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 '  "boot_server":"'`cat /mnt/cdrom/bootme/BOOTSERVER`'",'
 241                                         echo '  "nm_status":"'`ps ax | grep nm.py | grep -v grep`'",'
 242                                         echo '  "dns_status":"'`host boot.planet-lab.org 2>&1`'",'
 243                                         echo '  "iptables_status":"'`iptables -t mangle -nL | awk '$1~/^[A-Z]+$/ {modules[$1]=1;}END{for (k in modules) {if (k) printf "%s ",k;}}'`'",'
 244                                         echo '  "princeton_comon_dir":"'`ls -d /vservers/princeton_comon`'",'
 245                                         echo '  "uptime":"'`cat /proc/uptime`'",'
 246
 247                                         ID=`grep princeton_comon /etc/passwd | awk -F : '{if ( $3 > 500 ) { print $3}}'`
 248                                         echo '  "princeton_comon_running":"'`ls -d /proc/virtual/$ID`'",'
 249                                         echo '  "princeton_comon_procs":"'`vps ax | grep $ID | grep -v grep | wc -l`'",'
 250                                         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`'",'
 251                                         echo '  "rpm_version":"'`if [ -x /usr/bin/timeout.pl ] ; then timeout.pl 30 rpm -q NodeManager ; fi`'",'
 252                                         echo '  "rpm_versions":"'`if [ -x /usr/bin/timeout.pl ] ; then timeout.pl 45 rpm -q -a ; fi`'",'
 253                                         echo "}"
 254 EOF                     """)
 255
 256                                 values['ssh_error'] = errval
 257                                 if len(oval) > 0:
 258                                         #print "OVAL: %s" % oval
 259                                         values.update(eval(oval))
 260                                         values['ssh_portused'] = port
 261                                         break
 262                                 else:
 263                                         values.update({'kernel_version': "", 'bmlog' : "", 'bootcd_version' : '',
 264                                                                         'boot_server' : '',
 265                                                                         'nm_status' : '',
 266                                                                         'fs_status' : '',
 267                                                                         'uptime' : '',
 268                                                                         'dns_status' : '',
 269                                                                         'rpm_version' : '',
 270                                                                         'rpm_versions' : '',
 271                                                                         'princeton_comon_dir' : "",
 272                                                                         'princeton_comon_running' : "",
 273                                                                         'princeton_comon_procs' : "", 'ssh_portused' : None})
 274
 275                         oval = values['nm_status']
 276                         if "nm.py" in oval:
 277                                 values['nm_status'] = "Y"
 278                         else:
 279                                 values['nm_status'] = "N"
 280
 281                         continue_slice_check = True
 282                         oval = values['princeton_comon_dir']
 283                         if "princeton_comon" in oval:
 284                                 values['princeton_comon_dir'] = True
 285                         else:
 286                                 values['princeton_comon_dir'] = False
 287                                 continue_slice_check = False
 288
 289                         if continue_slice_check:
 290                                 oval = values['princeton_comon_running']
 291                                 if len(oval) > len('/proc/virtual/'):
 292                                         values['princeton_comon_running'] = True
 293                                 else:
 294                                         values['princeton_comon_running'] = False
 295                                         continue_slice_check = False
 296                         else:
 297                                 values['princeton_comon_running'] = False
 298
 299                         if continue_slice_check:
 300                                 oval = values['princeton_comon_procs']
 301                                 values['princeton_comon_procs'] = int(oval)
 302                         else:
 303                                 values['princeton_comon_procs'] = None
 304                 except:
 305                         print traceback.print_exc()
 306                         sys.exit(1)
 307
 308                 return values
 309
 310         def collectPLC(self, nodename, cohash):
 311                 values = {}
 312                 ### GET PLC NODE ######################
 313                 d_node = plccache.GetNodeByName(nodename)
 314                 values['plc_node_stats'] = d_node
 315
 316                 ### GET PLC PCU ######################
 317                 site_id = -1
 318                 d_pcu = None
 319                 if d_node and len(d_node['pcu_ids']) > 0:
 320                         d_pcu = d_node['pcu_ids'][0]
 321
 322                 site_id = d_node['site_id']
 323
 324                 values['plc_pcuid'] = d_pcu
 325
 326                 ### GET PLC SITE ######################
 327                 print "SITEID: %s" % site_id
 328                 d_site = plccache.GetSitesById([ site_id ])[0]
 329                 values['loginbase'] = d_site['login_base']
 330                 values['plc_site_stats'] = d_site
 331
 332                 return values
 333
 334         def evaluate(self, nodename, values):
 335                 # TODO: this section can probably be reduced to a policy statement
 336                 #               using patterns and values collected so far.
 337                 # NOTE: A node is "DOWN" if
 338                 #       * cannot ssh into it.
 339                 #   * all ports are not open for a 'BOOT' node
 340                 #   * dns for hostname does not exist.
 341                 b_getbootcd_id = True
 342
 343                 oval = values['kernel_version']
 344                 values['ssh_status'] = True
 345                 if "2.6.17" in oval or "2.6.2" in oval:
 346                         values['observed_category'] = 'PROD'
 347                         if "bm.log" in values['bmlog']:
 348                                 values['observed_status'] = 'DEBUG'
 349                         else:
 350                                 values['observed_status'] = 'BOOT'
 351                 elif "2.6.12" in oval or "2.6.10" in oval:
 352                         values['observed_category'] = 'OLDPROD'
 353                         if "bm.log" in values['bmlog']:
 354                                 values['observed_status'] = 'DEBUG'
 355                         else:
 356                                 values['observed_status'] = 'BOOT'
 357
 358                 # NOTE: on 2.6.8 kernels, with 4.2 bootstrapfs, the chroot
 359                 #       command fails.  I have no idea why.
 360                 elif "2.4" in oval or "2.6.8" in oval:
 361                         b_getbootcd_id = False
 362                         values['observed_category'] = 'OLDBOOTCD'
 363                         values['observed_status'] = 'DEBUG'
 364                 elif oval != "":
 365                         values['observed_category'] = 'UNKNOWN'
 366                         if "bm.log" in values['bmlog']:
 367                                 values['observed_status'] = 'DEBUG'
 368                         else:
 369                                 values['observed_status'] = 'BOOT'
 370                 else:
 371                         # An error occurred.
 372                         b_getbootcd_id = False
 373                         values['ssh_status'] = False
 374                         values['observed_category'] = 'ERROR'
 375                         values['observed_status'] = 'DOWN'
 376                         values['kernel_version'] = ""
 377
 378                 values['firewall'] = False
 379
 380                 # NOTE: A node is down if some of the public ports are not open
 381                 if values['observed_status'] == "BOOT":
 382                         # verify that all ports are open.  Else, report node as down.
 383                         if not ( values['port_status']['22']  == "open" and \
 384                                          values['port_status']['80']  == "open" and \
 385                                          values['port_status']['806'] == "open") :
 386                                 #email_exception(nodename, "%s FILTERED HOST" % nodename)
 387                                 values['observed_status'] = 'DOWN'
 388                                 values['firewall'] = True
 389
 390                         #if   values['port_status']['22']  == "open" and \
 391                         #        values['port_status']['80']  == "closed" and \
 392                         #        values['port_status']['806'] == "open" :
 393                         #       email_exception("%s port 80 blocked" % nodename, "possible VSERVER ref blocked")
 394
 395                 #if not values['external_dns_status']:
 396                 #       email_exception("%s DNS down" % nodename)
 397
 398                 if b_getbootcd_id:
 399                         # try to get BootCD for all nodes that are not 2.4 nor inaccessible
 400                         oval = values['bootcd_version']
 401                         if "BootCD" in oval:
 402                                 values['bootcd_version'] = oval
 403                                 if "v2" in oval and \
 404                                         ( nodename is not "planetlab1.cs.unc.edu" and \
 405                                           nodename is not "planetlab2.cs.unc.edu" ):
 406                                         values['observed_category'] = 'OLDBOOTCD'
 407                         else:
 408                                 values['bootcd_version'] = ""
 409                 else:
 410                         values['bootcd_version'] = ""
 411
 412                 return values
 413
 414         def collectDNS(self, nodename, cohash):
 415                 values = {}
 416                 try:
 417                         ipaddr = socket.gethostbyname(nodename)
 418                         # TODO: check that IP returned matches IP in plc db.
 419                         values['external_dns_status'] = True
 420                 except Exception, err:
 421                         values['external_dns_status'] = False
 422
 423                 return values
 424
 425         def collectInternal(self, nodename, cohash):
 426                 try:
 427                         values = {}
 428
 429                         v = self.collectPING(nodename, cohash)
 430                         values.update(v)
 431
 432                         v = self.collectPorts(nodename)
 433                         values.update(v)
 434
 435                         v = self.collectSSH(nodename, cohash)
 436                         values.update(v)
 437
 438                         v = self.collectDNS(nodename, cohash)
 439                         values.update(v)
 440
 441                         v = self.collectTRACEROUTE(nodename, cohash)
 442                         values.update(v)
 443
 444                         v = self.collectPLC(nodename, cohash)
 445                         values.update(v)
 446
 447                         if nodename in cohash:
 448                                 values['comon_stats'] = cohash[nodename]
 449                         else:
 450                                 values['comon_stats'] = {'resptime':  '-1',
 451                                                                                 'uptime':    '-1',
 452                                                                                 'sshstatus': '-1',
 453                                                                                 'lastcotop': '-1',
 454                                                                                 'cpuspeed' : "null",
 455                                                                                 'disksize' : 'null',
 456                                                                                 'memsize'  : 'null'}
 457
 458                         values['rpms'] = values['rpm_versions']
 459                         print "ALLVERSIONS: %s %s" % (nodename, values['rpm_versions'])
 460                         print "RPMVERSION: %s %s" % (nodename, values['rpm_version'])
 461                         print "UPTIME: %s %s" % (nodename, values['uptime'])
 462
 463                         values = self.evaluate(nodename, values)
 464                         values['date_checked'] = datetime.now()
 465
 466                 except:
 467                         print traceback.print_exc()
 468
 469                 return (nodename, values)
 470
 471
 472 def internalprobe(hostname):
 473         scannode = ScanNodeInternal()
 474         try:
 475                 (nodename, values) = scannode.collectInternal(hostname, {})
 476                 scannode.record(None, (nodename, values))
 477                 session.flush()
 478                 return True
 479         except:
 480                 print traceback.print_exc()
 481                 return False
 482
 483 def externalprobe(hostname):
 484         scannode = ScanNodeInternal()
 485         try:
 486                 values = self.collectPorts(hostname)
 487                 scannode.record(None, (hostname, values))
 488                 session.flush()
 489                 return True
 490         except:
 491                 print traceback.print_exc()
 492                 return False
 493
 494 class ScanPCU(ScanInterface):
 495         recordclass = FindbadPCURecord
 496         syncclass = None
 497         primarykey = 'plc_pcuid'
 498
 499         def collectInternal(self, pcuname, cohash):
 500
 501                 continue_probe = True
 502                 errors = None
 503                 values = {'reboot_trial_status' : 'novalue'}
 504                 ### GET PCU ######################
 505                 try:
 506                         b_except = False
 507                         try:
 508                                 v = get_plc_pcu_values(pcuname)
 509                                 if v['hostname'] is not None: v['hostname'] = v['hostname'].strip()
 510                                 if v['ip'] is not None: v['ip'] = v['ip'].strip()
 511
 512                                 if v is not None:
 513                                         values['plc_pcu_stats'] = v
 514                                 else:
 515                                         continue_probe = False
 516                         except:
 517                                 b_except = True
 518                                 traceback.print_exc()
 519                                 continue_probe = False
 520
 521                         if b_except or not continue_probe: return (None, None, None)
 522
 523                         #### RUN NMAP ###############################
 524                         if continue_probe:
 525                                 nmap = command.CMD()
 526                                 print "nmap -oG - -P0 -p22,23,80,443,5869,9100,16992 %s | grep Host:" % reboot.pcu_name(values['plc_pcu_stats'])
 527                                 (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']))
 528                                 # NOTE: an empty / error value for oval, will still work.
 529                                 (values['port_status'], continue_probe) = nmap_port_status(oval)
 530                         else:
 531                                 values['port_status'] = None
 532
 533                         #### COMPLETE ENTRY   #######################
 534
 535                         values['entry_complete'] = []
 536                         #if values['protocol'] is None or values['protocol'] is "":
 537                         #       values['entry_complete'] += ["protocol"]
 538                         if values['plc_pcu_stats']['model'] is None or values['plc_pcu_stats']['model'] is "":
 539                                 values['entry_complete'] += ["model"]
 540                                 # Cannot continue due to this condition
 541                                 continue_probe = False
 542
 543                         if values['plc_pcu_stats']['password'] is None or values['plc_pcu_stats']['password'] is "":
 544                                 values['entry_complete'] += ["password"]
 545                                 # Cannot continue due to this condition
 546                                 continue_probe = False
 547
 548                         if len(values['entry_complete']) > 0:
 549                                 continue_probe = False
 550
 551                         if values['plc_pcu_stats']['hostname'] is None or values['plc_pcu_stats']['hostname'] is "":
 552                                 values['entry_complete'] += ["hostname"]
 553                         if values['plc_pcu_stats']['ip'] is None or values['plc_pcu_stats']['ip'] is "":
 554                                 values['entry_complete'] += ["ip"]
 555
 556                         # If there are no nodes associated with this PCU, then we cannot continue.
 557                         if len(values['plc_pcu_stats']['node_ids']) == 0:
 558                                 continue_probe = False
 559                                 values['entry_complete'] += ['nodeids']
 560
 561
 562                         #### DNS and IP MATCH #######################
 563                         if values['plc_pcu_stats']['hostname'] is not None and values['plc_pcu_stats']['hostname'] is not "" and \
 564                            values['plc_pcu_stats']['ip'] is not None and values['plc_pcu_stats']['ip'] is not "":
 565                                 try:
 566                                         ipaddr = socket.gethostbyname(values['plc_pcu_stats']['hostname'])
 567                                         if ipaddr == values['plc_pcu_stats']['ip']:
 568                                                 values['dns_status'] = "DNS-OK"
 569                                         else:
 570                                                 values['dns_status'] = "DNS-MISMATCH"
 571                                                 values['plc_pcu_stats']['hostname'] = values['plc_pcu_stats']['ip']
 572
 573                                 except Exception, err:
 574                                         values['dns_status'] = "DNS-NOENTRY"
 575                                         values['plc_pcu_stats']['hostname'] = values['plc_pcu_stats']['ip']
 576                         else:
 577                                 if values['plc_pcu_stats']['ip'] is not None and values['plc_pcu_stats']['ip'] is not "":
 578                                         values['dns_status'] = "NOHOSTNAME"
 579                                         values['plc_pcu_stats']['hostname'] = values['plc_pcu_stats']['ip']
 580                                 else:
 581                                         values['dns_status'] = "NO-DNS-OR-IP"
 582                                         values['plc_pcu_stats']['hostname'] = "No_entry_in_DB"
 583                                         continue_probe = False
 584
 585
 586                         ######  DRY RUN  ############################
 587                         if continue_probe and 'node_ids' in values['plc_pcu_stats'] and \
 588                                 len(values['plc_pcu_stats']['node_ids']) > 0:
 589                                 rb_ret = reboot.reboot_test_new(values['plc_pcu_stats']['nodenames'][0],
 590                                                                                                 values, 1, True)
 591                         else:
 592                                 rb_ret = "Not_Run" # No nodes to test"
 593
 594                         values['reboot_trial_status'] = rb_ret
 595
 596                 except:
 597                         print "____________________________________"
 598                         print values
 599                         errors = values
 600                         print "____________________________________"
 601                         errors['traceback'] = traceback.format_exc()
 602                         print errors['traceback']
 603                         values['reboot_trial_status'] = str(errors['traceback'])
 604                         print values
 605
 606                 values['entry_complete']=" ".join(values['entry_complete'])
 607
 608                 values['date_checked'] = datetime.now()
 609                 return (pcuname, values)
 610