monitor/scanapi.py

   1 #!/usr/bin/python
   2
   3 import os
   4 import sys
   5 import string
   6 import time
   7 from datetime import datetime,timedelta
   8 import threadpool
   9 import threading
  10
  11 import socket
  12 from pcucontrol import reboot
  13
  14 from pcucontrol.util import command
  15 from monitor import config
  16
  17 from monitor.database.info.model import *
  18
  19 from monitor.sources import comon
  20 from monitor.wrapper import plc, plccache
  21
  22 import traceback
  23 from monitor.common import nmap_port_status, email_exception
  24
  25 COMON_COTOPURL= "http://summer.cs.princeton.edu/status/tabulator.cgi?" + \
  26                         "table=table_nodeview&" + \
  27                         "dumpcols='name,resptime,sshstatus,uptime,lastcotop,cpuspeed,memsize,disksize'&" + \
  28                         "formatcsv"
  29
  30 api = plc.getAuthAPI()
  31 plc_lock = threading.Lock()
  32 round = 1
  33 global_round = round
  34 count = 0
  35
  36
  37 def get_pcu(pcuname):
  38         plc_lock.acquire()
  39         try:
  40                 #print "GetPCU from PLC %s" % pcuname
  41                 l_pcu  = plc.GetPCUs({'pcu_id' : pcuname})
  42                 #print l_pcu
  43                 if len(l_pcu) > 0:
  44                         l_pcu = l_pcu[0]
  45         except:
  46                 try:
  47                         #print "GetPCU from file %s" % pcuname
  48                         l_pcus = plccache.l_pcus
  49                         for i in l_pcus:
  50                                 if i['pcu_id'] == pcuname:
  51                                         l_pcu = i
  52                 except:
  53                         traceback.print_exc()
  54                         l_pcu = None
  55
  56         plc_lock.release()
  57         return l_pcu
  58
  59 def get_nodes(node_ids):
  60         plc_lock.acquire()
  61         l_node = []
  62         try:
  63                 l_node = plc.getNodes(node_ids, ['hostname', 'last_contact', 'node_id', 'ports'])
  64         except:
  65                 try:
  66                         plc_nodes = plccache.l_nodes
  67                         for n in plc_nodes:
  68                                 if n['node_id'] in node_ids:
  69                                         l_node.append(n)
  70                 except:
  71                         traceback.print_exc()
  72                         l_node = None
  73
  74         plc_lock.release()
  75         if l_node == []:
  76                 l_node = None
  77         return l_node
  78
  79
  80 def get_plc_pcu_values(pcuname):
  81         """
  82                 Try to contact PLC to get the PCU info.
  83                 If that fails, try a backup copy from the last run.
  84                 If that fails, return None
  85         """
  86         values = {}
  87
  88         l_pcu = get_pcu(pcuname)
  89
  90         if l_pcu is not None:
  91                 site_id = l_pcu['site_id']
  92                 node_ids = l_pcu['node_ids']
  93                 l_node = get_nodes(node_ids)
  94
  95                 if l_node is not None:
  96                         for node in l_node:
  97                                 values[node['hostname']] = node['ports'][0]
  98
  99                         values['nodenames'] = [node['hostname'] for node in l_node]
 100
 101                         # NOTE: this is for a dry run later. It doesn't matter which node.
 102                         values['node_id'] = l_node[0]['node_id']
 103
 104                 values.update(l_pcu)
 105         else:
 106                 values = None
 107
 108         return values
 109
 110 class ScanInterface(object):
 111         recordclass = None
 112         syncclass = None
 113         primarykey = 'hostname'
 114
 115         def __init__(self, round=1):
 116                 self.round = round
 117                 self.count = 1
 118
 119         def __getattr__(self, name):
 120                 if 'collect' in name or 'record' in name:
 121                         method = getattr(self, name, None)
 122                         if method is None:
 123                                 raise Exception("No such method %s" % name)
 124                         return method
 125                 else:
 126                         raise Exception("No such method %s" % name)
 127
 128         def collect(self, nodename, data):
 129                 pass
 130
 131         def record(self, request, (nodename, values) ):
 132
 133                 try:
 134                         if values is None:
 135                                 return
 136
 137                         if self.syncclass:
 138                                 fbnodesync = self.syncclass.findby_or_create(
 139                                                                                                 #if_new_set={'round' : self.round},
 140                                                                                                 **{ self.primarykey : nodename})
 141                         # NOTE: This code will either add a new record for the new self.round,
 142                         #       OR it will find the previous value, and update it with new information.
 143                         #       The data that is 'lost' is not that important, b/c older
 144                         #       history still exists.
 145                         fbrec = self.recordclass.findby_or_create(
 146                                                 **{ self.primarykey:nodename})
 147
 148                         fbrec.set( **values )
 149
 150                         fbrec.flush()
 151                         if self.syncclass:
 152                                 fbnodesync.round = self.round
 153                                 fbnodesync.flush()
 154
 155                         print "%d %s %s" % (self.count, nodename, values)
 156                         self.count += 1
 157
 158                 except:
 159                         print "ERROR:"
 160                         email_exception(str(nodename))
 161                         print traceback.print_exc()
 162                         pass
 163
 164 class ScanNodeInternal(ScanInterface):
 165         recordclass = FindbadNodeRecord
 166         #syncclass = FindbadNodeRecordSync
 167         syncclass = None
 168         primarykey = 'hostname'
 169
 170         def collectPorts(self, nodename, port_list=[22,80,806]):
 171                 values = {}
 172                 for port in port_list:
 173                         ret = os.system("nc -w 5 -z %s %s > /dev/null" % (nodename, port) )
 174                         if ret == 0:
 175                                 values[str(port)] = "open"
 176                         else:
 177                                 values[str(port)] = "closed"
 178                 return {'port_status' : values }
 179
 180         def collectNMAP(self, nodename, cohash):
 181                 #### RUN NMAP ###############################
 182                 # NOTE: run the same command three times and take the best of three
 183                 #               runs.  NMAP can drop packets, and especially so when it runs many
 184                 #               commands at once.
 185                 values = {}
 186                 nmap = command.CMD()
 187                 print "nmap -oG - -P0 -p22,80,806 %s | grep Host:" % nodename
 188                 (oval1,eval) = nmap.run_noexcept("nmap -oG - -P0 -p22,80,806 %s | grep Host:" % nodename)
 189                 (oval2,eval) = nmap.run_noexcept("nmap -oG - -P0 -p22,80,806 %s | grep Host:" % nodename)
 190                 (oval3,eval) = nmap.run_noexcept("nmap -oG - -P0 -p22,80,806 %s | grep Host:" % nodename)
 191                 # NOTE: an empty / error value for oval, will still work.
 192                 values['port_status'] = {}
 193                 (o1,continue_probe) = nmap_port_status(oval1)
 194                 (o2,continue_probe) = nmap_port_status(oval2)
 195                 (o3,continue_probe) = nmap_port_status(oval3)
 196                 for p in ['22', '80', '806']:
 197                         l = [ o1[p], o2[p], o3[p] ]
 198                         if len(filter(lambda x: x == 'open', l)) > 1:
 199                                 values['port_status'][p] = 'open'
 200                         else:
 201                                 values['port_status'][p] = o1[p]
 202
 203                 print values['port_status']
 204                 return (nodename, values)
 205
 206         def collectPING(self, nodename, cohash):
 207                 values = {}
 208                 ping = command.CMD()
 209                 (oval,errval) = ping.run_noexcept("ping -c 1 -q %s | grep rtt" % nodename)
 210
 211                 values = {}
 212                 if oval == "":
 213                         # An error occurred
 214                         values['ping_status'] = False
 215                 else:
 216                         values['ping_status'] = True
 217
 218                 return values
 219
 220         def collectTRACEROUTE(self, nodename, cohash):
 221                 values = {}
 222                 trace = command.CMD()
 223                 (oval,errval) = trace.run_noexcept("traceroute %s" % nodename)
 224
 225                 values['traceroute'] = oval
 226
 227                 return values
 228
 229         def collectSSH(self, nodename, cohash):
 230                 values = {}
 231                 try:
 232                         for port in [22, 806]:
 233                                 ssh = command.SSH('root', nodename, port)
 234
 235                                 (oval, errval) = ssh.run_noexcept2(""" <<\EOF
 236                                         echo "{"
 237                                         echo '  "kernel_version":"'`uname -a`'",'
 238                                         echo '  "bmlog":"'`ls /tmp/bm.log`'",'
 239                                         echo '  "bootcd_version":"'`cat /mnt/cdrom/bootme/ID`'",'
 240                                         echo '  "nm_status":"'`ps ax | grep nm.py | grep -v grep`'",'
 241                                         echo '  "dns_status":"'`host boot.planet-lab.org 2>&1`'",'
 242                                         echo '  "princeton_comon_dir":"'`ls -d /vservers/princeton_comon`'",'
 243                                         echo '  "uptime":"'`cat /proc/uptime`'",'
 244
 245                                         ID=`grep princeton_comon /etc/passwd | awk -F : '{if ( $3 > 500 ) { print $3}}'`
 246                                         echo '  "princeton_comon_running":"'`ls -d /proc/virtual/$ID`'",'
 247                                         echo '  "princeton_comon_procs":"'`vps ax | grep $ID | grep -v grep | wc -l`'",'
 248                                         echo '  "fs_status":"'`grep proc /proc/mounts | grep ro, ; if [ -x /usr/bin/timeout.pl ] ; then timeout.pl 20 touch /var/log/monitor 2>&1 ; if [ -d /vservers/ ] ; then timeout.pl 20 touch /vservers/monitor.log 2>&1  ; fi ; fi`'",'
 249                                         echo '  "rpm_version":"'`if [ -x /usr/bin/timeout.pl ] ; then timeout.pl 30 rpm -q NodeManager ; fi`'",'
 250                                         echo '  "rpm_versions":"'`if [ -x /usr/bin/timeout.pl ] ; then timeout.pl 45 rpm -q -a ; fi`'",'
 251                                         echo "}"
 252 EOF                     """)
 253
 254                                 values['ssh_error'] = errval
 255                                 if len(oval) > 0:
 256                                         #print "OVAL: %s" % oval
 257                                         values.update(eval(oval))
 258                                         values['ssh_portused'] = port
 259                                         break
 260                                 else:
 261                                         values.update({'kernel_version': "", 'bmlog' : "", 'bootcd_version' : '',
 262                                                                         'nm_status' : '',
 263                                                                         'fs_status' : '',
 264                                                                         'uptime' : '',
 265                                                                         'dns_status' : '',
 266                                                                         'rpm_version' : '',
 267                                                                         'rpm_versions' : '',
 268                                                                         'princeton_comon_dir' : "",
 269                                                                         'princeton_comon_running' : "",
 270                                                                         'princeton_comon_procs' : "", 'ssh_portused' : None})
 271
 272                         oval = values['nm_status']
 273                         if "nm.py" in oval:
 274                                 values['nm_status'] = "Y"
 275                         else:
 276                                 values['nm_status'] = "N"
 277
 278                         continue_slice_check = True
 279                         oval = values['princeton_comon_dir']
 280                         if "princeton_comon_dir" in oval:
 281                                 values['princeton_comon_dir'] = True
 282                         else:
 283                                 values['princeton_comon_dir'] = False
 284                                 continue_slice_check = False
 285
 286                         if continue_slice_check:
 287                                 oval = values['princeton_comon_running']
 288                                 if len(oval) > len('/proc/virtual/'):
 289                                         values['princeton_comon_running'] = True
 290                                 else:
 291                                         values['princeton_comon_running'] = False
 292                                         continue_slice_check = False
 293                         else:
 294                                 values['princeton_comon_running'] = False
 295
 296                         if continue_slice_check:
 297                                 oval = values['princeton_comon_procs']
 298                                 values['princeton_comon_procs'] = int(oval)
 299                         else:
 300                                 values['princeton_comon_procs'] = None
 301                 except:
 302                         print traceback.print_exc()
 303                         sys.exit(1)
 304
 305                 return values
 306
 307         def collectPLC(self, nodename, cohash):
 308                 values = {}
 309                 ### GET PLC NODE ######################
 310                 d_node = plccache.GetNodeByName(nodename)
 311                 values['plc_node_stats'] = d_node
 312
 313                 ### GET PLC PCU ######################
 314                 site_id = -1
 315                 d_pcu = None
 316                 if d_node and len(d_node['pcu_ids']) > 0:
 317                         d_pcu = d_node['pcu_ids'][0]
 318
 319                 site_id = d_node['site_id']
 320
 321                 values['plc_pcuid'] = d_pcu
 322
 323                 ### GET PLC SITE ######################
 324                 print "SITEID: %s" % site_id
 325                 d_site = plccache.GetSitesById([ site_id ])[0]
 326                 values['loginbase'] = d_site['login_base']
 327                 values['plc_site_stats'] = d_site
 328
 329                 return values
 330
 331         def evaluate(self, nodename, values):
 332                 # TODO: this section can probably be reduced to a policy statement
 333                 #               using patterns and values collected so far.
 334                 # NOTE: A node is "DOWN" if
 335                 #       * cannot ssh into it.
 336                 #   * all ports are not open for a 'BOOT' node
 337                 #   * dns for hostname does not exist.
 338                 b_getbootcd_id = True
 339
 340                 oval = values['kernel_version']
 341                 values['ssh_status'] = True
 342                 if "2.6.17" in oval or "2.6.2" in oval:
 343                         values['observed_category'] = 'PROD'
 344                         if "bm.log" in values['bmlog']:
 345                                 values['observed_status'] = 'DEBUG'
 346                         else:
 347                                 values['observed_status'] = 'BOOT'
 348                 elif "2.6.12" in oval or "2.6.10" in oval:
 349                         values['observed_category'] = 'OLDPROD'
 350                         if "bm.log" in values['bmlog']:
 351                                 values['observed_status'] = 'DEBUG'
 352                         else:
 353                                 values['observed_status'] = 'BOOT'
 354
 355                 # NOTE: on 2.6.8 kernels, with 4.2 bootstrapfs, the chroot
 356                 #       command fails.  I have no idea why.
 357                 elif "2.4" in oval or "2.6.8" in oval:
 358                         b_getbootcd_id = False
 359                         values['observed_category'] = 'OLDBOOTCD'
 360                         values['observed_status'] = 'DEBUG'
 361                 elif oval != "":
 362                         values['observed_category'] = 'UNKNOWN'
 363                         if "bm.log" in values['bmlog']:
 364                                 values['observed_status'] = 'DEBUG'
 365                         else:
 366                                 values['observed_status'] = 'BOOT'
 367                 else:
 368                         # An error occurred.
 369                         b_getbootcd_id = False
 370                         values['ssh_status'] = False
 371                         values['observed_category'] = 'ERROR'
 372                         values['observed_status'] = 'DOWN'
 373                         values['kernel_version'] = ""
 374
 375                 values['firewall'] = False
 376
 377                 # NOTE: A node is down if some of the public ports are not open
 378                 if values['observed_status'] == "BOOT":
 379                         # verify that all ports are open.  Else, report node as down.
 380                         if not ( values['port_status']['22']  == "open" and \
 381                                          values['port_status']['80']  == "open" and \
 382                                          values['port_status']['806'] == "open") :
 383                                 #email_exception(nodename, "%s FILTERED HOST" % nodename)
 384                                 values['observed_status'] = 'DOWN'
 385                                 values['firewall'] = True
 386
 387                         #if   values['port_status']['22']  == "open" and \
 388                         #        values['port_status']['80']  == "closed" and \
 389                         #        values['port_status']['806'] == "open" :
 390                         #       email_exception("%s port 80 blocked" % nodename, "possible VSERVER ref blocked")
 391
 392                 #if not values['external_dns_status']:
 393                 #       email_exception("%s DNS down" % nodename)
 394
 395                 if b_getbootcd_id:
 396                         # try to get BootCD for all nodes that are not 2.4 nor inaccessible
 397                         oval = values['bootcd_version']
 398                         if "BootCD" in oval:
 399                                 values['bootcd_version'] = oval
 400                                 if "v2" in oval and \
 401                                         ( nodename is not "planetlab1.cs.unc.edu" and \
 402                                           nodename is not "planetlab2.cs.unc.edu" ):
 403                                         values['observed_category'] = 'OLDBOOTCD'
 404                         else:
 405                                 values['bootcd_version'] = ""
 406                 else:
 407                         values['bootcd_version'] = ""
 408
 409                 return values
 410
 411         def collectDNS(self, nodename, cohash):
 412                 values = {}
 413                 try:
 414                         ipaddr = socket.gethostbyname(nodename)
 415                         # TODO: check that IP returned matches IP in plc db.
 416                         values['external_dns_status'] = True
 417                 except Exception, err:
 418                         values['external_dns_status'] = False
 419
 420                 return values
 421
 422         def collectInternal(self, nodename, cohash):
 423                 try:
 424                         values = {}
 425
 426                         v = self.collectPING(nodename, cohash)
 427                         values.update(v)
 428
 429                         v = self.collectPorts(nodename)
 430                         values.update(v)
 431
 432                         v = self.collectSSH(nodename, cohash)
 433                         values.update(v)
 434
 435                         v = self.collectDNS(nodename, cohash)
 436                         values.update(v)
 437
 438                         v = self.collectTRACEROUTE(nodename, cohash)
 439                         values.update(v)
 440
 441                         v = self.collectPLC(nodename, cohash)
 442                         values.update(v)
 443
 444                         if nodename in cohash:
 445                                 values['comon_stats'] = cohash[nodename]
 446                         else:
 447                                 values['comon_stats'] = {'resptime':  '-1',
 448                                                                                 'uptime':    '-1',
 449                                                                                 'sshstatus': '-1',
 450                                                                                 'lastcotop': '-1',
 451                                                                                 'cpuspeed' : "null",
 452                                                                                 'disksize' : 'null',
 453                                                                                 'memsize'  : 'null'}
 454
 455                         values['rpms'] = values['rpm_versions']
 456                         print "ALLVERSIONS: %s %s" % (nodename, values['rpm_versions'])
 457                         print "RPMVERSION: %s %s" % (nodename, values['rpm_version'])
 458                         print "UPTIME: %s %s" % (nodename, values['uptime'])
 459
 460                         values = self.evaluate(nodename, values)
 461                         values['date_checked'] = datetime.now()
 462
 463                 except:
 464                         print traceback.print_exc()
 465
 466                 return (nodename, values)
 467
 468
 469 def internalprobe(hostname):
 470         scannode = ScanNodeInternal()
 471         try:
 472                 (nodename, values) = scannode.collectInternal(hostname, {})
 473                 scannode.record(None, (nodename, values))
 474                 session.flush()
 475                 return True
 476         except:
 477                 print traceback.print_exc()
 478                 return False
 479
 480 def externalprobe(hostname):
 481         scannode = ScanNodeInternal()
 482         try:
 483                 values = self.collectPorts(hostname)
 484                 scannode.record(None, (hostname, values))
 485                 session.flush()
 486                 return True
 487         except:
 488                 print traceback.print_exc()
 489                 return False
 490
 491 class ScanPCU(ScanInterface):
 492         recordclass = FindbadPCURecord
 493         syncclass = None
 494         primarykey = 'plc_pcuid'
 495
 496         def collectInternal(self, pcuname, cohash):
 497
 498                 continue_probe = True
 499                 errors = None
 500                 values = {'reboot_trial_status' : 'novalue'}
 501                 ### GET PCU ######################
 502                 try:
 503                         b_except = False
 504                         try:
 505                                 v = get_plc_pcu_values(pcuname)
 506                                 if v['hostname'] is not None: v['hostname'] = v['hostname'].strip()
 507                                 if v['ip'] is not None: v['ip'] = v['ip'].strip()
 508
 509                                 if v is not None:
 510                                         values['plc_pcu_stats'] = v
 511                                 else:
 512                                         continue_probe = False
 513                         except:
 514                                 b_except = True
 515                                 traceback.print_exc()
 516                                 continue_probe = False
 517
 518                         if b_except or not continue_probe: return (None, None, None)
 519
 520                         #### RUN NMAP ###############################
 521                         if continue_probe:
 522                                 nmap = command.CMD()
 523                                 print "nmap -oG - -P0 -p22,23,80,443,5869,9100,16992 %s | grep Host:" % reboot.pcu_name(values['plc_pcu_stats'])
 524                                 (oval,eval) = nmap.run_noexcept("nmap -oG - -P0 -p22,23,80,443,5869,9100,16992 %s | grep Host:" % reboot.pcu_name(values['plc_pcu_stats']))
 525                                 # NOTE: an empty / error value for oval, will still work.
 526                                 (values['port_status'], continue_probe) = nmap_port_status(oval)
 527                         else:
 528                                 values['port_status'] = None
 529
 530                         #### COMPLETE ENTRY   #######################
 531
 532                         values['entry_complete'] = []
 533                         #if values['protocol'] is None or values['protocol'] is "":
 534                         #       values['entry_complete'] += ["protocol"]
 535                         if values['plc_pcu_stats']['model'] is None or values['plc_pcu_stats']['model'] is "":
 536                                 values['entry_complete'] += ["model"]
 537                                 # Cannot continue due to this condition
 538                                 continue_probe = False
 539
 540                         if values['plc_pcu_stats']['password'] is None or values['plc_pcu_stats']['password'] is "":
 541                                 values['entry_complete'] += ["password"]
 542                                 # Cannot continue due to this condition
 543                                 continue_probe = False
 544
 545                         if len(values['entry_complete']) > 0:
 546                                 continue_probe = False
 547
 548                         if values['plc_pcu_stats']['hostname'] is None or values['plc_pcu_stats']['hostname'] is "":
 549                                 values['entry_complete'] += ["hostname"]
 550                         if values['plc_pcu_stats']['ip'] is None or values['plc_pcu_stats']['ip'] is "":
 551                                 values['entry_complete'] += ["ip"]
 552
 553                         # If there are no nodes associated with this PCU, then we cannot continue.
 554                         if len(values['plc_pcu_stats']['node_ids']) == 0:
 555                                 continue_probe = False
 556                                 values['entry_complete'] += ['nodeids']
 557
 558
 559                         #### DNS and IP MATCH #######################
 560                         if values['plc_pcu_stats']['hostname'] is not None and values['plc_pcu_stats']['hostname'] is not "" and \
 561                            values['plc_pcu_stats']['ip'] is not None and values['plc_pcu_stats']['ip'] is not "":
 562                                 try:
 563                                         ipaddr = socket.gethostbyname(values['plc_pcu_stats']['hostname'])
 564                                         if ipaddr == values['plc_pcu_stats']['ip']:
 565                                                 values['dns_status'] = "DNS-OK"
 566                                         else:
 567                                                 values['dns_status'] = "DNS-MISMATCH"
 568                                                 values['plc_pcu_stats']['hostname'] = values['plc_pcu_stats']['ip']
 569
 570                                 except Exception, err:
 571                                         values['dns_status'] = "DNS-NOENTRY"
 572                                         values['plc_pcu_stats']['hostname'] = values['plc_pcu_stats']['ip']
 573                         else:
 574                                 if values['plc_pcu_stats']['ip'] is not None and values['plc_pcu_stats']['ip'] is not "":
 575                                         values['dns_status'] = "NOHOSTNAME"
 576                                         values['plc_pcu_stats']['hostname'] = values['plc_pcu_stats']['ip']
 577                                 else:
 578                                         values['dns_status'] = "NO-DNS-OR-IP"
 579                                         values['plc_pcu_stats']['hostname'] = "No_entry_in_DB"
 580                                         continue_probe = False
 581
 582
 583                         ######  DRY RUN  ############################
 584                         if continue_probe and 'node_ids' in values['plc_pcu_stats'] and \
 585                                 len(values['plc_pcu_stats']['node_ids']) > 0:
 586                                 rb_ret = reboot.reboot_test_new(values['plc_pcu_stats']['nodenames'][0],
 587                                                                                                 values, 1, True)
 588                         else:
 589                                 rb_ret = "Not_Run" # No nodes to test"
 590
 591                         values['reboot_trial_status'] = rb_ret
 592
 593                 except:
 594                         print "____________________________________"
 595                         print values
 596                         errors = values
 597                         print "____________________________________"
 598                         errors['traceback'] = traceback.format_exc()
 599                         print errors['traceback']
 600                         values['reboot_trial_status'] = str(errors['traceback'])
 601                         print values
 602
 603                 values['entry_complete']=" ".join(values['entry_complete'])
 604
 605                 values['date_checked'] = datetime.now()
 606                 return (pcuname, values)
 607