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         # also gather non-enabled user accounts so as to issue relevant warnings
 157         disabled_persons = shell.GetPersons({'peer_id': None, 'enabled': False}, ['person_id'])
 158         disabled_person_ids = [ person['person_id'] for person in disabled_persons ]
 159         # Get all plc public keys
 160         # accumulate key ids for keys retrieval
 161         key_ids = []
 162         for person in persons:
 163             key_ids.extend(person['key_ids'])
 164         keys = shell.GetKeys( {'peer_id': None, 'key_id': key_ids,
 165                                'key_type': 'ssh'} )
 166         # create a hash of keys by key_id
 167         keys_by_id = dict ( [ ( key['key_id'], key ) for key in keys ] )
 168         # create a dict person_id -> [ (plc)keys ]
 169         keys_by_person_id = {}
 170         for person in persons:
 171             pubkeys = []
 172             for key_id in person['key_ids']:
 173                 # by construction all the keys we fetched are ssh keys
 174                 # so gpg keys won't be in there
 175                 try:
 176                     key = keys_by_id[key_id]
 177                     pubkeys.append(key)
 178                 except:
 179                     self.logger.warning("Could not spot key %d - probably non-ssh"%key_id)
 180             keys_by_person_id[person['person_id']] = pubkeys
 181         # Get all plc nodes
 182         nodes = shell.GetNodes( {'peer_id': None}, ['node_id', 'hostname', 'site_id'])
 183         # create hash by node_id
 184         nodes_by_id = dict ( [ ( node['node_id'], node, ) for node in nodes ] )
 185         # Get all plc slices
 186         slices = shell.GetSlices( {'peer_id': None}, ['slice_id', 'name', 'person_ids'])
 187         # create hash by slice_id
 188         slices_by_id = dict ( [ (slice['slice_id'], slice ) for slice in slices ] )
 189
 190         # isolate special vini case in separate method
 191         self.create_special_vini_record (interface_hrn)
 192
 193         # start importing
 194         for site in sites:
 195             site_hrn = _get_site_hrn(interface_hrn, site)
 196             # import if hrn is not in list of existing hrns or if the hrn exists
 197             # but its not a site record
 198             site_record=self.locate_by_type_hrn ('authority', site_hrn)
 199             if not site_record:
 200                 try:
 201                     urn = hrn_to_urn(site_hrn, 'authority')
 202                     if not self.auth_hierarchy.auth_exists(urn):
 203                         self.auth_hierarchy.create_auth(urn)
 204                     auth_info = self.auth_hierarchy.get_auth_info(urn)
 205                     site_record = RegAuthority(hrn=site_hrn, gid=auth_info.get_gid_object(),
 206                                                pointer=site['site_id'],
 207                                                authority=get_authority(site_hrn))
 208                     site_record.just_created()
 209                     dbsession.add(site_record)
 210                     dbsession.commit()
 211                     self.logger.info("PlImporter: imported authority (site) : %s" % site_record)
 212                     self.remember_record (site_record)
 213                 except:
 214                     # if the site import fails then there is no point in trying to import the
 215                     # site's child records (node, slices, persons), so skip them.
 216                     self.logger.log_exc("PlImporter: failed to import site %s. Skipping child records"%site_hrn)
 217                     continue
 218             else:
 219                 # xxx update the record ...
 220                 pass
 221             site_record.stale=False
 222
 223             # import node records
 224             for node_id in site['node_ids']:
 225                 try:
 226                     node = nodes_by_id[node_id]
 227                 except:
 228                     self.logger.warning ("PlImporter: cannot find node_id %s - ignored"%node_id)
 229                     continue
 230                 site_auth = get_authority(site_hrn)
 231                 site_name = site['login_base']
 232                 node_hrn =  hostname_to_hrn(site_auth, site_name, node['hostname'])
 233                 # xxx this sounds suspicious
 234                 if len(node_hrn) > 64: node_hrn = node_hrn[:64]
 235                 node_record = self.locate_by_type_hrn ( 'node', node_hrn )
 236                 if not node_record:
 237                     try:
 238                         pkey = Keypair(create=True)
 239                         urn = hrn_to_urn(node_hrn, 'node')
 240                         node_gid = self.auth_hierarchy.create_gid(urn, create_uuid(), pkey)
 241                         node_record = RegNode (hrn=node_hrn, gid=node_gid,
 242                                                pointer =node['node_id'],
 243                                                authority=get_authority(node_hrn))
 244                         node_record.just_created()
 245                         dbsession.add(node_record)
 246                         dbsession.commit()
 247                         self.logger.info("PlImporter: imported node: %s" % node_record)
 248                         self.remember_record (node_record)
 249                     except:
 250                         self.logger.log_exc("PlImporter: failed to import node %s"%node_hrn)
 251                         continue
 252                 else:
 253                     # xxx update the record ...
 254                     pass
 255                 node_record.stale=False
 256
 257             site_pis=[]
 258             # import persons
 259             for person_id in site['person_ids']:
 260                 proceed=False
 261                 if person_id in persons_by_id:
 262                     person=persons_by_id[person_id]
 263                     proceed=True
 264                 elif person_id in disabled_person_ids:
 265                     pass
 266                 else:
 267                     self.logger.warning ("PlImporter: cannot locate person_id %s in site %s - ignored"%(person_id,site_hrn))
 268                 # make sure to NOT run this if anything is wrong
 269                 if not proceed: continue
 270
 271                 person_hrn = email_to_hrn(site_hrn, person['email'])
 272                 # xxx suspicious again
 273                 if len(person_hrn) > 64: person_hrn = person_hrn[:64]
 274                 person_urn = hrn_to_urn(person_hrn, 'user')
 275
 276                 user_record = self.locate_by_type_hrn ( 'user', person_hrn)
 277
 278                 # return a tuple pubkey (a plc key object) and pkey (a Keypair object)
 279                 def init_person_key (person, plc_keys):
 280                     pubkey=None
 281                     if  person['key_ids']:
 282                         # randomly pick first key in set
 283                         pubkey = plc_keys[0]
 284                         try:
 285                             pkey = convert_public_key(pubkey['key'])
 286                         except:
 287                             self.logger.warn('PlImporter: unable to convert public key for %s' % person_hrn)
 288                             pkey = Keypair(create=True)
 289                     else:
 290                         # the user has no keys. Creating a random keypair for the user's gid
 291                         self.logger.warn("PlImporter: person %s does not have a PL public key"%person_hrn)
 292                         pkey = Keypair(create=True)
 293                     return (pubkey, pkey)
 294
 295                 # new person
 296                 try:
 297                     plc_keys = keys_by_person_id.get(person['person_id'],[])
 298                     if not user_record:
 299                         (pubkey,pkey) = init_person_key (person, plc_keys )
 300                         person_gid = self.auth_hierarchy.create_gid(person_urn, create_uuid(), pkey)
 301                         person_gid.set_email(person['email'])
 302                         user_record = RegUser (hrn=person_hrn, gid=person_gid,
 303                                                  pointer=person['person_id'],
 304                                                  authority=get_authority(person_hrn),
 305                                                  email=person['email'])
 306                         if pubkey:
 307                             user_record.reg_keys=[RegKey (pubkey['key'], pubkey['key_id'])]
 308                         else:
 309                             self.logger.warning("No key found for user %s"%user_record)
 310                         user_record.just_created()
 311                         dbsession.add (user_record)
 312                         dbsession.commit()
 313                         self.logger.info("PlImporter: imported person: %s" % user_record)
 314                         self.remember_record ( user_record )
 315                     else:
 316                         # update the record ?
 317                         # if user's primary key has changed then we need to update the
 318                         # users gid by forcing an update here
 319                         sfa_keys = user_record.reg_keys
 320                         def key_in_list (key,sfa_keys):
 321                             for reg_key in sfa_keys:
 322                                 if reg_key.key==key['key']: return True
 323                             return False
 324                         # is there a new key in myplc ?
 325                         new_keys=False
 326                         for key in plc_keys:
 327                             if not key_in_list (key,sfa_keys):
 328                                 new_keys = True
 329                         if new_keys:
 330                             (pubkey,pkey) = init_person_key (person, plc_keys)
 331                             person_gid = self.auth_hierarchy.create_gid(person_urn, create_uuid(), pkey)
 332                             if not pubkey:
 333                                 user_record.reg_keys=[]
 334                             else:
 335                                 user_record.reg_keys=[ RegKey (pubkey['key'], pubkey['key_id'])]
 336                             self.logger.info("PlImporter: updated person: %s" % user_record)
 337                     user_record.email = person['email']
 338                     dbsession.commit()
 339                     user_record.stale=False
 340                     # accumulate PIs - PLCAPI has a limitation that when someone has PI role
 341                     # this is valid for all sites she is in..
 342                     # PI is coded with role_id==20
 343                     if 20 in person['role_ids']:
 344                         site_pis.append (user_record)
 345                 except:
 346                     self.logger.log_exc("PlImporter: failed to import person %d %s"%(person['person_id'],person['email']))
 347
 348             # maintain the list of PIs for a given site
 349             # for the record, Jordan had proposed the following addition as a welcome hotfix to a previous version:
 350             # site_pis = list(set(site_pis))
 351             # this was likely due to a bug in the above logic,
 352             # that had to do with enabled persons, and where the whole loop on persons
 353             # could be performed twice with the same person...
 354             # so hopefully we do not need to eliminate duplicates explicitly here anymore
 355             site_record.reg_pis = site_pis
 356             dbsession.commit()
 357
 358             # import slices
 359             for slice_id in site['slice_ids']:
 360                 try:
 361                     slice = slices_by_id[slice_id]
 362                 except:
 363                     self.logger.warning ("PlImporter: cannot locate slice_id %s - ignored"%slice_id)
 364                 slice_hrn = slicename_to_hrn(interface_hrn, slice['name'])
 365                 slice_record = self.locate_by_type_hrn ('slice', slice_hrn)
 366                 if not slice_record:
 367                     try:
 368                         pkey = Keypair(create=True)
 369                         urn = hrn_to_urn(slice_hrn, 'slice')
 370                         slice_gid = self.auth_hierarchy.create_gid(urn, create_uuid(), pkey)
 371                         slice_record = RegSlice (hrn=slice_hrn, gid=slice_gid,
 372                                                  pointer=slice['slice_id'],
 373                                                  authority=get_authority(slice_hrn))
 374                         slice_record.just_created()
 375                         dbsession.add(slice_record)
 376                         dbsession.commit()
 377                         self.logger.info("PlImporter: imported slice: %s" % slice_record)
 378                         self.remember_record ( slice_record )
 379                     except:
 380                         self.logger.log_exc("PlImporter: failed to import slice %s (%s)"%(slice_hrn,slice['name']))
 381                 else:
 382                     # xxx update the record ...
 383                     # given that we record the current set of users anyways, there does not seem to be much left to do here
 384                     # self.logger.warning ("Slice update not yet implemented on slice %s (%s)"%(slice_hrn,slice['name']))
 385                     pass
 386                 # record current users affiliated with the slice
 387                 slice_record.reg_researchers = \
 388                     [ self.locate_by_type_pointer ('user',user_id) for user_id in slice['person_ids'] ]
 389                 dbsession.commit()
 390                 slice_record.stale=False
 391
 392         ### remove stale records
 393         # special records must be preserved
 394         system_hrns = [interface_hrn, root_auth, interface_hrn + '.slicemanager']
 395         for record in all_records:
 396             if record.hrn in system_hrns:
 397                 record.stale=False
 398             if record.peer_authority:
 399                 record.stale=False
 400             if ".vini" in interface_hrn and interface_hrn.endswith('vini') and \
 401                 record.hrn.endswith("internet2"):
 402                 record.stale=False
 403
 404         for record in all_records:
 405             try:        stale=record.stale
 406             except:
 407                 stale=True
 408                 self.logger.warning("stale not found with %s"%record)
 409             if stale:
 410                 self.logger.info("PlImporter: deleting stale record: %s" % record)
 411                 dbsession.delete(record)
 412                 dbsession.commit()