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 (%s,%s)"%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 (%s,%s)"%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) %s"%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'])
 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'])
 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 %d - probably non-ssh"%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'])
 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         # start importing
 196         for site in sites:
 197             try:
 198                site_sfa_created = shell.GetSiteSfaCreated(site['site_id'])
 199             except:
 200                site_sfa_created = None
 201             if site['name'].startswith('sfa:') or site_sfa_created == 'True':
 202                 continue
 203
 204             site_hrn = _get_site_hrn(interface_hrn, site)
 205             # import if hrn is not in list of existing hrns or if the hrn exists
 206             # but its not a site record
 207             site_record=self.locate_by_type_hrn ('authority', site_hrn)
 208             if not site_record:
 209                 try:
 210                     urn = hrn_to_urn(site_hrn, 'authority')
 211                     if not self.auth_hierarchy.auth_exists(urn):
 212                         self.auth_hierarchy.create_auth(urn)
 213                     auth_info = self.auth_hierarchy.get_auth_info(urn)
 214                     site_record = RegAuthority(hrn=site_hrn, gid=auth_info.get_gid_object(),
 215                                                pointer=site['site_id'],
 216                                                authority=get_authority(site_hrn))
 217                     site_record.just_created()
 218                     global_dbsession.add(site_record)
 219                     global_dbsession.commit()
 220                     self.logger.info("PlImporter: imported authority (site) : %s" % site_record)
 221                     self.remember_record (site_record)
 222                 except:
 223                     # if the site import fails then there is no point in trying to import the
 224                     # site's child records (node, slices, persons), so skip them.
 225                     self.logger.log_exc("PlImporter: failed to import site %s. Skipping child records"%site_hrn)
 226                     continue
 227             else:
 228                 # xxx update the record ...
 229                 pass
 230             site_record.stale=False
 231
 232             # import node records
 233             for node_id in site['node_ids']:
 234                 try:
 235                     node = nodes_by_id[node_id]
 236                 except:
 237                     self.logger.warning ("PlImporter: cannot find node_id %s - ignored"%node_id)
 238                     continue
 239                 site_auth = get_authority(site_hrn)
 240                 site_name = site['login_base']
 241                 node_hrn =  hostname_to_hrn(site_auth, site_name, node['hostname'])
 242                 # xxx this sounds suspicious
 243                 if len(node_hrn) > 64: node_hrn = node_hrn[:64]
 244                 node_record = self.locate_by_type_hrn ( 'node', node_hrn )
 245                 if not node_record:
 246                     try:
 247                         pkey = Keypair(create=True)
 248                         urn = hrn_to_urn(node_hrn, 'node')
 249                         node_gid = self.auth_hierarchy.create_gid(urn, create_uuid(), pkey)
 250                         node_record = RegNode (hrn=node_hrn, gid=node_gid,
 251                                                pointer =node['node_id'],
 252                                                authority=get_authority(node_hrn))
 253                         node_record.just_created()
 254                         global_dbsession.add(node_record)
 255                         global_dbsession.commit()
 256                         self.logger.info("PlImporter: imported node: %s" % node_record)
 257                         self.remember_record (node_record)
 258                     except:
 259                         self.logger.log_exc("PlImporter: failed to import node %s"%node_hrn)
 260                         continue
 261                 else:
 262                     # xxx update the record ...
 263                     pass
 264                 node_record.stale=False
 265
 266             site_pis=[]
 267             # import persons
 268             for person_id in site['person_ids']:
 269                 proceed=False
 270                 if person_id in persons_by_id:
 271                     person=persons_by_id[person_id]
 272                     proceed=True
 273                 elif person_id in disabled_person_ids:
 274                     pass
 275                 else:
 276                     self.logger.warning ("PlImporter: cannot locate person_id %s in site %s - ignored"%(person_id,site_hrn))
 277                 # make sure to NOT run this if anything is wrong
 278                 if not proceed: continue
 279
 280                 person_hrn = email_to_hrn(site_hrn, person['email'])
 281                 # xxx suspicious again
 282                 if len(person_hrn) > 64: person_hrn = person_hrn[:64]
 283                 person_urn = hrn_to_urn(person_hrn, 'user')
 284
 285                 user_record = self.locate_by_type_hrn ( 'user', person_hrn)
 286
 287                 # return a tuple pubkey (a plc key object) and pkey (a Keypair object)
 288                 def init_person_key (person, plc_keys):
 289                     pubkey=None
 290                     if  person['key_ids']:
 291                         # randomly pick first key in set
 292                         pubkey = plc_keys[0]
 293                         try:
 294                             pkey = convert_public_key(pubkey['key'])
 295                         except:
 296                             self.logger.warn('PlImporter: unable to convert public key for %s' % person_hrn)
 297                             pkey = Keypair(create=True)
 298                     else:
 299                         # the user has no keys. Creating a random keypair for the user's gid
 300                         self.logger.warn("PlImporter: person %s does not have a PL public key"%person_hrn)
 301                         pkey = Keypair(create=True)
 302                     return (pubkey, pkey)
 303
 304                 # new person
 305                 try:
 306                     plc_keys = keys_by_person_id.get(person['person_id'],[])
 307                     if not user_record:
 308                         (pubkey,pkey) = init_person_key (person, plc_keys )
 309                         person_gid = self.auth_hierarchy.create_gid(person_urn, create_uuid(), pkey, email=person['email'])
 310                         user_record = RegUser (hrn=person_hrn, gid=person_gid,
 311                                                pointer=person['person_id'],
 312                                                authority=get_authority(person_hrn),
 313                                                email=person['email'])
 314                         if pubkey:
 315                             user_record.reg_keys=[RegKey (pubkey['key'], pubkey['key_id'])]
 316                         else:
 317                             self.logger.warning("No key found for user %s"%user_record)
 318                         user_record.just_created()
 319                         global_dbsession.add (user_record)
 320                         global_dbsession.commit()
 321                         self.logger.info("PlImporter: imported person: %s" % user_record)
 322                         self.remember_record ( user_record )
 323                     else:
 324                         # update the record ?
 325                         #
 326                         # if a user key has changed then we need to update the
 327                         # users gid by forcing an update here
 328                         #
 329                         # right now, SFA only has *one* key attached to a user, and this is
 330                         # the key that the GID was made with
 331                         # so the logic here is, we consider that things are OK (unchanged) if
 332                         # all the SFA keys are present as PLC keys
 333                         # otherwise we trigger the creation of a new gid from *some* plc key
 334                         # and record this on the SFA side
 335                         # it would make sense to add a feature in PLC so that one could pick a 'primary'
 336                         # key but this is not available on the myplc side for now
 337                         # = or = it would be much better to support several keys in SFA but that
 338                         # does not seem doable without a major overhaul in the data model as
 339                         # a GID is attached to a hrn, but it's also linked to a key, so...
 340                         # NOTE: with this logic, the first key entered in PLC remains the one
 341                         # current in SFA until it is removed from PLC
 342                         sfa_keys = user_record.reg_keys
 343                         def sfa_key_in_list (sfa_key,plc_keys):
 344                             for plc_key in plc_keys:
 345                                 if plc_key['key']==sfa_key.key:
 346                                     return True
 347                             return False
 348                         # are all the SFA keys known to PLC ?
 349                         new_keys=False
 350                         if not sfa_keys and plc_keys:
 351                             new_keys=True
 352                         else:
 353                             for sfa_key in sfa_keys:
 354                                  if not sfa_key_in_list (sfa_key,plc_keys):
 355                                      new_keys = True
 356                         if new_keys:
 357                             (pubkey,pkey) = init_person_key (person, plc_keys)
 358                             person_gid = self.auth_hierarchy.create_gid(person_urn, create_uuid(), pkey)
 359                             person_gid.set_email(person['email'])
 360                             if not pubkey:
 361                                 user_record.reg_keys=[]
 362                             else:
 363                                 user_record.reg_keys=[ RegKey (pubkey['key'], pubkey['key_id'])]
 364                             user_record.gid = person_gid
 365                             user_record.just_updated()
 366                             self.logger.info("PlImporter: updated person: %s" % user_record)
 367                     user_record.email = person['email']
 368                     global_dbsession.commit()
 369                     user_record.stale=False
 370                     # accumulate PIs - PLCAPI has a limitation that when someone has PI role
 371                     # this is valid for all sites she is in..
 372                     # PI is coded with role_id==20
 373                     if 20 in person['role_ids']:
 374                         site_pis.append (user_record)
 375                 except:
 376                     self.logger.log_exc("PlImporter: failed to import person %d %s"%(person['person_id'],person['email']))
 377
 378             # maintain the list of PIs for a given site
 379             # for the record, Jordan had proposed the following addition as a welcome hotfix to a previous version:
 380             # site_pis = list(set(site_pis))
 381             # this was likely due to a bug in the above logic, that had to do with disabled persons
 382             # being improperly handled, and where the whole loop on persons
 383             # could be performed twice with the same person...
 384             # so hopefully we do not need to eliminate duplicates explicitly here anymore
 385             site_record.reg_pis = list(set(site_pis))
 386             global_dbsession.commit()
 387
 388             # import slices
 389             for slice_id in site['slice_ids']:
 390                 try:
 391                     slice = slices_by_id[slice_id]
 392                 except:
 393                     self.logger.warning ("PlImporter: cannot locate slice_id %s - ignored"%slice_id)
 394                 slice_hrn = slicename_to_hrn(interface_hrn, slice['name'])
 395                 slice_record = self.locate_by_type_hrn ('slice', slice_hrn)
 396                 if not slice_record:
 397                     try:
 398                         pkey = Keypair(create=True)
 399                         urn = hrn_to_urn(slice_hrn, 'slice')
 400                         slice_gid = self.auth_hierarchy.create_gid(urn, create_uuid(), pkey)
 401                         slice_record = RegSlice (hrn=slice_hrn, gid=slice_gid,
 402                                                  pointer=slice['slice_id'],
 403                                                  authority=get_authority(slice_hrn))
 404                         slice_record.just_created()
 405                         global_dbsession.add(slice_record)
 406                         global_dbsession.commit()
 407                         self.logger.info("PlImporter: imported slice: %s" % slice_record)
 408                         self.remember_record ( slice_record )
 409                     except:
 410                         self.logger.log_exc("PlImporter: failed to import slice %s (%s)"%(slice_hrn,slice['name']))
 411                 else:
 412                     # xxx update the record ...
 413                     # given that we record the current set of users anyways, there does not seem to be much left to do here
 414                     # self.logger.warning ("Slice update not yet implemented on slice %s (%s)"%(slice_hrn,slice['name']))
 415                     pass
 416                 # record current users affiliated with the slice
 417                 slice_record.reg_researchers = \
 418                     [ self.locate_by_type_pointer ('user',user_id) for user_id in slice['person_ids'] ]
 419                 global_dbsession.commit()
 420                 slice_record.stale=False
 421
 422         ### remove stale records
 423         # special records must be preserved
 424         system_hrns = [interface_hrn, root_auth, interface_hrn + '.slicemanager']
 425         for record in all_records:
 426             if record.hrn in system_hrns:
 427                 record.stale=False
 428             if record.peer_authority:
 429                 record.stale=False
 430             if ".vini" in interface_hrn and interface_hrn.endswith('vini') and \
 431                 record.hrn.endswith("internet2"):
 432                 record.stale=False
 433
 434         for record in all_records:
 435             try:        stale=record.stale
 436             except:
 437                 stale=True
 438                 self.logger.warning("stale not found with %s"%record)
 439             if stale:
 440                 self.logger.info("PlImporter: deleting stale record: %s" % record)
 441                 global_dbsession.delete(record)
 442                 global_dbsession.commit()