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