sfa/importer/plimporter.py

   1 #
   2 # PlanetLab importer
   3 #
   4 # requirements
   5 #
   6 # read the planetlab database and update the local registry database accordingly
   7 # (in other words, with this testbed, the SFA registry is *not* authoritative)
   8 # so we update the following collections
   9 # . authorities                 (from pl sites)
  10 # . node                        (from pl nodes)
  11 # . users+keys                  (from pl persons and attached keys)
  12 #                       known limitation : *one* of the ssh keys is chosen at random here
  13 #                       xxx todo/check xxx at the very least, when a key is known to the registry
  14 #                       and is still current in plc
  15 #                       then we should definitely make sure to keep that one in sfa...
  16 # . slice+researchers           (from pl slices and attached users)
  17 #
  18
  19 import os
  20
  21 from sfa.util.config import Config
  22 from sfa.util.xrn import Xrn, get_leaf, get_authority, hrn_to_urn
  23
  24 from sfa.trust.gid import create_uuid
  25 from sfa.trust.certificate import convert_public_key, Keypair
  26
  27 # using global alchemy.session() here is fine
  28 # as importer is on standalone one-shot process
  29 from sfa.storage.alchemy import global_dbsession
  30 from sfa.storage.model import RegRecord, RegAuthority, RegSlice, RegNode, RegUser, RegKey
  31
  32 from sfa.planetlab.plshell import PlShell
  33 from sfa.planetlab.plxrn import hostname_to_hrn, slicename_to_hrn, email_to_hrn, hrn_to_pl_slicename
  34
  35
  36 def _get_site_hrn(interface_hrn, site):
  37     # Hardcode 'internet2' into the hrn for sites hosting
  38     # internet2 nodes. This is a special operation for some vini
  39     # sites only
  40     hrn = ".".join([interface_hrn, site['login_base']])
  41     if ".vini" in interface_hrn and interface_hrn.endswith('vini'):
  42         if site['login_base'].startswith("i2") or site['login_base'].startswith("nlr"):
  43             hrn = ".".join([interface_hrn, "internet2", site['login_base']])
  44     return hrn
  45
  46
  47 class PlImporter:
  48
  49     def __init__(self, auth_hierarchy, logger):
  50         self.auth_hierarchy = auth_hierarchy
  51         self.logger = logger
  52
  53     def add_options(self, parser):
  54         # we don't have any options for now
  55         pass
  56
  57     # hrn hash is initialized from current db
  58     # remember just-created records as we go
  59     # xxx might make sense to add a UNIQUE constraint in the db itself
  60     def remember_record_by_hrn(self, record):
  61         tuple = (record.type, record.hrn)
  62         if tuple in self.records_by_type_hrn:
  63             self.logger.warning(
  64                 "PlImporter.remember_record_by_hrn: duplicate {}".format(tuple))
  65             return
  66         self.records_by_type_hrn[tuple] = record
  67
  68     # ditto for pointer hash
  69     def remember_record_by_pointer(self, record):
  70         if record.pointer == -1:
  71             self.logger.warning(
  72                 "PlImporter.remember_record_by_pointer: pointer is void")
  73             return
  74         tuple = (record.type, record.pointer)
  75         if tuple in self.records_by_type_pointer:
  76             self.logger.warning(
  77                 "PlImporter.remember_record_by_pointer: duplicate {}".format(tuple))
  78             return
  79         self.records_by_type_pointer[(record.type, record.pointer,)] = record
  80
  81     def remember_record(self, record):
  82         self.remember_record_by_hrn(record)
  83         self.remember_record_by_pointer(record)
  84
  85     def locate_by_type_hrn(self, type, hrn):
  86         return self.records_by_type_hrn.get((type, hrn), None)
  87
  88     def locate_by_type_pointer(self, type, pointer):
  89         return self.records_by_type_pointer.get((type, pointer), None)
  90
  91     # a convenience/helper function to see if a record is already known
  92     # a former, broken, attempt (in 2.1-9) had been made
  93     # to try and use 'pointer' as a first, most significant attempt
  94     # the idea being to preserve stuff as much as possible, and thus
  95     # to avoid creating a new gid in the case of a simple hrn rename
  96     # however this of course doesn't work as the gid depends on the hrn...
  97     # def locate (self, type, hrn=None, pointer=-1):
  98     #    if pointer!=-1:
  99     #        attempt = self.locate_by_type_pointer (type, pointer)
 100     #        if attempt : return attempt
 101     #    if hrn is not None:
 102     #        attempt = self.locate_by_type_hrn (type, hrn,)
 103     #        if attempt : return attempt
 104     #    return None
 105
 106     # this makes the run method a bit abtruse - out of the way
 107     def create_special_vini_record(self, interface_hrn):
 108         # special case for vini
 109         if ".vini" in interface_hrn and interface_hrn.endswith('vini'):
 110             # create a fake internet2 site first
 111             i2site = {'name': 'Internet2',
 112                       'login_base': 'internet2', 'site_id': -1}
 113             site_hrn = _get_site_hrn(interface_hrn, i2site)
 114             # import if hrn is not in list of existing hrns or if the hrn exists
 115             # but its not a site record
 116             if ('authority', site_hrn, ) not in self.records_by_type_hrn:
 117                 urn = hrn_to_urn(site_hrn, 'authority')
 118                 if not self.auth_hierarchy.auth_exists(urn):
 119                     self.auth_hierarchy.create_auth(urn)
 120                 auth_info = self.auth_hierarchy.get_auth_info(urn)
 121                 auth_record = RegAuthority(hrn=site_hrn, gid=auth_info.get_gid_object(),
 122                                            pointer=site['site_id'],
 123                                            authority=get_authority(site_hrn))
 124                 auth_record.just_created()
 125                 global_dbsession.add(auth_record)
 126                 global_dbsession.commit()
 127                 self.logger.info(
 128                     "PlImporter: Imported authority (vini site) {}".format(auth_record))
 129                 self.remember_record(site_record)
 130
 131     def run(self, options):
 132         config = Config()
 133         interface_hrn = config.SFA_INTERFACE_HRN
 134         root_auth = config.SFA_REGISTRY_ROOT_AUTH
 135         shell = PlShell(config)
 136
 137         # retrieve all existing SFA objects
 138         all_records = global_dbsession.query(RegRecord).all()
 139
 140         # create hash by (type,hrn)
 141         # we essentially use this to know if a given record is already known to
 142         # SFA
 143         self.records_by_type_hrn = \
 144             dict([((record.type, record.hrn), record)
 145                   for record in all_records])
 146         # create hash by (type,pointer)
 147         self.records_by_type_pointer = \
 148             dict([((record.type, record.pointer), record) for record in all_records
 149                   if record.pointer != -1])
 150
 151         # initialize record.stale to True by default, then mark stale=False on
 152         # the ones that are in use
 153         for record in all_records:
 154             record.stale = True
 155
 156         # retrieve PLC data
 157         # Get all plc sites
 158         # retrieve only required stuf
 159         sites = shell.GetSites({'peer_id': None, 'enabled': True},
 160                                ['site_id', 'login_base', 'node_ids', 'slice_ids', 'person_ids', 'name', 'hrn'])
 161         # create a hash of sites by login_base
 162 #        sites_by_login_base = dict ( [ ( site['login_base'], site ) for site in sites ] )
 163         # Get all plc users
 164         persons = shell.GetPersons({'peer_id': None, 'enabled': True},
 165                                    ['person_id', 'email', 'key_ids', 'site_ids', 'role_ids', 'hrn'])
 166         # create a hash of persons by person_id
 167         persons_by_id = dict([(person['person_id'], person)
 168                               for person in persons])
 169         # also gather non-enabled user accounts so as to issue relevant
 170         # warnings
 171         disabled_persons = shell.GetPersons(
 172             {'peer_id': None, 'enabled': False}, ['person_id'])
 173         disabled_person_ids = [person['person_id']
 174                                for person in disabled_persons]
 175         # Get all plc public keys
 176         # accumulate key ids for keys retrieval
 177         key_ids = []
 178         for person in persons:
 179             key_ids.extend(person['key_ids'])
 180         keys = shell.GetKeys({'peer_id': None, 'key_id': key_ids,
 181                               'key_type': 'ssh'})
 182         # create a hash of keys by key_id
 183         keys_by_id = dict([(key['key_id'], key) for key in keys])
 184         # create a dict person_id -> [ (plc)keys ]
 185         keys_by_person_id = {}
 186         for person in persons:
 187             pubkeys = []
 188             for key_id in person['key_ids']:
 189                 # by construction all the keys we fetched are ssh keys
 190                 # so gpg keys won't be in there
 191                 try:
 192                     key = keys_by_id[key_id]
 193                     pubkeys.append(key)
 194                 except:
 195                     self.logger.warning(
 196                         "Could not spot key {} - probably non-ssh".format(key_id))
 197             keys_by_person_id[person['person_id']] = pubkeys
 198         # Get all plc nodes
 199         nodes = shell.GetNodes({'peer_id': None}, [
 200                                'node_id', 'hostname', 'site_id'])
 201         # create hash by node_id
 202         nodes_by_id = dict([(node['node_id'], node, ) for node in nodes])
 203         # Get all plc slices
 204         slices = shell.GetSlices(
 205             {'peer_id': None}, ['slice_id', 'name', 'person_ids', 'hrn'])
 206         # create hash by slice_id
 207         slices_by_id = dict([(slice['slice_id'], slice) for slice in slices])
 208
 209         # isolate special vini case in separate method
 210         self.create_special_vini_record(interface_hrn)
 211
 212         # Get top authority record
 213         top_auth_record = self.locate_by_type_hrn('authority', root_auth)
 214         admins = []
 215
 216         # start importing
 217         for site in sites:
 218             try:
 219                 site_sfa_created = shell.GetSiteSfaCreated(site['site_id'])
 220             except:
 221                 site_sfa_created = None
 222             if site['name'].startswith('sfa:') or site_sfa_created == 'True':
 223                 continue
 224
 225             #site_hrn = _get_site_hrn(interface_hrn, site)
 226             site_hrn = site['hrn']
 227             # import if hrn is not in list of existing hrns or if the hrn exists
 228             # but its not a site record
 229             site_record = self.locate_by_type_hrn('authority', site_hrn)
 230             if not site_record:
 231                 try:
 232                     urn = hrn_to_urn(site_hrn, 'authority')
 233                     if not self.auth_hierarchy.auth_exists(urn):
 234                         self.auth_hierarchy.create_auth(urn)
 235                     auth_info = self.auth_hierarchy.get_auth_info(urn)
 236                     site_record = RegAuthority(hrn=site_hrn, gid=auth_info.get_gid_object(),
 237                                                pointer=site['site_id'],
 238                                                authority=get_authority(
 239                                                    site_hrn),
 240                                                name=site['name'])
 241                     site_record.just_created()
 242                     global_dbsession.add(site_record)
 243                     global_dbsession.commit()
 244                     self.logger.info(
 245                         "PlImporter: imported authority (site) : {}".format(site_record))
 246                     self.remember_record(site_record)
 247                 except:
 248                     # if the site import fails then there is no point in trying to import the
 249                     # site's child records (node, slices, persons), so skip
 250                     # them.
 251                     self.logger.log_exc("PlImporter: failed to import site {}. Skipping child records"
 252                                         .format(site_hrn))
 253                     continue
 254             else:
 255                 # xxx update the record ...
 256                 site_record.name = site['name']
 257                 pass
 258             site_record.stale = False
 259
 260             # import node records
 261             for node_id in site['node_ids']:
 262                 try:
 263                     node = nodes_by_id[node_id]
 264                 except:
 265                     self.logger.warning("PlImporter: cannot find node_id {} - ignored"
 266                                         .format(node_id))
 267                     continue
 268                 site_auth = get_authority(site_hrn)
 269                 site_name = site['login_base']
 270                 node_hrn = hostname_to_hrn(
 271                     site_auth, site_name, node['hostname'])
 272                 # xxx this sounds suspicious
 273                 if len(node_hrn) > 64:
 274                     node_hrn = node_hrn[:64]
 275                 node_record = self.locate_by_type_hrn('node', node_hrn)
 276                 if not node_record:
 277                     try:
 278                         pkey = Keypair(create=True)
 279                         urn = hrn_to_urn(node_hrn, 'node')
 280                         node_gid = self.auth_hierarchy.create_gid(
 281                             urn, create_uuid(), pkey)
 282                         node_record = RegNode(hrn=node_hrn, gid=node_gid,
 283                                               pointer=node['node_id'],
 284                                               authority=get_authority(node_hrn))
 285                         node_record.just_created()
 286                         global_dbsession.add(node_record)
 287                         global_dbsession.commit()
 288                         self.logger.info(
 289                             "PlImporter: imported node: {}".format(node_record))
 290                         self.remember_record(node_record)
 291                     except:
 292                         self.logger.log_exc(
 293                             "PlImporter: failed to import node {}".format(node_hrn))
 294                         continue
 295                 else:
 296                     # xxx update the record ...
 297                     pass
 298                 node_record.stale = False
 299
 300             site_pis = []
 301             # import persons
 302             for person_id in site['person_ids']:
 303                 proceed = False
 304                 if person_id in persons_by_id:
 305                     person = persons_by_id[person_id]
 306                     proceed = True
 307                 elif person_id in disabled_person_ids:
 308                     pass
 309                 else:
 310                     self.logger.warning("PlImporter: cannot locate person_id {} in site {} - ignored"
 311                                         .format(person_id, site_hrn))
 312                 # make sure to NOT run this if anything is wrong
 313                 if not proceed:
 314                     continue
 315
 316                 #person_hrn = email_to_hrn(site_hrn, person['email'])
 317                 person_hrn = person['hrn']
 318                 if person_hrn is None:
 319                     self.logger.warn(
 320                         "Person {} has no hrn - skipped".format(person['email']))
 321                     continue
 322                 # xxx suspicious again
 323                 if len(person_hrn) > 64:
 324                     person_hrn = person_hrn[:64]
 325                 person_urn = hrn_to_urn(person_hrn, 'user')
 326
 327                 user_record = self.locate_by_type_hrn('user', person_hrn)
 328
 329                 # return a tuple pubkey (a plc key object) and pkey (a Keypair
 330                 # object)
 331                 def init_person_key(person, plc_keys):
 332                     pubkey = None
 333                     if person['key_ids']:
 334                         # randomly pick first key in set
 335                         pubkey = plc_keys[0]
 336                         try:
 337                             pkey = convert_public_key(pubkey['key'])
 338                         except:
 339                             self.logger.warn('PlImporter: unable to convert public key for {}'
 340                                              .format(person_hrn))
 341                             pkey = Keypair(create=True)
 342                     else:
 343                         # the user has no keys. Creating a random keypair for
 344                         # the user's gid
 345                         self.logger.warn("PlImporter: person {} does not have a PL public key"
 346                                          .format(person_hrn))
 347                         pkey = Keypair(create=True)
 348                     return (pubkey, pkey)
 349
 350                 # new person
 351                 try:
 352                     plc_keys = keys_by_person_id.get(person['person_id'], [])
 353                     if not user_record:
 354                         (pubkey, pkey) = init_person_key(person, plc_keys)
 355                         person_gid = self.auth_hierarchy.create_gid(person_urn, create_uuid(), pkey,
 356                                                                     email=person['email'])
 357                         user_record = RegUser(hrn=person_hrn, gid=person_gid,
 358                                               pointer=person['person_id'],
 359                                               authority=get_authority(
 360                                                   person_hrn),
 361                                               email=person['email'])
 362                         if pubkey:
 363                             user_record.reg_keys = [
 364                                 RegKey(pubkey['key'], pubkey['key_id'])]
 365                         else:
 366                             self.logger.warning(
 367                                 "No key found for user {}".format(user_record))
 368                         user_record.just_created()
 369                         global_dbsession.add(user_record)
 370                         global_dbsession.commit()
 371                         self.logger.info(
 372                             "PlImporter: imported person: {}".format(user_record))
 373                         self.remember_record(user_record)
 374                     else:
 375                         # update the record ?
 376                         #
 377                         # if a user key has changed then we need to update the
 378                         # users gid by forcing an update here
 379                         #
 380                         # right now, SFA only has *one* key attached to a user, and this is
 381                         # the key that the GID was made with
 382                         # so the logic here is, we consider that things are OK (unchanged) if
 383                         # all the SFA keys are present as PLC keys
 384                         # otherwise we trigger the creation of a new gid from *some* plc key
 385                         # and record this on the SFA side
 386                         # it would make sense to add a feature in PLC so that one could pick a 'primary'
 387                         # key but this is not available on the myplc side for now
 388                         # = or = it would be much better to support several keys in SFA but that
 389                         # does not seem doable without a major overhaul in the data model as
 390                         # a GID is attached to a hrn, but it's also linked to a key, so...
 391                         # NOTE: with this logic, the first key entered in PLC remains the one
 392                         # current in SFA until it is removed from PLC
 393                         sfa_keys = user_record.reg_keys
 394
 395                         def sfa_key_in_list(sfa_key, plc_keys):
 396                             for plc_key in plc_keys:
 397                                 if plc_key['key'] == sfa_key.key:
 398                                     return True
 399                             return False
 400                         # are all the SFA keys known to PLC ?
 401                         new_keys = False
 402                         if not sfa_keys and plc_keys:
 403                             new_keys = True
 404                         else:
 405                             for sfa_key in sfa_keys:
 406                                 if not sfa_key_in_list(sfa_key, plc_keys):
 407                                     new_keys = True
 408                         if new_keys:
 409                             (pubkey, pkey) = init_person_key(person, plc_keys)
 410                             person_gid = self.auth_hierarchy.create_gid(
 411                                 person_urn, create_uuid(), pkey)
 412                             person_gid.set_email(person['email'])
 413                             if not pubkey:
 414                                 user_record.reg_keys = []
 415                             else:
 416                                 user_record.reg_keys = [
 417                                     RegKey(pubkey['key'], pubkey['key_id'])]
 418                             user_record.gid = person_gid
 419                             user_record.just_updated()
 420                             self.logger.info(
 421                                 "PlImporter: updated person: {}".format(user_record))
 422                     user_record.email = person['email']
 423                     global_dbsession.commit()
 424                     user_record.stale = False
 425                     # accumulate PIs - PLCAPI has a limitation that when someone has PI role
 426                     # this is valid for all sites she is in..
 427                     # PI is coded with role_id == 20
 428                     if 20 in person['role_ids']:
 429                         site_pis.append(user_record)
 430
 431                     # PL Admins need to marked as PI of the top authority
 432                     # record
 433                     if 10 in person['role_ids'] and user_record not in top_auth_record.reg_pis:
 434                         admins.append(user_record)
 435
 436                 except:
 437                     self.logger.log_exc("PlImporter: failed to import person {} {}"
 438                                         .format(person['person_id'], person['email']))
 439
 440             # maintain the list of PIs for a given site
 441             # for the record, Jordan had proposed the following addition as a welcome hotfix to a previous version:
 442             # site_pis = list(set(site_pis))
 443             # this was likely due to a bug in the above logic, that had to do with disabled persons
 444             # being improperly handled, and where the whole loop on persons
 445             # could be performed twice with the same person...
 446             # so hopefully we do not need to eliminate duplicates explicitly
 447             # here anymore
 448             site_record.reg_pis = list(set(site_pis))
 449             global_dbsession.commit()
 450
 451             # import slices
 452             for slice_id in site['slice_ids']:
 453                 try:
 454                     slice = slices_by_id[slice_id]
 455                 except:
 456                     self.logger.warning("PlImporter: cannot locate slice_id {} - ignored"
 457                                         .format(slice_id))
 458                     continue
 459                 #slice_hrn = slicename_to_hrn(interface_hrn, slice['name'])
 460                 slice_hrn = slice['hrn']
 461                 if slice_hrn is None:
 462                     self.logger.warning("Slice {} has no hrn - skipped"
 463                                         .format(slice['name']))
 464                     continue
 465                 slice_record = self.locate_by_type_hrn('slice', slice_hrn)
 466                 if not slice_record:
 467                     try:
 468                         pkey = Keypair(create=True)
 469                         urn = hrn_to_urn(slice_hrn, 'slice')
 470                         slice_gid = self.auth_hierarchy.create_gid(
 471                             urn, create_uuid(), pkey)
 472                         slice_record = RegSlice(hrn=slice_hrn, gid=slice_gid,
 473                                                 pointer=slice['slice_id'],
 474                                                 authority=get_authority(slice_hrn))
 475                         slice_record.just_created()
 476                         global_dbsession.add(slice_record)
 477                         global_dbsession.commit()
 478                         self.logger.info(
 479                             "PlImporter: imported slice: {}".format(slice_record))
 480                         self.remember_record(slice_record)
 481                     except:
 482                         self.logger.log_exc("PlImporter: failed to import slice {} ({})"
 483                                             .format(slice_hrn, slice['name']))
 484                 else:
 485                     # xxx update the record ...
 486                     # given that we record the current set of users anyways, there does not seem to be much left to do here
 487                     # self.logger.warning ("Slice update not yet implemented on slice {} ({})"
 488                     #                      .format(slice_hrn, slice['name']))
 489                     pass
 490                 # record current users affiliated with the slice
 491                 slice_record.reg_researchers = \
 492                     [self.locate_by_type_pointer('user', user_id) for user_id in slice[
 493                         'person_ids']]
 494                 # remove any weird value (looks like we can get 'None' here
 495                 slice_record.reg_researchers = [
 496                     x for x in slice_record.reg_researchers if x]
 497                 global_dbsession.commit()
 498                 slice_record.stale = False
 499
 500         # Set PL Admins as PI's of the top authority
 501         if admins:
 502             top_auth_record.reg_pis = list(set(admins))
 503             global_dbsession.commit()
 504             self.logger.info('PlImporter: set PL admins {} as PIs of {}'
 505                              .format(admins, top_auth_record.hrn))
 506
 507         # remove stale records
 508         # special records must be preserved
 509         system_hrns = [interface_hrn, root_auth,
 510                        interface_hrn + '.slicemanager']
 511         for record in all_records:
 512             if record.hrn in system_hrns:
 513                 record.stale = False
 514             if record.peer_authority:
 515                 record.stale = False
 516             if ".vini" in interface_hrn and interface_hrn.endswith('vini') and \
 517                     record.hrn.endswith("internet2"):
 518                 record.stale = False
 519
 520         for record in all_records:
 521             try:
 522                 stale = record.stale
 523             except:
 524                 stale = True
 525                 self.logger.warning("stale not found with {}".format(record))
 526             if stale:
 527                 self.logger.info(
 528                     "PlImporter: deleting stale record: {}".format(record))
 529                 global_dbsession.delete(record)
 530                 global_dbsession.commit()