 # populate sickdb
  71                 self.accumSickSites()
  72                 # read data from findbad and act_all
  73                 self.mergeActionsAndBadDB()
  74                 # pass node_records to RT
  75                 return self.getRecordList()
  76
  77         def accumSickSites(self):
  78                 """
  79                 Take all nodes, from l_diagnose, look them up in the act_all database,
  80                 and insert them into sickdb[] as:
  81
  82                         sickdb[loginbase][nodename] = fb_record
  83                 """
  84                 # look at all problems reported by findbad
  85                 l_nodes = self.findbad['nodes'].keys()
  86                 count = 0
  87                 for nodename in l_nodes:
  88                         if nodename not in self.merge_list:
  89                                 continue                # skip this node, since it's not wanted
  90
  91                         count += 1
  92                         loginbase = self.plcdb_hn2lb[nodename]
  93                         values = self.findbad['nodes'][nodename]['values']
  94
  95                         fb_record = {}
  96                         fb_record['nodename'] = nodename
  97                         try:
  98                                 fb_record['category'] = values['category']
  99                         except:
 100                                 print values
 101                                 print nodename
 102                                 print self.findbad['nodes'][nodename]
 103                                 count -= 1
 104                                 continue
 105                         fb_record['state'] = values['state']
 106                         fb_record['comonstats'] = values['comonstats']
 107                         fb_record['plcnode'] = values['plcnode']
 108                         fb_record['kernel'] = self.getKernel(values['kernel'])
 109                         fb_record['stage'] = "findbad"
 110                         fb_record['message'] = None
 111                         fb_record['bootcd'] = values['bootcd']
 112                         fb_record['args'] = None
 113                         fb_record['info'] = None
 114                         fb_record['time'] = time.time()
 115                         fb_record['date_created'] = time.time()
 116
 117                         if loginbase not in self.sickdb:
 118                                 self.sickdb[loginbase] = {}
 119
 120                         self.sickdb[loginbase][nodename] = fb_record
 121
 122                 print "Found %d nodes" % count
 123
 124         def getKernel(self, unamestr):
 125                 s = unamestr.split()
 126                 if len(s) > 2:
 127                         return s[2]
 128                 else:
 129                         return ""
 130
 131         def mergeActionsAndBadDB(self):
 132                 """
 133                 - Look at the sick node_records as reported in findbad,
 134                 - Then look at the node_records in act_all.
 135
 136                 There are four cases:
 137                 1) Problem in findbad, no problem in act_all
 138                         this ok, b/c it just means it's a new problem
 139                 2) Problem in findbad, problem in act_all
 140                         -Did the problem get better or worse?
 141                                 -If Same, or Worse, then continue looking for open tickets.
 142                                 -If Better, or No problem, then "back-off" penalties.
 143                                         This judgement may need to wait until 'Diagnose()'
 144
 145                 3) No problem in findbad, problem in act_all
 146                         The the node is operational again according to Findbad()
 147
 148                 4) No problem in findbad, no problem in act_all
 149                         There won't be a record in either db, so there's no code.
 150                 """
 151
 152                 sorted_sites = self.sickdb.keys()
 153                 sorted_sites.sort()
 154                 # look at all problems reported by findbad
 155                 for loginbase in sorted_sites:
 156                         d_fb_nodes = self.sickdb[loginbase]
 157                         sorted_nodes = d_fb_nodes.keys()
 158                         sorted_nodes.sort()
 159                         for nodename in sorted_nodes:
 160                                 fb_record = self.sickdb[loginbase][nodename]
 161                                 x = fb_record
 162                                 if loginbase not in self.mergedb:
 163                                         self.mergedb[loginbase] = {}
 164
 165                                 # take the info either from act_all or fb-record.
 166                                 # if node not in act_all
 167                                 #       then take it from fbrecord, obviously.
 168                                 # else node in act_all
 169                                 #   if act_all == 0 length (no previous records)
 170                                 #               then take it from fbrecord.
 171                                 #   else
 172                                 #           take it from act_all.
 173                                 #
 174
 175                                 # We must compare findbad state with act_all state
 176                                 if nodename not in self.act_all:
 177                                         # 1) ok, b/c it's a new problem. set ticket_id to null
 178                                         self.mergedb[loginbase][nodename] = {}
 179                                         self.mergedb[loginbase][nodename].update(x)
 180                                         self.mergedb[loginbase][nodename]['ticket_id'] = ""
 181                                         self.mergedb[loginbase][nodename]['prev_category'] = "NORECORD"
 182                                 else:
 183                                         if len(self.act_all[nodename]) == 0:
 184                                                 self.mergedb[loginbase][nodename] = {}
 185                                                 self.mergedb[loginbase][nodename].update(x)
 186                                                 self.mergedb[loginbase][nodename]['ticket_id'] = ""
 187                                                 self.mergedb[loginbase][nodename]['prev_category'] = "NORECORD"
 188                                         else:
 189                                                 y = self.act_all[nodename][0]
 190                                                 y['prev_category'] = y['category']
 191
 192                                                 self.mergedb[loginbase][nodename] = {}
 193                                                 self.mergedb[loginbase][nodename].update(y)
 194                                                 self.mergedb[loginbase][nodename]['comonstats'] = x['comonstats']
 195                                                 self.mergedb[loginbase][nodename]['category']   = x['category']
 196                                                 self.mergedb[loginbase][nodename]['state'] = x['state']
 197                                                 self.mergedb[loginbase][nodename]['kernel']=x['kernel']
 198                                                 self.mergedb[loginbase][nodename]['bootcd']=x['bootcd']
 199                                                 self.mergedb[loginbase][nodename]['plcnode']=x['plcnode']
 200                                                 ticket = get_ticket_id(self.mergedb[loginbase][nodename])
 201                                                 self.mergedb[loginbase][nodename]['rt'] = mailer.getTicketStatus(ticket)
 202
 203                                         # delete the entry from cache_all to keep it out of case 3)
 204                                         del self.cache_all[nodename]
 205
 206                 # 3) nodes that remin in cache_all were not identified by findbad.
 207                 #        Do we keep them or not?
 208                 #   NOTE: i think that since the categories are performed before this
 209                 #               step now, and by a monitor-controlled agent.
 210
 211                 return
 212
 213         def getRecordList(self):
 214                 sorted_sites = self.mergedb.keys()
 215                 sorted_sites.sort()
 216                 ret_list = []
 217
 218                 # look at all problems reported by merge
 219                 for loginbase in sorted_sites:
 220                         d_merge_nodes = self.mergedb[loginbase]
 221                         for nodename in d_merge_nodes.keys():
 222                                 record = self.mergedb[loginbase][nodename]
 223                                 ret_list.append(record)
 224
 225                 return ret_list
 226
 227 class RT:
 228         def __init__(self, record_list, dbTickets, l_ticket_blacklist, target = None):
 229                 # Time of last update of ticket DB
 230                 self.record_list = record_list
 231                 self.dbTickets = dbTickets
 232                 self.lastupdated = 0
 233                 self.l_ticket_blacklist = l_ticket_blacklist
 234                 self.tickets = {}
 235
 236         def run(self):
 237                 self.count = 0
 238                 ret_list = []
 239                 for diag_node in self.record_list:
 240                         if diag_node != None:
 241                                 host = diag_node['nodename']
 242                                 (b_host_inticket, r_ticket) = is_host_in_rt_tickets(host, \
 243                                                                                                         self.l_ticket_blacklist, \
 244                                                                                                         self.dbTickets)
 245                                 diag_node['found_rt_ticket'] = None
 246                                 if b_host_inticket:
 247                                         #logger.debug("RT: found tickets for %s" %host)
 248                                         diag_node['found_rt_ticket'] = r_ticket['ticket_id']
 249
 250                                 else:
 251                                         if r_ticket is not None:
 252                                                 print "Ignoring ticket %s" % r_ticket['ticket_id']
 253                                                 # TODO: why do i return the ticket id for a
 254                                                 #               blacklisted ticket id?
 255                                                 #diag_node['found_rt_ticket'] = r_ticket['ticket_id']
 256                                         self.count = self.count + 1
 257
 258                                 ret_list.append(diag_node)
 259
 260                 #print "RT processed %d nodes with noticket" % self.count
 261                 #logger.debug("RT filtered %d noticket nodes" % self.count)
 262                 return ret_list
 263
 264 class Diagnose:
 265         def __init__(self, record_list):
 266                 self.record_list = record_list
 267                 self.plcdb_hn2lb = database.dbLoad("plcdb_hn2lb")
 268                 self.findbad = database.if_cached_else(1, "findbad", lambda : {})
 269
 270                 self.diagnose_in = {}
 271                 self.diagnose_out = {}
 272
 273         def run(self):
 274                 self.accumSickSites()
 275
 276                 #logger.debug("Accumulated %d sick sites" % len(self.diagnose_in.keys()))
 277
 278                 try:
 279                         stats = self.diagnoseAll()
 280                 except Exception, err:
 281                         print "----------------"
 282                         import traceback
 283                         print traceback.print_exc()
 284                         from nodecommon import email_exception
 285                         email_exception()
 286                         print err
 287                         #if config.policysavedb:
 288                         sys.exit(1)
 289
 290                 #print_stats("sites_observed", stats)
 291                 #print_stats("sites_diagnosed", stats)
 292                 #print_stats("nodes_diagnosed", stats)
 293
 294                 return self.diagnose_out
 295
 296         def accumSickSites(self):
 297                 """
 298                 Take all nodes, from l_diagnose, look them up in the diagnose_out database,
 299                 and insert them into diagnose_in[] as:
 300
 301                         diagnose_in[loginbase] = [diag_node1, diag_node2, ...]
 302                 """
 303                 for node_record in self.record_list:
 304
 305                         nodename = node_record['nodename']
 306                         loginbase = self.plcdb_hn2lb[nodename]
 307
 308                         if loginbase not in self.diagnose_in:
 309                                 self.diagnose_in[loginbase] = {}
 310
 311                         self.diagnose_in[loginbase][nodename] = node_record
 312
 313                 return
 314
 315         def diagnoseAll(self):
 316                 i_sites_observed = 0
 317                 i_sites_diagnosed = 0
 318                 i_nodes_diagnosed = 0
 319                 i_nodes_actedon = 0
 320                 i_sites_emailed = 0
 321                 l_allsites = []
 322
 323                 sorted_sites = self.diagnose_in.keys()
 324                 sorted_sites.sort()
 325                 self.diagnose_out= {}
 326                 for loginbase in sorted_sites:
 327                         l_allsites += [loginbase]
 328
 329                         d_diag_nodes = self.diagnose_in[loginbase]
 330                         d_act_records = self.__diagnoseSite(loginbase, d_diag_nodes)
 331                         # store records in diagnose_out, for saving later.
 332                         self.diagnose_out.update(d_act_records)
 333
 334                         if len(d_act_records[loginbase]['nodes'].keys()) > 0:
 335                                 i_nodes_diagnosed += (len(d_act_records[loginbase]['nodes'].keys()))
 336                                 i_sites_diagnosed += 1
 337                         i_sites_observed += 1
 338
 339                 return {'sites_observed': i_sites_observed,
 340                                 'sites_diagnosed': i_sites_diagnosed,
 341                                 'nodes_diagnosed': i_nodes_diagnosed,
 342                                 'allsites':l_allsites}
 343
 344                 pass
 345
 346         def __getDaysDown(self, diag_record, nodename):
 347                 daysdown = -1
 348                 if diag_record['comonstats']['sshstatus'] != "null":
 349                         daysdown = int(diag_record['comonstats']['sshstatus']) // (60*60*24)
 350                 elif diag_record['comonstats']['lastcotop'] != "null":
 351                         daysdown = int(diag_record['comonstats']['lastcotop']) // (60*60*24)
 352                 else:
 353                         now = time.time()
 354                         last_contact = diag_record['plcnode']['last_contact']
 355                         if last_contact == None:
 356                                 # the node has never been up, so give it a break
 357                                 daysdown = -1
 358                         else:
 359                                 diff = now - last_contact
 360                                 daysdown = diff // (60*60*24)
 361                 return daysdown
 362
 363         def __getStrDaysDown(self, diag_record, nodename):
 364                 daysdown = self.__getDaysDown(diag_record, nodename)
 365                 if daysdown > 0:
 366                         return "(%d days down)"%daysdown
 367                 else:
 368                         return "Unknown number of days"
 369
 370         def __getCDVersion(self, diag_record, nodename):
 371                 cdversion = ""
 372                 #print "Getting kernel for: %s" % diag_record['nodename']
 373                 cdversion = diag_record['kernel']
 374                 return cdversion
 375
 376         def __diagnoseSite(self, loginbase, d_diag_nodes):
 377                 """
 378                 d_diag_nodes are diagnose_in entries.
 379                 """
 380                 d_diag_site = {loginbase : { 'config' :
 381                                                                                                 {'squeeze': False,
 382                                                                                                  'email': False
 383                                                                                                 },
 384                                                                         'nodes': {}
 385                                                                         }
 386                                            }
 387                 sorted_nodes = d_diag_nodes.keys()
 388                 sorted_nodes.sort()
 389                 for nodename in sorted_nodes:
 390                         node_record = d_diag_nodes[nodename]
 391                         diag_record = self.__diagnoseNode(loginbase, node_record)
 392
 393                         if diag_record != None:
 394                                 d_diag_site[loginbase]['nodes'][nodename] = diag_record
 395
 396                                 # NOTE: improvement means, we need to act/squeeze and email.
 397                                 #print "DIAG_RECORD", diag_record
 398                                 if 'monitor-end-record' in diag_record['stage'] or \
 399                                    'nmreset' in diag_record['stage']:
 400                                 #       print "resetting loginbase!"
 401                                         d_diag_site[loginbase]['config']['squeeze'] = True
 402                                         d_diag_site[loginbase]['config']['email'] = True
 403                                 #else:
 404                                 #       print "NO IMPROVEMENT!!!!"
 405                         else:
 406                                 pass # there is nothing to do for this node.
 407
 408                 # NOTE: these settings can be overridden by command line arguments,
 409                 #       or the state of a record, i.e. if already in RT's Support Queue.
 410                 pf = PersistFlags(loginbase, 1, db='site_persistflags')
 411                 nodes_up = pf.nodes_up
 412                 if nodes_up < MINUP:
 413                         d_diag_site[loginbase]['config']['squeeze'] = True
 414
 415                 max_slices = self.getMaxSlices(loginbase)
 416                 num_nodes = pf.nodes_total #self.getNumNodes(loginbase)
 417                 # NOTE: when max_slices == 0, this is either a new site (the old way)
 418                 #       or an old disabled site from previous monitor (before site['enabled'])
 419                 if nodes_up < num_nodes and max_slices != 0:
 420                         d_diag_site[loginbase]['config']['email'] = True
 421
 422                 if len(d_diag_site[loginbase]['nodes'].keys()) > 0:
 423                         print "SITE: %20s : %d nodes up, at most" % (loginbase, nodes_up)
 424
 425                 return d_diag_site
 426
 427         def diagRecordByCategory(self, node_record):
 428                 nodename = node_record['nodename']
 429                 category = node_record['category']
 430                 state    = node_record['state']
 431                 loginbase = self.plcdb_hn2lb[nodename]
 432                 diag_record = None
 433
 434                 if  "ERROR" in category:        # i.e. "DOWN"
 435                         diag_record = {}
 436                         diag_record.update(node_record)
 437                         daysdown = self.__getDaysDown(diag_record, nodename)
 438                         #if daysdown < 7:
 439                         #       format = "DIAG: %20s : %-40s Down only %s days  NOTHING DONE"
 440                         #       print format % (loginbase, nodename, daysdown)
 441                         #       return None
 442
 443                         s_daysdown = self.__getStrDaysDown(diag_record, nodename)
 444                         diag_record['message'] = emailTxt.mailtxt.newdown
 445                         diag_record['args'] = {'nodename': nodename}
 446                         diag_record['info'] = (nodename, s_daysdown, "")
 447
 448                         #if 'reboot_node_failed' in node_record:
 449                         #       # there was a previous attempt to use the PCU.
 450                         #       if node_record['reboot_node_failed'] == False:
 451                         #               # then the last attempt apparently, succeeded.
 452                         #               # But, the category is still 'ERROR'.  Therefore, the
 453                         #               # PCU-to-Node mapping is broken.
 454                         #               #print "Setting message for ERROR node to PCU2NodeMapping: %s" % nodename
 455                         #               diag_record['message'] = emailTxt.mailtxt.pcutonodemapping
 456                         #               diag_record['email_pcu'] = True
 457
 458                         if diag_record['ticket_id'] == "":
 459                                 diag_record['log'] = "DOWN: %20s : %-40s == %20s %s" % \
 460                                         (loginbase, nodename, diag_record['info'][1:], diag_record['found_rt_ticket'])
 461                         else:
 462                                 diag_record['log'] = "DOWN: %20s : %-40s == %20s %s" % \
 463                                         (loginbase, nodename, diag_record['info'][1:], diag_record['ticket_id'])
 464
 465                 elif "OLDBOOTCD" in category:
 466                         # V2 boot cds as determined by findbad
 467                         s_daysdown = self.__getStrDaysDown(node_record, nodename)
 468                         s_cdversion = self.__getCDVersion(node_record, nodename)
 469                         diag_record = {}
 470                         diag_record.update(node_record)
 471                         #if "2.4" in diag_record['kernel'] or "v2" in diag_record['bootcd']:
 472                         diag_record['message'] = emailTxt.mailtxt.newbootcd
 473                         diag_record['args'] = {'nodename': nodename}
 474                         diag_record['info'] = (nodename, s_daysdown, s_cdversion)
 475                         if diag_record['ticket_id'] == "":
 476                                 diag_record['log'] = "BTCD: %20s : %-40s == %20s %20s %s" % \
 477                                                                         (loginbase, nodename, diag_record['kernel'],
 478                                                                          diag_record['bootcd'], diag_record['found_rt_ticket'])
 479                         else:
 480                                 diag_record['log'] = "BTCD: %20s : %-40s == %20s %20s %s" % \
 481                                                                         (loginbase, nodename, diag_record['kernel'],
 482                                                                          diag_record['bootcd'], diag_record['ticket_id'])
 483
 484                 elif "PROD" in category:
 485                         if "DEBUG" in state:
 486                                 # Not sure what to do with these yet.  Probably need to
 487                                 # reboot, and email.
 488                                 print "DEBG: %20s : %-40s  NOTHING DONE" % (loginbase, nodename)
 489                                 return None
 490                         elif "BOOT" in state:
 491                                 # no action needed.
 492                                 # TODO: remove penalties, if any are applied.
 493                                 now = time.time()
 494                                 last_contact = node_record['plcnode']['last_contact']
 495                                 if last_contact == None:
 496                                         time_diff = 0
 497                                 else:
 498                                         time_diff = now - last_contact;
 499
 500                                 if 'improvement' in node_record['stage']:
 501                                         # then we need to pass this on to 'action'
 502                                         diag_record = {}
 503                                         diag_record.update(node_record)
 504                                         diag_record['message'] = emailTxt.mailtxt.newthankyou
 505                                         diag_record['args'] = {'nodename': nodename}
 506                                         diag_record['info'] = (nodename, node_record['prev_category'],
 507                                                                                                          node_record['category'])
 508                                         #if 'email_pcu' in diag_record:
 509                                         #       if diag_record['email_pcu']:
 510                                         #               # previously, the pcu failed to reboot, so send
 511                                         #               # email. Now, reset these values to try the reboot
 512                                         #               # again.
 513                                         #               diag_record['email_pcu'] = False
 514                                         #               del diag_record['reboot_node_failed']
 515
 516                                         if diag_record['ticket_id'] == "":
 517                                                 diag_record['log'] = "IMPR: %20s : %-40s == %20s %20s %s %s" % \
 518                                                                         (loginbase, nodename, diag_record['stage'],
 519                                                                          state, category, diag_record['found_rt_ticket'])
 520                                         else:
 521                                                 diag_record['log'] = "IMPR: %20s : %-40s == %20s %20s %s %s" % \
 522                                                                         (loginbase, nodename, diag_record['stage'],
 523                                                                          state, category, diag_record['ticket_id'])
 524                                         return diag_record
 525                                 #elif time_diff >= 6*SPERHOUR:
 526                                 #       # heartbeat is older than 30 min.
 527                                 #       # then reset NM.
 528                                 #       #print "Possible NM problem!! %s - %s = %s" % (now, last_contact, time_diff)
 529                                 #       diag_record = {}
 530                                 #       diag_record.update(node_record)
 531                                 #       diag_record['message'] = emailTxt.mailtxt.NMReset
 532                                 #       diag_record['args'] = {'nodename': nodename}
 533                                 #       diag_record['stage'] = "nmreset"
 534                                 #       diag_record['info'] = (nodename,
 535                                 #                                                       node_record['prev_category'],
 536                                 #                                                       node_record['category'])
 537                                 #       if diag_record['ticket_id'] == "":
 538                                 #               diag_record['log'] = "NM  : %20s : %-40s == %20s %20s %s %s" % \
 539                                 #                                       (loginbase, nodename, diag_record['stage'],
 540                                 #                                        state, category, diag_record['found_rt_ticket'])
 541                                 #       else:
 542                                 #               diag_record['log'] = "NM  : %20s : %-40s == %20s" % \
 543                                 #                                       (loginbase, nodename, diag_record['stage'])
 544 #
 545 #                                       return diag_record
 546                                 else:
 547                                         return None
 548                         else:
 549                                 # unknown
 550                                 pass
 551                 elif "ALPHA"    in category:
 552                         pass
 553                 elif "clock_drift" in category:
 554                         pass
 555                 elif "dns"    in category:
 556                         pass
 557                 elif "filerw"    in category:
 558                         pass
 559                 else:
 560                         print "Unknown category!!!! %s" % category
 561                         sys.exit(1)
 562
 563                 return diag_record
 564
 565         def __diagnoseNode(self, loginbase, node_record):
 566                 # TODO: change the format of the hostname in this
 567                 #               record to something more natural.
 568                 nodename                = node_record['nodename']
 569                 category                = node_record['category']
 570                 prev_category   = node_record['prev_category']
 571                 state                   = node_record['state']
 572                 #if 'prev_category' in node_record:
 573                 #       prev_category = node_record['prev_category']
 574                 #else:
 575                 #       prev_category = "ERROR"
 576                 if node_record['prev_category'] != "NORECORD":
 577
 578                         val = cmpCategoryVal(category, prev_category)
 579                         print "%s went from %s -> %s" % (nodename, prev_category, category)
 580                         if prev_category == "UNKNOWN" and category == "PROD":
 581                                 # sending too many thank you notes to people that don't
 582                                 # deserve them.
 583                                 # TODO: not sure what effect this will have on the node
 584                                 # status, though...
 585                                 return None
 586
 587                         if val == 1:
 588                                 # improved
 589                                 if node_record['ticket_id'] == "" or node_record['ticket_id'] == None:
 590                                         print "closing record with no ticket: ", node_record['nodename']
 591                                         node_record['action'] = ['close_rt']
 592                                         node_record['message'] = None
 593                                         node_record['stage'] = 'monitor-end-record'
 594                                         return node_record
 595                                 else:
 596                                         node_record['stage'] = 'improvement'
 597
 598                                 #if 'monitor-end-record' in node_record['stage']:
 599                                 #       # just ignore it if it's already ended.
 600                                 #       # otherwise, the status should be worse, and we won't get
 601                                 #       # here.
 602                                 #       print "monitor-end-record: ignoring ", node_record['nodename']
 603                                 #       return None
 604 #
 605 #                                       #return None
 606                         elif val == -1:
 607                                 # current category is worse than previous, carry on
 608                                 pass
 609                         else:
 610                                 #values are equal, carry on.
 611                                 #print "why are we here?"
 612                                 pass
 613
 614                 if 'rt' in node_record and 'Status' in node_record['rt']:
 615                         if node_record['stage'] == 'ticket_waitforever':
 616                                 if 'resolved' in node_record['rt']['Status']:
 617                                         print "ending waitforever record for: ", node_record['nodename']
 618                                         node_record['action'] = ['noop']
 619                                         node_record['message'] = None
 620                                         node_record['stage'] = 'monitor-end-record'
 621                                         print "oldlog: %s" % node_record['log'],
 622                                         print "%15s" % node_record['action']
 623                                         return node_record
 624                                 if 'new' in node_record['rt']['Status'] and \
 625                                         'Queue' in node_record['rt'] and \
 626                                         'Monitor' in node_record['rt']['Queue']:
 627
 628                                         print "RESETTING stage to findbad"
 629                                         node_record['stage'] = 'findbad'
 630
 631                 #### COMPARE category and prev_category
 632                 # if not_equal
 633                 #       then assign a stage based on relative priorities
 634                 # else equal
 635                 #       then check category for stats.
 636                 diag_record = self.diagRecordByCategory(node_record)
 637                 if diag_record == None:
 638                         #print "diag_record == None"
 639                         return None
 640
 641                 #### found_RT_ticket
 642                 # TODO: need to record time found, and maybe add a stage for acting on it...
 643                 # NOTE: after found, if the support ticket is resolved, the block is
 644                 #               not removed. How to remove the block on this?
 645
 646                 #if 'found_rt_ticket' in diag_record and \
 647                 #       diag_record['found_rt_ticket'] is not None:
 648                 #       if diag_record['stage'] is not 'improvement':
 649                 #               diag_record['stage'] = 'ticket_waitforever'
 650
 651                 current_time = time.time()
 652                 # take off four days, for the delay that database caused.
 653                 # TODO: generalize delays at PLC, and prevent enforcement when there
 654                 #               have been no emails.
 655                 # NOTE: 7*SPERDAY exists to offset the 'bad week'
 656                 #delta = current_time - diag_record['time'] - 7*SPERDAY
 657                 delta = current_time - diag_record['time']
 658
 659                 message = diag_record['message']
 660                 act_record = {}
 661                 act_record.update(diag_record)
 662
 663                 #### DIAGNOSE STAGES
 664                 if   'findbad' in diag_record['stage']:
 665                         # The node is bad, and there's no previous record of it.
 666                         act_record['email'] = TECH
 667                         act_record['action'] = ['noop']
 668                         act_record['message'] = message[0]
 669                         act_record['stage'] = 'stage_actinoneweek'
 670
 671                 elif 'nmreset' in diag_record['stage']:
 672                         act_record['email']  = ADMIN
 673                         act_record['action'] = ['reset_nodemanager']
 674                         act_record['message'] = message[0]
 675                         act_record['stage']  = 'nmreset'
 676                         return None
 677
 678                 elif 'reboot_node' in diag_record['stage']:
 679                         act_record['email'] = TECH
 680                         act_record['action'] = ['noop']
 681                         act_record['message'] = message[0]
 682                         act_record['stage'] = 'stage_actinoneweek'
 683
 684                 elif 'improvement' in diag_record['stage']:
 685                         # - backoff previous squeeze actions (slice suspend, nocreate)
 686                         # TODO: add a backoff_squeeze section... Needs to runthrough
 687                         print "backing off of %s" % nodename
 688                         act_record['action'] = ['close_rt']
 689                         act_record['message'] = message[0]
 690                         act_record['stage'] = 'monitor-end-record'
 691
 692                 elif 'actinoneweek' in diag_record['stage']:
 693                         if delta >= 7 * SPERDAY:
 694                                 act_record['email'] = TECH | PI
 695                                 act_record['stage'] = 'stage_actintwoweeks'
 696                                 act_record['message'] = message[1]
 697                                 act_record['action'] = ['nocreate' ]
 698                                 act_record['time'] = current_time               # reset clock for waitforever
 699                         elif delta >= 3* SPERDAY and not 'second-mail-at-oneweek' in act_record:
 700                                 act_record['email'] = TECH
 701                                 act_record['message'] = message[0]
 702                                 act_record['action'] = ['sendmailagain-waitforoneweekaction' ]
 703                                 act_record['second-mail-at-oneweek'] = True
 704                         else:
 705                                 act_record['message'] = None
 706                                 act_record['action'] = ['waitforoneweekaction' ]
 707                                 print "ignoring this record for: %s" % act_record['nodename']
 708                                 return None                     # don't send if there's no action
 709
 710                 elif 'actintwoweeks' in diag_record['stage']:
 711                         if delta >= 7 * SPERDAY:
 712                                 act_record['email'] = TECH | PI | USER
 713                                 act_record['stage'] = 'stage_waitforever'
 714                                 act_record['message'] = message[2]
 715                                 act_record['action'] = ['suspendslices']
 716                                 act_record['time'] = current_time               # reset clock for waitforever
 717                         elif delta >= 3* SPERDAY and not 'second-mail-at-twoweeks' in act_record:
 718                                 act_record['email'] = TECH | PI
 719                                 act_record['message'] = message[1]
 720                                 act_record['action'] = ['sendmailagain-waitfortwoweeksaction' ]
 721                                 act_record['second-mail-at-twoweeks'] = True
 722                         else:
 723                                 act_record['message'] = None
 724                                 act_record['action'] = ['waitfortwoweeksaction']
 725                                 return None                     # don't send if there's no action
 726
 727                 elif 'ticket_waitforever' in diag_record['stage']:
 728                         act_record['email'] = TECH
 729                         if 'first-found' not in act_record:
 730                                 act_record['first-found'] = True
 731                                 act_record['log'] += " firstfound"
 732                                 act_record['action'] = ['ticket_waitforever']
 733                                 act_record['message'] = None
 734                                 act_record['time'] = current_time
 735                         else:
 736                                 if delta >= 7*SPERDAY:
 737                                         act_record['action'] = ['ticket_waitforever']
 738                                         act_record['message'] = None
 739                                         act_record['time'] = current_time               # reset clock
 740                                 else:
 741                                         act_record['action'] = ['ticket_waitforever']
 742                                         act_record['message'] = None
 743                                         return None
 744
 745                 elif 'waitforever' in diag_record['stage']:
 746                         # more than 3 days since last action
 747                         # TODO: send only on weekdays.
 748                         # NOTE: expects that 'time' has been reset before entering waitforever stage
 749                         if delta >= 3*SPERDAY:
 750                                 act_record['action'] = ['email-againwaitforever']
 751                                 act_record['message'] = message[2]
 752                                 act_record['time'] = current_time               # reset clock
 753                         else:
 754                                 act_record['action'] = ['waitforever']
 755                                 act_record['message'] = None
 756                                 return None                     # don't send if there's no action
 757
 758                 else:
 759                         # There is no action to be taken, possibly b/c the stage has
 760                         # already been performed, but diagnose picked it up again.
 761                         # two cases,
 762                         #       1. stage is unknown, or
 763                         #       2. delta is not big enough to bump it to the next stage.
 764                         # TODO: figure out which. for now assume 2.
 765                         print "UNKNOWN stage for %s; nothing done" % nodename
 766                         act_record['action'] = ['unknown']
 767                         act_record['message'] = message[0]
 768
 769                         act_record['email'] = TECH
 770                         act_record['action'] = ['noop']
 771                         act_record['message'] = message[0]
 772                         act_record['stage'] = 'stage_actinoneweek'
 773                         act_record['time'] = current_time               # reset clock
 774                         #print "Exiting..."
 775                         #return None
 776                         #sys.exit(1)
 777
 778                 print "%s" % act_record['log'],
 779                 print "%15s" % act_record['action']
 780                 return act_record
 781
 782         def getMaxSlices(self, loginbase):
 783                 # if sickdb has a loginbase, then it will have at least one node.
 784                 site_stats = None
 785
 786                 for nodename in self.diagnose_in[loginbase].keys():
 787                         if nodename in self.findbad['nodes']:
 788                                 site_stats = self.findbad['nodes'][nodename]['values']['plcsite']
 789                                 break
 790
 791                 if site_stats == None:
 792                         raise Exception, "loginbase with no nodes in findbad"
 793                 else:
 794                         return site_stats['max_slices']
 795
 796         def getNumNodes(self, loginbase):
 797                 # if sickdb has a loginbase, then it will have at least one node.
 798                 site_stats = None
 799
 800                 for nodename in self.diagnose_in[loginbase].keys():
 801                         if nodename in self.findbad['nodes']:
 802                                 site_stats = self.findbad['nodes'][nodename]['values']['plcsite']
 803                                 break
 804
 805                 if site_stats == None:
 806                         raise Exception, "loginbase with no nodes in findbad"
 807                 else:
 808                         return site_stats['num_nodes']
 809
 810         """
 811         Returns number of up nodes as the total number *NOT* in act_all with a
 812         stage other than 'steady-state' .
 813         """
 814         def getUpAtSite(self, loginbase, d_diag_site):
 815                 # TODO: THIS DOESN"T WORK!!! it misses all the 'debug' state nodes
 816                 #               that aren't recorded yet.
 817
 818                 numnodes = self.getNumNodes(loginbase)
 819                 # NOTE: assume nodes we have no record of are ok. (too conservative)
 820                 # TODO: make the 'up' value more representative
 821                 up = numnodes
 822                 for nodename in d_diag_site[loginbase]['nodes'].keys():
 823
 824                         rec = d_diag_site[loginbase]['nodes'][nodename]
 825                         if rec['stage'] != 'monitor-end-record':
 826                                 up -= 1
 827                         else:
 828                                 pass # the node is assumed to be up.
 829
 830                 #if up != numnodes:
 831                 #       print "ERROR: %s total nodes up and down != %d" % (loginbase, numnodes)
 832
 833                 return up
 834
 835 def close_rt_backoff(args):
 836         if 'ticket_id' in args and (args['ticket_id'] != "" and args['ticket_id'] != None):
 837                 mailer.closeTicketViaRT(args['ticket_id'],
 838                                                                 "Ticket CLOSED automatically by SiteAssist.")
 839                 plc.enableSlices(args['hostname'])
 840                 plc.enableSliceCreation(args['hostname'])
 841         return
 842
 843 def reboot_node(args):
 844         host = args['hostname']
 845         return reboot.reboot_policy(host, True, config.debug)
 846
 847 class Action:
 848         def __init__(self, diagnose_out):
 849                 # the hostname to loginbase mapping
 850                 self.plcdb_hn2lb = database.dbLoad("plcdb_hn2lb")
 851
 852                 # Actions to take.
 853                 self.diagnose_db = diagnose_out
 854                 # Actions taken.
 855                 self.act_all   = database.if_cached_else(1, "act_all", lambda : {})
 856
 857                 # A dict of actions to specific functions. PICKLE doesnt' like lambdas.
 858                 self.actions = {}
 859                 self.actions['suspendslices'] = lambda args: plc.suspendSlices(args['hostname'])
 860                 self.actions['nocreate'] = lambda args: plc.removeSliceCreation(args['hostname'])
 861                 self.actions['close_rt'] = lambda args: close_rt_backoff(args)
 862                 self.actions['rins'] = lambda args: plc.nodeBootState(args['hostname'], "rins")
 863                 self.actions['noop'] = lambda args: args
 864                 self.actions['reboot_node'] = lambda args: reboot_node(args)
 865                 self.actions['reset_nodemanager'] = lambda args: args # reset_nodemanager(args)
 866
 867                 self.actions['ticket_waitforever'] = lambda args: args
 868                 self.actions['waitforever'] = lambda args: args
 869                 self.actions['unknown'] = lambda args: args
 870                 self.actions['waitforoneweekaction'] = lambda args: args
 871                 self.actions['waitfortwoweeksaction'] = lambda args: args
 872                 self.actions['sendmailagain-waitforoneweekaction'] = lambda args: args
 873                 self.actions['sendmailagain-waitfortwoweeksaction'] = lambda args: args
 874                 self.actions['email-againwaitforever'] = lambda args: args
 875                 self.actions['email-againticket_waitforever'] = lambda args: args
 876
 877                 self.sickdb = {}
 878
 879         def run(self):
 880                 self.accumSites()
 881                 #logger.debug("Accumulated %d sick sites" % len(self.sickdb.keys()))
 882
 883                 try:
 884                         stats = self.analyseSites()
 885                 except Exception, err:
 886                         print "----------------"
 887                         import traceback
 888                         print traceback.print_exc()
 889                         from nodecommon import email_exception
 890                         email_exception()
 891                         print err
 892                         if config.policysavedb:
 893                                 print "Saving Databases... act_all"
 894                                 database.dbDump("act_all", self.act_all)
 895                                 database.dbDump("diagnose_out", self.diagnose_db)
 896                         sys.exit(1)
 897
 898                 #print_stats("sites_observed", stats)
 899                 #print_stats("sites_diagnosed", stats)
 900                 #print_stats("nodes_diagnosed", stats)
 901                 self.print_stats("sites_emailed", stats)
 902                 #print_stats("nodes_actedon", stats)
 903                 print string.join(stats['allsites'], ",")
 904
 905                 if config.policysavedb:
 906                         print "Saving Databases... act_all"
 907                         #database.dbDump("policy.eventlog", self.eventlog)
 908                         # TODO: remove 'diagnose_out',
 909                         #       or at least the entries that were acted on.
 910                         database.dbDump("act_all", self.act_all)
 911                         database.dbDump("diagnose_out", self.diagnose_db)
 912
 913         def accumSites(self):
 914                 """
 915                 Take all nodes, from l_action, look them up in the diagnose_db database,
 916                 and insert them into sickdb[] as:
 917
 918                 This way only the given l_action nodes will be acted on regardless
 919                 of how many from diagnose_db are available.
 920
 921                         sickdb[loginbase][nodename] = diag_record
 922                 """
 923                 self.sickdb = self.diagnose_db
 924
 925         def __emailSite(self, loginbase, roles, message, args):
 926                 """
 927                 loginbase is the unique site abbreviation, prepended to slice names.
 928                 roles contains TECH, PI, USER roles, and derive email aliases.
 929                 record contains {'message': [<subj>,<body>], 'args': {...}}
 930                 """
 931                 ticket_id = 0
 932                 args.update({'loginbase':loginbase})
 933
 934                 if not config.mail and not config.debug and config.bcc:
 935                         roles = ADMIN
 936                 if config.mail and config.debug:
 937                         roles = ADMIN
 938
 939                 # build targets
 940                 contacts = []
 941                 if ADMIN & roles:
 942                         contacts += [config.email]
 943                 if TECH & roles:
 944                         #contacts += [TECHEMAIL % loginbase]
 945                         contacts += plc.getTechEmails(loginbase)
 946                 if PI & roles:
 947                         #contacts += [PIEMAIL % loginbase]
 948                         contacts += plc.getPIEmails(loginbase)
 949                 if USER & roles:
 950                         contacts += plc.getSliceUserEmails(loginbase)
 951                         slices = plc.slices(loginbase)
 952                         if len(slices) >= 1:
 953                                 print "SLIC: %20s : %d slices" % (loginbase, len(slices))
 954                         else:
 955                                 print "SLIC: %20s : 0 slices" % loginbase
 956
 957                 unique_contacts = set(contacts)
 958                 contacts = [ c for c in unique_contacts ]       # convert back into list
 959
 960                 try:
 961                         subject = message[0] % args
 962                         body = message[1] % args
 963                         if ADMIN & roles:
 964                                 # send only to admin
 965                                 if 'ticket_id' in args:
 966                                         subj = "Re: [PL #%s] %s" % (args['ticket_id'], subject)
 967                                 else:
 968                                         subj = "Re: [PL noticket] %s" % subject
 969                                 mailer.email(subj, body, contacts)
 970                                 ticket_id = args['ticket_id']
 971                         else:
 972                                 ticket_id = mailer.emailViaRT(subject, body, contacts, args['ticket_id'])
 973                 except Exception, err:
 974                         print "exception on message:"
 975                         import traceback
 976                         print traceback.print_exc()
 977                         from nodecommon import email_exception
 978                         email_exception()
 979                         print message
 980
 981                 return ticket_id
 982
 983
 984         def _format_diaginfo(self, diag_node):
 985                 info = diag_node['info']
 986                 if diag_node['stage'] == 'monitor-end-record':
 987                         hlist = "    %s went from '%s' to '%s'\n" % (info[0], info[1], info[2])
 988                 else:
 989                         hlist = "    %s %s - %s\n" % (info[0], info[2], info[1]) #(node,ver,daysdn)
 990                 return hlist
 991
 992
 993         def get_email_args(self, act_recordlist, loginbase=None):
 994
 995                 email_args = {}
 996                 email_args['hostname_list'] = ""
 997                 email_args['url_list'] = ""
 998
 999                 for act_record in act_recordlist:
1000                         email_args['hostname_list'] += act_record['msg_format']
1001                         email_args['hostname'] = act_record['nodename']
1002                         email_args['url_list'] += "\thttp://boot2.planet-lab.org/premade-bootcd-alpha/iso/%s.iso\n"
1003                         email_args['url_list'] += "\thttp://boot2.planet-lab.org/premade-bootcd-alpha/usb/%s.usb\n"
1004                         email_args['url_list'] += "\n"
1005                         if  'plcnode' in act_record and \
1006                                 'pcu_ids' in act_record['plcnode'] and \
1007                                 len(act_record['plcnode']['pcu_ids']) > 0:
1008                                 print "setting 'pcu_id' for email_args %s"%email_args['hostname']
1009                                 email_args['pcu_id'] = act_record['plcnode']['pcu_ids'][0]
1010                         else:
1011                                 email_args['pcu_id'] = "-1"
1012
1013                         if 'ticket_id' in act_record:
1014                                 if act_record['ticket_id'] == 0 or act_record['ticket_id'] == '0':
1015                                         print "Enter the ticket_id for %s @ %s" % (loginbase, act_record['nodename'])
1016                                         sys.stdout.flush()
1017                                         line = sys.stdin.readline()
1018                                         try:
1019                                                 ticket_id = int(line)
1020                                         except:
1021                                                 print "could not get ticket_id from stdin..."
1022                                                 os._exit(1)
1023                                 else:
1024                                         ticket_id = act_record['ticket_id']
1025
1026                                 email_args['ticket_id'] = ticket_id
1027
1028                 return email_args
1029
1030         def get_unique_issues(self, act_recordlist):
1031                 # NOTE: only send one email per site, per problem...
1032                 unique_issues = {}
1033                 for act_record in act_recordlist:
1034                         act_key = act_record['action'][0]
1035                         if act_key not in unique_issues:
1036                                 unique_issues[act_key] = []
1037
1038                         unique_issues[act_key] += [act_record]
1039
1040                 return unique_issues
1041
1042
1043         def __actOnSite(self, loginbase, site_record):
1044                 i_nodes_actedon = 0
1045                 i_nodes_emailed = 0
1046
1047                 act_recordlist = []
1048
1049                 for nodename in site_record['nodes'].keys():
1050                         diag_record = site_record['nodes'][nodename]
1051                         act_record  = self.__actOnNode(diag_record)
1052                         #print "nodename: %s %s" % (nodename, act_record)
1053                         if act_record is not None:
1054                                 act_recordlist += [act_record]
1055
1056                 unique_issues = self.get_unique_issues(act_recordlist)
1057
1058                 for issue in unique_issues.keys():
1059                         print "\tworking on issue: %s" % issue
1060                         issue_record_list = unique_issues[issue]
1061                         email_args = self.get_email_args(issue_record_list, loginbase)
1062
1063                         # for each record.
1064                         #for act_record in issue_record_list:
1065                         #       # if there's a pcu record and email config is set
1066                         #       if 'email_pcu' in act_record:
1067                         #               if act_record['message'] != None and act_record['email_pcu'] and site_record['config']['email']:
1068                         #                       # and 'reboot_node' in act_record['stage']:
1069
1070                         #                       email_args['hostname'] = act_record['nodename']
1071                         #                       ticket_id = self.__emailSite(loginbase,
1072                         #                                                               act_record['email'],
1073                         #                                                               emailTxt.mailtxt.pcudown[0],
1074                         #                                                               email_args)
1075                         #                       if ticket_id == 0:
1076                         #                               # error.
1077                         #                               print "got a ticket_id == 0!!!! %s" % act_record['nodename']
1078                         #                               os._exit(1)
1079                         #                               pass
1080                         #                       email_args['ticket_id'] = ticket_id
1081
1082
1083                         act_record = issue_record_list[0]
1084                         # send message before squeezing
1085                         print "\t\tconfig.email: %s and %s" % (act_record['message'] != None,
1086                                                                                                 site_record['config']['email'])
1087                         if act_record['message'] != None and site_record['config']['email']:
1088                                 ticket_id = self.__emailSite(loginbase, act_record['email'],
1089                                                                                          act_record['message'], email_args)
1090
1091                                 if ticket_id == 0:
1092                                         # error.
1093                                         print "ticket_id == 0 for %s %s" % (loginbase, act_record['nodename'])
1094                                         import os
1095                                         os._exit(1)
1096                                         pass
1097
1098                                 # Add ticket_id to ALL nodenames
1099                                 for act_record in issue_record_list:
1100                                         nodename = act_record['nodename']
1101                                         # update node record with RT ticket_id
1102                                         if nodename in self.act_all:
1103                                                 self.act_all[nodename][0]['ticket_id'] = "%s" % ticket_id
1104                                                 # if the ticket was previously resolved, reset it to new.
1105                                                 if 'rt' in act_record and \
1106                                                         'Status' in act_record['rt'] and \
1107                                                         act_record['rt']['Status'] == 'resolved':
1108                                                         mailer.setTicketStatus(ticket_id, "new")
1109                                                 status = mailer.getTicketStatus(ticket_id)
1110                                                 self.act_all[nodename][0]['rt'] = status
1111                                         if config.mail: i_nodes_emailed += 1
1112
1113                         print "\t\tconfig.squeeze: %s and %s" % (config.squeeze,
1114                                                                                                         site_record['config']['squeeze'])
1115                         if config.squeeze and site_record['config']['squeeze']:
1116                                 for act_key in act_record['action']:
1117                                         self.actions[act_key](email_args)
1118                                 i_nodes_actedon += 1
1119
1120                 if config.policysavedb:
1121                         #print "Saving Databases... act_all, diagnose_out"
1122                         #database.dbDump("act_all", self.act_all)
1123                         # remove site record from diagnose_out, it's in act_all as done.
1124                         del self.diagnose_db[loginbase]
1125                         #database.dbDump("diagnose_out", self.diagnose_db)
1126
1127                 print "sleeping for 1 sec"
1128                 time.sleep(1)
1129                 #print "Hit enter to continue..."
1130                 #sys.stdout.flush()
1131                 #line = sys.stdin.readline()
1132
1133                 return (i_nodes_actedon, i_nodes_emailed)
1134
1135         def __actOnNode(self, diag_record):
1136                 nodename = diag_record['nodename']
1137                 message = diag_record['message']
1138
1139                 act_record = {}
1140                 act_record.update(diag_record)
1141                 act_record['nodename'] = nodename
1142                 act_record['msg_format'] = self._format_diaginfo(diag_record)
1143                 print "act_record['stage'] == %s " % act_record['stage']
1144
1145                 # avoid end records, and nmreset records
1146                 # reboot_node_failed, is set below, so don't reboot repeatedly.
1147
1148                 #if 'monitor-end-record' not in act_record['stage'] and \
1149                 #   'nmreset' not in act_record['stage'] and \
1150                 #   'reboot_node_failed' not in act_record:
1151
1152                 #       if "DOWN" in act_record['log'] and \
1153                 #                       'pcu_ids' in act_record['plcnode'] and \
1154                 #                       len(act_record['plcnode']['pcu_ids']) > 0:
1155 #
1156 #                               print "%s" % act_record['log'],
1157 #                               print "%15s" % (['reboot_node'],)
1158 #                               # Set node to re-install
1159 #                               plc.nodeBootState(act_record['nodename'], "rins")
1160 #                               try:
1161 #                                       ret = reboot_node({'hostname': act_record['nodename']})
1162 #                               except Exception, exc:
1163 #                                       print "exception on reboot_node:"
1164 #                                       import traceback
1165 #                                       print traceback.print_exc()
1166 #                                       ret = False
1167 #
1168 #                               if ret: # and ( 'reboot_node_failed' not in act_record or act_record['reboot_node_failed'] == False):
1169 #                                       # Reboot Succeeded
1170 #                                       print "reboot succeeded for %s" % act_record['nodename']
1171 #                                       act_record2 = {}
1172 #                                       act_record2.update(act_record)
1173 #                                       act_record2['action'] = ['reboot_node']
1174 #                                       act_record2['stage'] = "reboot_node"
1175 #                                       act_record2['reboot_node_failed'] = False
1176 #                                       act_record2['email_pcu'] = False
1177 #
1178 #                                       if nodename not in self.act_all:
1179 #                                               self.act_all[nodename] = []
1180 #                                       print "inserting 'reboot_node' record into act_all"
1181 #                                       self.act_all[nodename].insert(0,act_record2)
1182 #
1183 #                                       # return None to avoid further action
1184 #                                       print "Taking no further action"
1185 #                                       return None
1186 #                               else:
1187 #                                       print "reboot failed for %s" % act_record['nodename']
1188 #                                       # set email_pcu to also send pcu notice for this record.
1189 #                                       act_record['reboot_node_failed'] = True
1190 #                                       act_record['email_pcu'] = True
1191 #
1192 #                       print "%s" % act_record['log'],
1193 #                       print "%15s" % act_record['action']
1194
1195                 if act_record['stage'] is not 'monitor-end-record' and \
1196                    act_record['stage'] is not 'nmreset':
1197                         if nodename not in self.act_all:
1198                                 self.act_all[nodename] = []
1199
1200                         self.act_all[nodename].insert(0,act_record)
1201                 else:
1202                         print "Not recording %s in act_all" % nodename
1203
1204                 return act_record
1205
1206         def analyseSites(self):
1207                 i_sites_observed = 0
1208                 i_sites_diagnosed = 0
1209                 i_nodes_diagnosed = 0
1210                 i_nodes_actedon = 0
1211                 i_sites_emailed = 0
1212                 l_allsites = []
1213
1214                 sorted_sites = self.sickdb.keys()
1215                 sorted_sites.sort()
1216                 for loginbase in sorted_sites:
1217                         site_record = self.sickdb[loginbase]
1218                         print "sites: %s" % loginbase
1219
1220                         i_nodes_diagnosed += len(site_record.keys())
1221                         i_sites_diagnosed += 1
1222
1223                         (na,ne) = self.__actOnSite(loginbase, site_record)
1224
1225                         i_sites_observed += 1
1226                         i_nodes_actedon += na
1227                         i_sites_emailed += ne
1228
1229                         l_allsites += [loginbase]
1230
1231                 return {'sites_observed': i_sites_observed,
1232                                 'sites_diagnosed': i_sites_diagnosed,
1233                                 'nodes_diagnosed': i_nodes_diagnosed,
1234                                 'sites_emailed': i_sites_emailed,
1235                                 'nodes_actedon': i_nodes_actedon,
1236                                 'allsites':l_allsites}
1237
1238         def print_stats(self, key, stats):
1239                 print "%20s : %d" % (key, stats[key])
1240