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 from sfa.storage.alchemy import dbsession
  28 from sfa.storage.model import RegRecord, RegAuthority, RegSlice, RegNode, RegUser, RegKey
  29
  30 from sfa.planetlab.plshell import PlShell
  31 from sfa.planetlab.plxrn import hostname_to_hrn, slicename_to_hrn, email_to_hrn, hrn_to_pl_slicename
  32
  33 def _get_site_hrn(interface_hrn, site):
  34     # Hardcode 'internet2' into the hrn for sites hosting
  35     # internet2 nodes. This is a special operation for some vini
  36     # sites only
  37     hrn = ".".join([interface_hrn, site['login_base']])
  38     if ".vini" in interface_hrn and interface_hrn.endswith('vini'):
  39         if site['login_base'].startswith("i2") or site['login_base'].startswith("nlr"):
  40             hrn = ".".join([interface_hrn, "internet2", site['login_base']])
  41     return hrn
  42
  43
  44 class PlImporter:
  45
  46     def __init__ (self, auth_hierarchy, logger):
  47         self.auth_hierarchy = auth_hierarchy
  48         self.logger=logger
  49
  50     def add_options (self, parser):
  51         # we don't have any options for now
  52         pass
  53
  54     # hrn hash is initialized from current db
  55     # remember just-created records as we go
  56     # xxx might make sense to add a UNIQUE constraint in the db itself
  57     def remember_record_by_hrn (self, record):
  58         tuple = (record.type, record.hrn)
  59         if tuple in self.records_by_type_hrn:
  60             self.logger.warning ("PlImporter.remember_record_by_hrn: duplicate (%s,%s)"%tuple)
  61             return
  62         self.records_by_type_hrn [ tuple ] = record
  63
  64     # ditto for pointer hash
  65     def remember_record_by_pointer (self, record):
  66         if record.pointer == -1:
  67             self.logger.warning ("PlImporter.remember_record_by_pointer: pointer is void")
  68             return
  69         tuple = (record.type, record.pointer)
  70         if tuple in self.records_by_type_pointer:
  71             self.logger.warning ("PlImporter.remember_record_by_pointer: duplicate (%s,%s)"%tuple)
  72             return
  73         self.records_by_type_pointer [ ( record.type, record.pointer,) ] = record
  74
  75     def remember_record (self, record):
  76         self.remember_record_by_hrn (record)
  77         self.remember_record_by_pointer (record)
  78
  79     def locate_by_type_hrn (self, type, hrn):
  80         return self.records_by_type_hrn.get ( (type, hrn), None)
  81
  82     def locate_by_type_pointer (self, type, pointer):
  83         return self.records_by_type_pointer.get ( (type, pointer), None)
  84
  85     # a convenience/helper function to see if a record is already known
  86     # a former, broken, attempt (in 2.1-9) had been made
  87     # to try and use 'pointer' as a first, most significant attempt
  88     # the idea being to preserve stuff as much as possible, and thus
  89     # to avoid creating a new gid in the case of a simple hrn rename
  90     # however this of course doesn't work as the gid depends on the hrn...
  91     #def locate (self, type, hrn=None, pointer=-1):
  92     #    if pointer!=-1:
  93     #        attempt = self.locate_by_type_pointer (type, pointer)
  94     #        if attempt : return attempt
  95     #    if hrn is not None:
  96     #        attempt = self.locate_by_type_hrn (type, hrn,)
  97     #        if attempt : return attempt
  98     #    return None
  99
 100     # this makes the run method a bit abtruse - out of the way
 101     def create_special_vini_record (self, interface_hrn):
 102         # special case for vini
 103         if ".vini" in interface_hrn and interface_hrn.endswith('vini'):
 104             # create a fake internet2 site first
 105             i2site = {'name': 'Internet2', 'login_base': 'internet2', 'site_id': -1}
 106             site_hrn = _get_site_hrn(interface_hrn, i2site)
 107             # import if hrn is not in list of existing hrns or if the hrn exists
 108             # but its not a site record
 109             if ( 'authority', site_hrn, ) not in self.records_by_type_hrn:
 110                 urn = hrn_to_urn(site_hrn, 'authority')
 111                 if not self.auth_hierarchy.auth_exists(urn):
 112                     self.auth_hierarchy.create_auth(urn)
 113                 auth_info = self.auth_hierarchy.get_auth_info(urn)
 114                 auth_record = RegAuthority(hrn=site_hrn, gid=auth_info.get_gid_object(),
 115                                            pointer=site['site_id'],
 116                                            authority=get_authority(site_hrn))
 117                 auth_record.just_created()
 118                 dbsession.add(auth_record)
 119                 dbsession.commit()
 120                 self.logger.info("PlImporter: Imported authority (vini site) %s"%auth_record)
 121                 self.remember_record ( site_record )
 122
 123     def run (self, options):
 124         config = Config ()
 125         interface_hrn = config.SFA_INTERFACE_HRN
 126         root_auth = config.SFA_REGISTRY_ROOT_AUTH
 127         shell = PlShell (config)
 128
 129         ######## retrieve all existing SFA objects
 130         all_records = dbsession.query(RegRecord).all()
 131
 132         # create hash by (type,hrn)
 133         # we essentially use this to know if a given record is already known to SFA
 134         self.records_by_type_hrn = \
 135             dict ( [ ( (record.type, record.hrn) , record ) for record in all_records ] )
 136         # create hash by (type,pointer)
 137         self.records_by_type_pointer = \
 138             dict ( [ ( (record.type, record.pointer) , record ) for record in all_records
 139                      if record.pointer != -1] )
 140
 141         # initialize record.stale to True by default, then mark stale=False on the ones that are in use
 142         for record in all_records: record.stale=True
 143
 144         ######## retrieve PLC data
 145         # Get all plc sites
 146         # retrieve only required stuf
 147         sites = shell.GetSites({'peer_id': None, 'enabled' : True},
 148                                ['site_id','login_base','node_ids','slice_ids','person_ids',])
 149         # create a hash of sites by login_base
 150 #        sites_by_login_base = dict ( [ ( site['login_base'], site ) for site in sites ] )
 151         # Get all plc users
 152         persons = shell.GetPersons({'peer_id': None, 'enabled': True},
 153                                    ['person_id', 'email', 'key_ids', 'site_ids', 'role_ids'])
 154         # create a hash of persons by person_id
 155         persons_by_id = dict ( [ ( person['person_id'], person) for person in persons ] )
 156         # Get all plc public keys
 157         # accumulate key ids for keys retrieval
 158         key_ids = []
 159         for person in persons:
 160             key_ids.extend(person['key_ids'])
 161         keys = shell.GetKeys( {'peer_id': None, 'key_id': key_ids,
 162                                'key_type': 'ssh'} )
 163         # create a hash of keys by key_id
 164         keys_by_id = dict ( [ ( key['key_id'], key ) for key in keys ] )
 165         # create a dict person_id -> [ (plc)keys ]
 166         keys_by_person_id = {}
 167         for person in persons:
 168             pubkeys = []
 169             for key_id in person['key_ids']:
 170                 # by construction all the keys we fetched are ssh keys
 171                 # so gpg keys won't be in there
 172                 try:
 173                     key = keys_by_id[key_id]
 174                     pubkeys.append(key)
 175                 except:
 176                     self.logger.warning("Could not spot key %d - probably non-ssh"%key_id)
 177             keys_by_person_id[person['person_id']] = pubkeys
 178         # Get all plc nodes
 179         nodes = shell.GetNodes( {'peer_id': None}, ['node_id', 'hostname', 'site_id'])
 180         # create hash by node_id
 181         nodes_by_id = dict ( [ ( node['node_id'], node, ) for node in nodes ] )
 182         # Get all plc slices
 183         slices = shell.GetSlices( {'peer_id': None}, ['slice_id', 'name', 'person_ids'])
 184         # create hash by slice_id
 185         slices_by_id = dict ( [ (slice['slice_id'], slice ) for slice in slices ] )
 186
 187         # isolate special vini case in separate method
 188         self.create_special_vini_record (interface_hrn)
 189
 190         # start importing
 191         for site in sites:
 192             site_hrn = _get_site_hrn(interface_hrn, site)
 193             # import if hrn is not in list of existing hrns or if the hrn exists
 194             # but its not a site record
 195             site_record=self.locate_by_type_hrn ('authority', site_hrn)
 196             if not site_record:
 197                 try:
 198                     urn = hrn_to_urn(site_hrn, 'authority')
 199                     if not self.auth_hierarchy.auth_exists(urn):
 200                         self.auth_hierarchy.create_auth(urn)
 201                     auth_info = self.auth_hierarchy.get_auth_info(urn)
 202                     site_record = RegAuthority(hrn=site_hrn, gid=auth_info.get_gid_object(),
 203                                                pointer=site['site_id'],
 204                                                authority=get_authority(site_hrn))
 205                     site_record.just_created()
 206                     dbsession.add(site_record)
 207                     dbsession.commit()
 208                     self.logger.info("PlImporter: imported authority (site) : %s" % site_record)
 209                     self.remember_record (site_record)
 210                 except:
 211                     # if the site import fails then there is no point in trying to import the
 212                     # site's child records (node, slices, persons), so skip them.
 213                     self.logger.log_exc("PlImporter: failed to import site. Skipping child records")
 214                     continue
 215             else:
 216                 # xxx update the record ...
 217                 pass
 218             site_record.stale=False
 219
 220             # import node records
 221             for node_id in site['node_ids']:
 222                 try:
 223                     node = nodes_by_id[node_id]
 224                 except:
 225                     self.logger.warning ("PlImporter: cannot find node_id %s - ignored"%node_id)
 226                     continue
 227                 site_auth = get_authority(site_hrn)
 228                 site_name = site['login_base']
 229                 node_hrn =  hostname_to_hrn(site_auth, site_name, node['hostname'])
 230                 # xxx this sounds suspicious
 231                 if len(node_hrn) > 64: node_hrn = node_hrn[:64]
 232                 node_record = self.locate_by_type_hrn ( 'node', node_hrn )
 233                 if not node_record:
 234                     try:
 235                         pkey = Keypair(create=True)
 236                         urn = hrn_to_urn(node_hrn, 'node')
 237                         node_gid = self.auth_hierarchy.create_gid(urn, create_uuid(), pkey)
 238                         node_record = RegNode (hrn=node_hrn, gid=node_gid,
 239                                                pointer =node['node_id'],
 240                                                authority=get_authority(node_hrn))
 241                         node_record.just_created()
 242                         dbsession.add(node_record)
 243                         dbsession.commit()
 244                         self.logger.info("PlImporter: imported node: %s" % node_record)
 245                         self.remember_record (node_record)
 246                     except:
 247                         self.logger.log_exc("PlImporter: failed to import node")
 248                 else:
 249                     # xxx update the record ...
 250                     pass
 251                 node_record.stale=False
 252
 253             site_pis=set()
 254             # import persons
 255             for person_id in site['person_ids']:
 256                 try:
 257                     person = persons_by_id[person_id]
 258                 except:
 259                     self.logger.warning ("PlImporter: cannot locate person_id %s - ignored"%person_id)
 260                 person_hrn = email_to_hrn(site_hrn, person['email'])
 261                 # xxx suspicious again
 262                 if len(person_hrn) > 64: person_hrn = person_hrn[:64]
 263                 person_urn = hrn_to_urn(person_hrn, 'user')
 264
 265                 user_record = self.locate_by_type_hrn ( 'user', person_hrn)
 266
 267                 # return a tuple pubkey (a plc key object) and pkey (a Keypair object)
 268                 def init_person_key (person, plc_keys):
 269                     pubkey=None
 270                     if  person['key_ids']:
 271                         # randomly pick first key in set
 272                         pubkey = plc_keys[0]
 273                         try:
 274                             pkey = convert_public_key(pubkey['key'])
 275                         except:
 276                             self.logger.warn('PlImporter: unable to convert public key for %s' % person_hrn)
 277                             pkey = Keypair(create=True)
 278                     else:
 279                         # the user has no keys. Creating a random keypair for the user's gid
 280                         self.logger.warn("PlImporter: person %s does not have a PL public key"%person_hrn)
 281                         pkey = Keypair(create=True)
 282                     return (pubkey, pkey)
 283
 284                 # new person
 285                 try:
 286                     plc_keys = keys_by_person_id.get(person['person_id'],[])
 287                     if not user_record:
 288                         (pubkey,pkey) = init_person_key (person, plc_keys )
 289                         person_gid = self.auth_hierarchy.create_gid(person_urn, create_uuid(), pkey)
 290                         person_gid.set_email(person['email'])
 291                         user_record = RegUser (hrn=person_hrn, gid=person_gid,
 292                                                  pointer=person['person_id'],
 293                                                  authority=get_authority(person_hrn),
 294                                                  email=person['email'])
 295                         if pubkey:
 296                             user_record.reg_keys=[RegKey (pubkey['key'], pubkey['key_id'])]
 297                         else:
 298                             self.logger.warning("No key found for user %s"%user_record)
 299                         user_record.just_created()
 300                         dbsession.add (user_record)
 301                         dbsession.commit()
 302                         self.logger.info("PlImporter: imported person: %s" % user_record)
 303                         self.remember_record ( user_record )
 304                     else:
 305                         # update the record ?
 306                         # if user's primary key has changed then we need to update the
 307                         # users gid by forcing an update here
 308                         sfa_keys = user_record.reg_keys
 309                         def key_in_list (key,sfa_keys):
 310                             for reg_key in sfa_keys:
 311                                 if reg_key.key==key['key']: return True
 312                             return False
 313                         # is there a new key in myplc ?
 314                         new_keys=False
 315                         for key in plc_keys:
 316                             if not key_in_list (key,sfa_keys):
 317                                 new_keys = True
 318                         if new_keys:
 319                             (pubkey,pkey) = init_person_key (person, plc_keys)
 320                             person_gid = self.auth_hierarchy.create_gid(person_urn, create_uuid(), pkey)
 321                             if not pubkey:
 322                                 user_record.reg_keys=[]
 323                             else:
 324                                 user_record.reg_keys=[ RegKey (pubkey['key'], pubkey['key_id'])]
 325                             self.logger.info("PlImporter: updated person: %s" % user_record)
 326                     user_record.email = person['email']
 327                     dbsession.commit()
 328                     user_record.stale=False
 329                     # accumulate PIs - PLCAPI has a limitation that when someone has PI role
 330                     # this is valid for all sites she is in..
 331                     # PI is coded with role_id==20
 332                     if 20 in person['role_ids']:
 333                         site_pis.add (user_record)
 334                 except:
 335                     self.logger.log_exc("PlImporter: failed to import person %d %s"%(person['person_id'],person['email']))
 336
 337             # maintain the list of PIs for a given site
 338             site_record.reg_pis = list(site_pis)
 339             dbsession.commit()
 340
 341             # import slices
 342             for slice_id in site['slice_ids']:
 343                 try:
 344                     slice = slices_by_id[slice_id]
 345                 except:
 346                     self.logger.warning ("PlImporter: cannot locate slice_id %s - ignored"%slice_id)
 347                 slice_hrn = slicename_to_hrn(interface_hrn, slice['name'])
 348                 slice_record = self.locate_by_type_hrn ('slice', slice_hrn)
 349                 if not slice_record:
 350                     try:
 351                         pkey = Keypair(create=True)
 352                         urn = hrn_to_urn(slice_hrn, 'slice')
 353                         slice_gid = self.auth_hierarchy.create_gid(urn, create_uuid(), pkey)
 354                         slice_record = RegSlice (hrn=slice_hrn, gid=slice_gid,
 355                                                  pointer=slice['slice_id'],
 356                                                  authority=get_authority(slice_hrn))
 357                         slice_record.just_created()
 358                         dbsession.add(slice_record)
 359                         dbsession.commit()
 360                         self.logger.info("PlImporter: imported slice: %s" % slice_record)
 361                         self.remember_record ( slice_record )
 362                     except:
 363                         self.logger.log_exc("PlImporter: failed to import slice")
 364                 else:
 365                     # update the pointer if it has changed
 366                     if slice_id != slice_record.pointer:
 367                         self.logger.info("updating record (slice) pointer")
 368                         slice_record.pointer = slice_id
 369                         dbsession.commit()
 370                     # xxx update the record ...
 371                     #self.logger.warning ("Slice update not yet implemented")
 372                 # record current users affiliated with the slice
 373                 slice_record.reg_researchers = \
 374                     [ self.locate_by_type_pointer ('user',user_id) for user_id in slice['person_ids'] ]
 375                 dbsession.commit()
 376                 slice_record.stale=False
 377
 378         ### remove stale records
 379         # special records must be preserved
 380         system_hrns = [interface_hrn, root_auth, interface_hrn + '.slicemanager']
 381         for record in all_records:
 382             if record.hrn in system_hrns:
 383                 record.stale=False
 384             if record.peer_authority:
 385                 record.stale=False
 386             if ".vini" in interface_hrn and interface_hrn.endswith('vini') and \
 387                 record.hrn.endswith("internet2"):
 388                 record.stale=False
 389
 390         for record in all_records:
 391             try:        stale=record.stale
 392             except:
 393                 stale=True
 394                 self.logger.warning("stale not found with %s"%record)
 395             if stale:
 396                 self.logger.info("PlImporter: deleting stale record: %s" % record)
 397                 dbsession.delete(record)
 398                 dbsession.commit()