monitor_policy.py

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