planetstack/openstack/observer.py

   1 import time
   2 import traceback
   3 import commands
   4 import threading
   5
   6 from datetime import datetime
   7 from collections import defaultdict
   8 from core.models import *
   9 from django.db.models import F, Q
  10 from openstack.manager import OpenStackManager
  11 from util.logger import Logger, logging, logger
  12 #from timeout import timeout
  13
  14
  15 logger = Logger(logfile='observer.log', level=logging.INFO)
  16
  17 class PlanetStackObserver:
  18
  19     def __init__(self):
  20         self.manager = OpenStackManager()
  21         # The Condition object that gets signalled by Feefie events
  22         self.event_cond = threading.Condition()
  23
  24     def wait_for_event(self, timeout):
  25         self.event_cond.acquire()
  26         self.event_cond.wait(timeout)
  27         self.event_cond.release()
  28
  29     def wake_up(self):
  30         logger.info('Wake up routine called. Event cond %r'%self.event_cond)
  31         self.event_cond.acquire()
  32         self.event_cond.notify()
  33         self.event_cond.release()
  34
  35     def run(self):
  36         if not self.manager.enabled or not self.manager.has_openstack:
  37             return
  38         while True:
  39             try:
  40                 start_time=time.time()
  41                 logger.info('Observer run loop')
  42                 #self.sync_roles()
  43
  44                 logger.info('Calling sync tenants')
  45                 try:
  46                     self.sync_tenants()
  47                 except:
  48                     logger.log_exc("Exception in sync_tenants")
  49                     traceback.print_exc()
  50                 finish_time = time.time()
  51                 logger.info('Sync tenants took %f seconds'%(finish_time-start_time))
  52
  53                 logger.info('Calling sync users')
  54                 try:
  55                     self.sync_users()
  56                 except:
  57                     logger.log_exc("Exception in sync_users")
  58                     traceback.print_exc()
  59                 finish_time = time.time()
  60                 logger.info('Sync users took %f seconds'%(finish_time-start_time))
  61
  62                 logger.info('Calling sync tenant roles')
  63                 try:
  64                     self.sync_user_tenant_roles()
  65                 except:
  66                     logger.log_exc("Exception in sync_users")
  67                     traceback.print_exc()
  68
  69                 logger.info('Calling sync slivers')
  70                 try:
  71                     self.sync_slivers()
  72                 except:
  73                     logger.log_exc("Exception in sync slivers")
  74                     traceback.print_exc()
  75                 finish_time = time.time()
  76                 logger.info('Sync slivers took %f seconds'%(finish_time-start_time))
  77
  78                 logger.info('Calling sync sliver ips')
  79                 try:
  80                     self.sync_sliver_ips()
  81                 except:
  82                     logger.log_exc("Exception in sync_sliver_ips")
  83                     traceback.print_exc()
  84                 finish_time = time.time()
  85                 logger.info('Sync sliver ips took %f seconds'%(finish_time-start_time))
  86
  87                 logger.info('Calling sync networks')
  88                 try:
  89                     self.sync_networks()
  90                 except:
  91                     logger.log_exc("Exception in sync_networks")
  92                     traceback.print_exc()
  93                 finish_time = time.time()
  94                 logger.info('Sync networks took %f seconds'%(finish_time-start_time))
  95
  96                 logger.info('Calling sync network slivers')
  97                 try:
  98                     self.sync_network_slivers()
  99                 except:
 100                     logger.log_exc("Exception in sync_network_slivers")
 101                     traceback.print_exc()
 102                 finish_time = time.time()
 103                 logger.info('Sync network sliver ips took %f seconds'%(finish_time-start_time))
 104
 105                 logger.info('Calling sync external routes')
 106                 try:
 107                     self.sync_external_routes()
 108                 except:
 109                      logger.log_exc("Exception in sync_external_routes")
 110                      traceback.print_exc()
 111                 finish_time = time.time()
 112                 logger.info('Sync external routes took %f seconds'%(finish_time-start_time))
 113
 114                 logger.info('Waiting for event')
 115                 tBeforeWait = time.time()
 116                 self.wait_for_event(timeout=300)
 117
 118                 # Enforce 5 minutes between wakeups
 119                 tSleep = 300 - (time.time() - tBeforeWait)
 120                 if tSleep > 0:
 121                     logger.info('Sleeping for %d seconds' % tSleep)
 122                     time.sleep(tSleep)
 123
 124                 logger.info('Observer woken up')
 125             except:
 126                 logger.log_exc("Exception in observer run loop")
 127                 traceback.print_exc()
 128
 129     def sync_roles(self):
 130         """
 131         save all role that don't already exist in keystone. Remove keystone roles that
 132         don't exist in planetstack
 133         """
 134         # sync all roles that don't already in keystone
 135         keystone_roles = self.manager.driver.shell.keystone.roles.findall()
 136         keystone_role_names = [kr.name for kr in keystone_roles]
 137         pending_roles = Role.objects.all()
 138         pending_role_names = [r.role_type for r in pending_roles]
 139         for role in pending_roles:
 140             if role.role_type not in keystone_role_names:
 141                 try:
 142                     self.manager.save_role(role)
 143                     logger.info("save role: %s" % (role))
 144                 except:
 145                     logger.log_exc("save role failed: %s" % role)
 146                     traceback.print_exc()
 147
 148         # don't delete roles for now
 149         """
 150         # delete keystone roles that don't exist in planetstack
 151         for keystone_role in keystone_roles:
 152             if keystone_role.name == 'admin':
 153                 continue
 154             if keystone_role.name not in pending_role_names:
 155                 try:
 156                     self.manager.driver.delete_role({id: keystone_role.id})
 157                 except:
 158                     traceback.print_exc()
 159         """
 160
 161     def sync_tenants(self):
 162         """
 163         Save all sites and sliceswhere enacted < updated or enacted == None.
 164         Remove sites and slices that no don't exist in openstack db if they
 165         have an enacted time (enacted != None).
 166         """
 167         # get all sites that need to be synced (enacted < updated or enacted is None)
 168         pending_sites = Site.objects.filter(Q(enacted__lt=F('updated')) | Q(enacted=None))
 169         for site in pending_sites:
 170             try:
 171                 self.manager.save_site(site)
 172                 logger.info("saved site %s" % site)
 173             except:
 174                 logger.log_exc("save site failed: %s" % site)
 175
 176         # get all slices that need to be synced (enacted < updated or enacted is None)
 177         pending_slices = Slice.objects.filter(Q(enacted__lt=F('updated')) | Q(enacted=None))
 178         for slice in pending_slices:
 179             try:
 180                 self.manager.init_caller(slice.creator, slice.creator.site.login_base)
 181                 self.manager.save_slice(slice)
 182                 logger.info("saved slice %s" % slice)
 183             except:
 184                 logger.log_exc("save slice failed: %s" % slice)
 185
 186         # get all sites that where enacted != null. We can assume these sites
 187         # have previously been synced and need to be checed for deletion.
 188         sites = Site.objects.filter(enacted__isnull=False)
 189         site_dict = {}
 190         for site in sites:
 191             site_dict[site.login_base] = site
 192
 193         # get all slices that where enacted != null. We can assume these slices
 194         # have previously been synced and need to be checed for deletion.
 195         slices = Slice.objects.filter(enacted__isnull=False)
 196         slice_dict = {}
 197         for slice in slices:
 198             slice_dict[slice.name] = slice
 199
 200         # delete keystone tenants that don't have a site record
 201         tenants = self.manager.driver.shell.keystone.tenants.findall()
 202         system_tenants = ['admin','service']
 203         for tenant in tenants:
 204             if tenant.name in system_tenants:
 205                 continue
 206             if tenant.name not in site_dict and tenant.name not in slice_dict:
 207                 try:
 208                     self.manager.driver.delete_tenant(tenant.id)
 209                     logger.info("deleted tenant: %s" % (tenant))
 210                 except:
 211                     logger.log_exc("delete tenant failed: %s" % tenant)
 212
 213
 214     def sync_users(self):
 215         """
 216         save all users where enacted < updated or enacted == None. Remove users that
 217         no don't exist in openstack db if they have an enacted time (enacted != None).
 218         """
 219         # get all users that need to be synced (enacted < updated or enacted is None)
 220         pending_users = User.objects.filter(Q(enacted__lt=F('updated')) | Q(enacted=None))
 221         for user in pending_users:
 222             try:
 223                 self.manager.save_user(user)
 224                 logger.info("saved user: %s" % (user))
 225             except:
 226                 logger.log_exc("save user failed: %s" %user)
 227
 228         # get all users that where enacted != null. We can assume these users
 229         # have previously been synced and need to be checed for deletion.
 230         users = User.objects.filter(enacted__isnull=False)
 231         user_dict = {}
 232         for user in users:
 233             user_dict[user.kuser_id] = user
 234
 235         # delete keystone users that don't have a user record
 236         system_users = ['admin', 'nova', 'quantum', 'glance', 'cinder', 'swift', 'service']
 237         users = self.manager.driver.shell.keystone.users.findall()
 238         for user in users:
 239             if user.name in system_users:
 240                 continue
 241             if user.id not in user_dict:
 242                 try:
 243                     #self.manager.driver.delete_user(user.id)
 244                     logger.info("deleted user: %s" % user)
 245                 except:
 246                     logger.log_exc("delete user failed: %s" % user)
 247
 248
 249     def sync_user_tenant_roles(self):
 250         """
 251         Save all site privileges and slice memberships wheree enacted < updated or
 252         enacted == None. Remove ones that don't exist in openstack db if they have
 253         an enacted time (enacted != None).
 254         """
 255         # sync site privileges
 256         pending_site_privileges = SitePrivilege.objects.filter(Q(enacted__lt=F('updated')) | Q(enacted=None))
 257         for site_priv in pending_site_privileges:
 258             try:
 259                 self.manager.save_site_privilege(site_priv)
 260                 logger.info("saved site privilege: %s" % (site_priv))
 261             except: logger.log_exc("save site privilege failed: %s " % site_priv)
 262
 263         # sync slice memberships
 264         pending_slice_memberships = SliceMembership.objects.filter(Q(enacted__lt=F('updated')) | Q(enacted=None))
 265         for slice_memb in pending_slice_memberships:
 266             try:
 267                 self.manager.save_slice_membership(slice_memb)
 268                 logger.info("saved slice membership: %s" % (slice_memb))
 269             except: logger.log_exc("save slice membership failed: %s" % slice_memb)
 270
 271         # get all site privileges and slice memberships that have been enacted
 272         user_tenant_roles = defaultdict(list)
 273         for site_priv in SitePrivilege.objects.filter(enacted__isnull=False):
 274             user_tenant_roles[(site_priv.user.kuser_id, site_priv.site.tenant_id)].append(site_priv.role.role)
 275         for slice_memb in SliceMembership.objects.filter(enacted__isnull=False):
 276             user_tenant_roles[(slice_memb.user.kuser_id, slice_memb.slice.tenant_id)].append(slice_memb.role.role)
 277
 278         # Some user tenant role aren't stored in planetstack but they must be preserved.
 279         # Role that fall in this category are
 280         # 1. Never remove a user's role that their home site
 281         # 2. Never remove a user's role at a slice they've created.
 282         # Keep track of all roles that must be preserved.
 283         users = User.objects.all()
 284         preserved_roles = {}
 285         for user in users:
 286             tenant_ids = [s['tenant_id'] for s in user.slices.values()]
 287             tenant_ids.append(user.site.tenant_id)
 288             preserved_roles[user.kuser_id] = tenant_ids
 289
 290
 291         # begin removing user tenant roles from keystone. This is stored in the
 292         # Metadata table.
 293         for metadata in self.manager.driver.shell.keystone_db.get_metadata():
 294             # skip admin roles
 295             if metadata.user_id == self.manager.driver.admin_user.id:
 296                 continue
 297             # skip preserved tenant ids
 298             if metadata.user_id in preserved_roles and \
 299                metadata.tenant_id in preserved_roles[metadata.user_id]:
 300                 continue
 301             # get roles for user at this tenant
 302             user_tenant_role_ids = user_tenant_roles.get((metadata.user_id, metadata.tenant_id), [])
 303
 304             if user_tenant_role_ids:
 305                 # The user has roles at the tenant. Check if roles need to
 306                 # be updated.
 307                 user_keystone_role_ids = metadata.data.get('roles', [])
 308                 for role_id in user_keystone_role_ids:
 309                     if role_id not in user_tenant_role_ids:
 310                         user_keystone_role_ids.pop(user_keystone_role_ids.index(role_id))
 311             else:
 312                 # The user has no roles at this tenant.
 313                 metadata.data['roles'] = []
 314             #session.add(metadata)
 315             logger.info("pruning metadata for %s at %s" % (metadata.user_id, metadata.tenant_id))
 316
 317     def sync_slivers(self):
 318         """
 319         save all slivers where enacted < updated or enacted == None. Remove slivers that
 320         no don't exist in openstack db if they have an enacted time (enacted != None).
 321         """
 322         # get all users that need to be synced (enacted < updated or enacted is None)
 323         pending_slivers = Sliver.objects.filter(Q(enacted__lt=F('updated')) | Q(enacted=None))
 324         for sliver in pending_slivers:
 325             if sliver.creator:
 326                 try:
 327                     # update manager context
 328                     self.manager.init_caller(sliver.creator, sliver.slice.name)
 329                     self.manager.save_sliver(sliver)
 330                     logger.info("saved sliver: %s" % (sliver))
 331                 except:
 332                     logger.log_exc("save sliver failed: %s" % sliver)
 333
 334         # get all slivers where enacted != null. We can assume these users
 335         # have previously been synced and need to be checed for deletion.
 336         slivers = Sliver.objects.filter(enacted__isnull=False)
 337         sliver_dict = {}
 338         for sliver in slivers:
 339             sliver_dict[sliver.instance_id] = sliver
 340
 341         # delete sliver that don't have a sliver record
 342         ctx = self.manager.driver.shell.nova_db.ctx
 343         instances = self.manager.driver.shell.nova_db.instance_get_all(ctx)
 344         for instance in instances:
 345             if instance.uuid not in sliver_dict:
 346                 try:
 347                     # lookup tenant and update context
 348                     try:
 349                         tenant = self.manager.driver.shell.keystone.tenants.find(id=instance.project_id)
 350                         tenant_name = tenant.name
 351                     except:
 352                         tenant_name = None
 353                         logger.info("exception while retrieving tenant %s. Deleting instance using root tenant." % instance.project_id)
 354                     self.manager.init_admin(tenant=tenant_name)
 355                     self.manager.driver.destroy_instance(instance.uuid)
 356                     logger.info("destroyed sliver: %s" % (instance.uuid))
 357                 except:
 358                     logger.log_exc("destroy sliver failed: %s" % instance)
 359
 360
 361     def sync_sliver_ips(self):
 362         # fill in null ip addresses
 363         slivers = Sliver.objects.filter(ip=None)
 364         for sliver in slivers:
 365             # update connection
 366             self.manager.init_admin(tenant=sliver.slice.name)
 367             servers = self.manager.driver.shell.nova.servers.findall(id=sliver.instance_id)
 368             if not servers:
 369                 continue
 370             server = servers[0]
 371             ips = server.addresses.get(sliver.slice.name, [])
 372             if not ips:
 373                 continue
 374             sliver.ip = ips[0]['addr']
 375             sliver.save()
 376             logger.info("saved sliver ip: %s %s" % (sliver, ips[0]))
 377
 378     def sync_external_routes(self):
 379         routes = self.manager.driver.get_external_routes()
 380         subnets = self.manager.driver.shell.quantum.list_subnets()['subnets']
 381         for subnet in subnets:
 382             try:
 383                 self.manager.driver.add_external_route(subnet, routes)
 384             except:
 385                 logger.log_exc("failed to add external route for subnet %s" % subnet)
 386
 387     def sync_network_slivers(self):
 388         networkSlivers = NetworkSliver.objects.all()
 389         networkSlivers_by_id = {}
 390         networkSlivers_by_port = {}
 391         for networkSliver in networkSlivers:
 392             networkSlivers_by_id[networkSliver.id] = networkSliver
 393             networkSlivers_by_port[networkSliver.port_id] = networkSliver
 394
 395         networks = Network.objects.all()
 396         networks_by_id = {}
 397         for network in networks:
 398             networks_by_id[network.network_id] = network
 399
 400         slivers = Sliver.objects.all()
 401         slivers_by_instance_id = {}
 402         for sliver in slivers:
 403             slivers_by_instance_id[sliver.instance_id] = sliver
 404
 405         ports = self.manager.driver.shell.quantum.list_ports()["ports"]
 406         for port in ports:
 407             if port["id"] in networkSlivers_by_port:
 408                 # we already have it
 409                 print "already accounted for port", port["id"]
 410                 continue
 411
 412             if port["device_owner"] != "compute:nova":
 413                 # we only want the ports that connect to instances
 414                 continue
 415
 416             network = networks_by_id.get(port['network_id'], None)
 417             if not network:
 418                 #print "no network for port", port["id"], "network", port["network_id"]
 419                 continue
 420
 421             sliver = slivers_by_instance_id.get(port['device_id'], None)
 422             if not sliver:
 423                 print "no sliver for port", port["id"], "device_id", port['device_id']
 424                 continue
 425
 426             if network.template.sharedNetworkId is not None:
 427                 # If it's a shared network template, then more than one network
 428                 # object maps to the quantum network. We have to do a whole bunch
 429                 # of extra work to find the right one.
 430                 networks = network.template.network_set.all()
 431                 network = None
 432                 for candidate_network in networks:
 433                     if (candidate_network.owner == sliver.slice):
 434                         print "found network", candidate_network
 435                         network = candidate_network
 436
 437                 if not network:
 438                     print "failed to find the correct network for a shared template for port", port["id"], "network", port["network_id"]
 439                     continue
 440
 441             if not port["fixed_ips"]:
 442                 print "port", port["id"], "has no fixed_ips"
 443                 continue
 444
 445 #            print "XXX", port
 446
 447             ns = NetworkSliver(network=network,
 448                                sliver=sliver,
 449                                ip=port["fixed_ips"][0]["ip_address"],
 450                                port_id=port["id"])
 451             ns.save()
 452
 453     def sync_networks(self):
 454         """
 455         save all networks where enacted < updated or enacted == None. Remove networks that
 456         no don't exist in openstack db if they have an enacted time (enacted != None).
 457         """
 458         # get all users that need to be synced (enacted < updated or enacted is None)
 459         pending_networks = Network.objects.filter(Q(enacted__lt=F('updated')) | Q(enacted=None))
 460         for network in pending_networks:
 461             if network.owner and network.owner.creator:
 462                 try:
 463                     # update manager context
 464                     self.manager.init_caller(network.owner.creator, network.owner.name)
 465                     self.manager.save_network(network)
 466                     logger.info("saved network: %s" % (network))
 467                 except:
 468                     logger.log_exc("save network failed: %s" % network)
 469
 470         # get all networks where enacted != null. We can assume these users
 471         # have previously been synced and need to be checed for deletion.
 472         networks = Network.objects.filter(enacted__isnull=False)
 473         network_dict = {}
 474         for network in networks:
 475             network_dict[network.network_id] = network
 476
 477         # TODO: delete Network objects if quantum network doesn't exist
 478         #       (need to write self.manager.driver.shell.quantum_db)
 479