class PlSlices:
- rspec_to_slice_tag = {'max_rate':'net_max_rate'}
+ rspec_to_slice_tag = {'max_rate' : 'net_max_rate'}
def __init__(self, driver):
self.driver = driver
person_ids = list(person_ids)
all_slice_tag_ids = list(all_slice_tag_ids)
# Get user information
- all_persons_list = self.driver.shell.GetPersons({'person_id':person_ids,'enabled':True}, ['person_id', 'enabled', 'key_ids'])
+ all_persons_list = self.driver.shell.GetPersons({'person_id':person_ids,'enabled':True},
+ ['person_id', 'enabled', 'key_ids'])
all_persons = {}
for person in all_persons_list:
all_persons[person['person_id']] = person
# XXX Sanity check; though technically this should be a system invariant
# checked with an assertion
- if slice['expires'] > MAXINT: slice['expires']= MAXINT
+ if slice['expires'] > MAXINT:
+ slice['expires'] = MAXINT
slivers.append({
'hrn': hrn,
def verify_slice_leases(self, slice, rspec_requested_leases):
- leases = self.driver.shell.GetLeases({'name':slice['name'], 'clip':int(time.time())}, ['lease_id','name', 'hostname', 't_from', 't_until'])
+ leases = self.driver.shell.GetLeases({'name':slice['name'], 'clip':int(time.time())},
+ ['lease_id','name', 'hostname', 't_from', 't_until'])
grain = self.driver.shell.GetLeaseGranularity()
requested_leases = []
for node in resulting_nodes:
client_id = slivers[node['hostname']]['client_id']
component_id = slivers[node['hostname']]['component_id']
- sliver_hrn = '%s.%s-%s' % (self.driver.hrn, slice['slice_id'], node['node_id'])
+ sliver_hrn = '{}.{}-{}'.format(self.driver.hrn, slice['slice_id'], node['node_id'])
sliver_id = Xrn(sliver_hrn, type='sliver').urn
record = SliverAllocation(sliver_id=sliver_id, client_id=client_id,
component_id=component_id,
slice_tags.append({'name': 'vini_topo', 'value': 'manual', 'node_id': node_id})
#self.driver.shell.AddSliceTag(slice['name'], 'topo_rspec', str([topo_rspec]), node_id)
- self.verify_slice_attributes(slice, slice_tags, {'append': True}, admin=True)
-
+ self.verify_slice_tags(slice, slice_tags, {'pltags':'append'}, admin=True)
- def verify_site(self, slice_xrn, slice_record={}, sfa_peer=None, options={}):
+ def verify_site(self, slice_xrn, slice_record=None, sfa_peer=None, options=None):
+ if slice_record is None: slice_record={}
+ if options is None: options={}
(slice_hrn, type) = urn_to_hrn(slice_xrn)
top_auth_hrn = top_auth(slice_hrn)
site_hrn = '.'.join(slice_hrn.split('.')[:-1])
login_base = slice_hrn.split('.')[-2][:12]
else:
login_base = hash_loginbase(site_hrn)
- #plxrn = PlXrn(xrn=slice_xrn)
- #slice_hrn = plxrn.get_hrn()
- #type = plxrn.get_type()
- #site_hrn = plxrn.get_authority_hrn()
- #authority_name = plxrn.pl_authname()
- #slicename = plxrn.pl_slicename()
- #login_base = plxrn.pl_login_base()
-
- sites = self.driver.shell.GetSites({'peer_id': None},['site_id','name','abbreviated_name','login_base','hrn'])
# filter sites by hrn
- site_exists = [site for site in sites if site['hrn'] == site_hrn]
+ sites = self.driver.shell.GetSites({'peer_id': None, 'hrn':site_hrn},
+ ['site_id','name','abbreviated_name','login_base','hrn'])
- if not site_exists:
+ # alredy exists
+ if sites:
+ site = sites[0]
+ else:
# create new site record
- site = {'name': 'sfa:%s' % site_hrn,
+ site = {'name': 'sfa:{}'.format(site_hrn),
'abbreviated_name': site_hrn,
'login_base': login_base,
'max_slices': 100,
'max_slivers': 1000,
'enabled': True,
- 'peer_site_id': None}
-
- site['site_id'] = self.driver.shell.AddSite(site)
- # Set site HRN
- self.driver.shell.SetSiteHrn(int(site['site_id']), site_hrn)
- # Tag this as created through SFA
- self.driver.shell.SetSiteSfaCreated(int(site['site_id']), 'True')
+ 'peer_site_id': None,
+ 'hrn':site_hrn,
+ 'sfa_created': 'True',
+ }
+ site_id = self.driver.shell.AddSite(site)
+ # plcapi tends to mess with the incoming hrn so let's make sure
+ self.driver.shell.SetSiteHrn (site_id, site_hrn)
+ site['site_id'] = site_id
# exempt federated sites from monitor policies
- self.driver.shell.AddSiteTag(int(site['site_id']), 'exempt_site_until', "20200101")
-
- else:
- site = site_exists[0]
+ self.driver.shell.AddSiteTag(site_id, 'exempt_site_until', "20200101")
return site
- def verify_slice(self, slice_hrn, slice_record, sfa_peer, expiration, options={}):
+ def verify_slice(self, slice_hrn, slice_record, sfa_peer, expiration, options=None):
+ if options is None: options={}
top_auth_hrn = top_auth(slice_hrn)
site_hrn = '.'.join(slice_hrn.split('.')[:-1])
slice_part = slice_hrn.split('.')[-1]
else:
login_base = hash_loginbase(site_hrn)
slice_name = '_'.join([login_base, slice_part])
- #plxrn = PlXrn(xrn=slice_hrn)
- #slice_hrn = plxrn.get_hrn()
- #type = plxrn.get_type()
- #site_hrn = plxrn.get_authority_hrn()
- #authority_name = plxrn.pl_authname()
- #slicename = plxrn.pl_slicename()
- #login_base = plxrn.pl_login_base()
-
- slices = self.driver.shell.GetSlices({'peer_id': None},['slice_id','name','hrn'])
- # Filter slices by HRN
- slice_exists = [slice for slice in slices if slice['hrn'] == slice_hrn]
+
expires = int(datetime_to_epoch(utcparse(expiration)))
- if not slice_exists:
+ # Filter slices by HRN
+ slices = self.driver.shell.GetSlices({'peer_id': None, 'hrn':slice_hrn},
+ ['slice_id','name','hrn','expires'])
+
+ if slices:
+ slice = slices[0]
+ slice_id = slice['slice_id']
+ #Update expiration if necessary
+ if slice.get('expires', None) != expires:
+ self.driver.shell.UpdateSlice( slice_id, {'expires' : expires})
+ else:
if slice_record:
url = slice_record.get('url', slice_hrn)
description = slice_record.get('description', slice_hrn)
description = slice_hrn
slice = {'name': slice_name,
'url': url,
- 'description': description}
- # add the slice
- slice['slice_id'] = self.driver.shell.AddSlice(slice)
- # set the slice HRN
- self.driver.shell.SetSliceHrn(int(slice['slice_id']), slice_hrn)
- # Tag this as created through SFA
- self.driver.shell.SetSliceSfaCreated(int(slice['slice_id']), 'True')
+ 'description': description,
+ 'hrn': slice_hrn,
+ 'sfa_created': 'True',
+ #'expires': expires,
+ }
+ # add the slice
+ slice_id = self.driver.shell.AddSlice(slice)
+ # plcapi tends to mess with the incoming hrn so let's make sure
+ self.driver.shell.SetSliceHrn (slice_id, slice_hrn)
+ # cannot be set with AddSlice
# set the expiration
- self.driver.shell.UpdateSlice(int(slice['slice_id']), {'expires': expires})
-
- else:
- slice = slice_exists[0]
- #Update expiration if necessary
- if slice.get('expires', None) != expires:
- self.driver.shell.UpdateSlice( int(slice['slice_id']), {'expires' : expires})
-
- return self.driver.shell.GetSlices(int(slice['slice_id']))[0]
+ self.driver.shell.UpdateSlice(slice_id, {'expires': expires})
+
+ return self.driver.shell.GetSlices(slice_id)[0]
+
+
+ # in the following code, we use
+ # 'person' to denote a PLCAPI-like record with typically 'person_id' and 'email'
+ # 'user' to denote an incoming record with typically 'urn' and 'email' - we add 'hrn' in there
+ # 'slice_record': it seems like the first of these 'users' also contains a 'slice_record'
+ # key that holds stuff like 'hrn', 'slice_id', 'authority',...
+ #
+ def create_person_from_user (self, user, site_id):
+ user_hrn = user['hrn']
+ # the value to use if 'user' has no 'email' attached - or if the attached email already exists
+ # typically
+ ( auth_hrn, _ , leaf ) = user_hrn.rpartition('.')
+ # somehow this has backslashes, get rid of them
+ auth_hrn = auth_hrn.replace('\\','')
+ default_email = "{}@{}.stub".format(leaf, auth_hrn)
+
+ person_record = {
+ # required
+ 'first_name': user.get('first_name',user_hrn),
+ 'last_name': user.get('last_name',user_hrn),
+ 'email': user.get('email', default_email),
+ # our additions
+ 'enabled': True,
+ 'sfa_created': 'True',
+ 'hrn': user_hrn,
+ }
+
+ logger.debug ("about to attempt to AddPerson with {}".format(person_record))
+ try:
+ # the thing is, the PLE db has a limitation on re-using the same e-mail
+ # in the case where people have an account on ple.upmc and then then come
+ # again from onelab.upmc, they will most likely have the same e-mail, and so kaboom..
+ # so we first try with the accurate email
+ person_id = int (self.driver.shell.AddPerson(person_record))
+ except:
+ logger.log_exc("caught during first attempt at AddPerson")
+ # and if that fails we start again with the email based on the hrn, which this time is unique..
+ person_record['email'] = default_email
+ logger.debug ("second chance with email={}".format(person_record['email']))
+ person_id = int (self.driver.shell.AddPerson(person_record))
+ self.driver.shell.AddRoleToPerson('user', person_id)
+ self.driver.shell.AddPersonToSite(person_id, site_id)
+ # plcapi tends to mess with the incoming hrn so let's make sure
+ self.driver.shell.SetPersonHrn (person_id, user_hrn)
+ # also 'enabled':True does not seem to pass through with AddPerson
+ self.driver.shell.UpdatePerson (person_id, {'enabled': True})
+
+ return person_id
+
+ def verify_persons(self, slice_hrn, slice_record, users, sfa_peer, options=None):
+ if options is None: options={}
+
+ # first we annotate the incoming users arg with a 'hrn' key
+ for user in users:
+ user['hrn'], _ = urn_to_hrn(user['urn'])
+ # this is for retrieving users from a hrn
+ users_by_hrn = { user['hrn'] : user for user in users }
+ for user in users: logger.debug("incoming user {}".format(user))
- def verify_persons(self, slice_hrn, slice_record, users, sfa_peer, options={}):
+ # compute the hrn's for the authority and site
top_auth_hrn = top_auth(slice_hrn)
site_hrn = '.'.join(slice_hrn.split('.')[:-1])
slice_part = slice_hrn.split('.')[-1]
- users_by_hrn = {}
- for user in users:
- user['hrn'], _ = urn_to_hrn(user['urn'])
- users_by_hrn[user['hrn']] = user
-
+ # deduce login_base and slice_name
if top_auth_hrn == self.driver.hrn:
login_base = slice_hrn.split('.')[-2][:12]
else:
login_base = hash_loginbase(site_hrn)
-
slice_name = '_'.join([login_base, slice_part])
- persons = self.driver.shell.GetPersons({'peer_id': None},['person_id','email','hrn'])
- sites = self.driver.shell.GetSites({'peer_id': None}, ['node_ids', 'site_id', 'name', 'person_ids', 'slice_ids', 'login_base', 'hrn'])
- site = [site for site in sites if site['hrn'] == site_hrn][0]
- slices = self.driver.shell.GetSlices({'peer_id': None}, ['slice_id', 'node_ids', 'person_ids', 'expires', 'site_id', 'name', 'hrn'])
- slice = [slice for slice in slices if slice['hrn'] == slice_hrn][0]
- slice_persons = self.driver.shell.GetPersons({'peer_id': None, 'person_id': slice['person_ids']},['person_id','email','hrn'])
-
- persons_by_hrn = {}
- persons_by_email = {}
- for person in persons:
- persons_by_hrn[person['hrn']] = person
- persons_by_email[person['email']] = person
- slice_persons_by_hrn = {}
- for slice_person in slice_persons:
- slice_persons_by_hrn[slice_person['hrn']] = slice_person
-
- # sort persons by HRN
- persons_to_add = set(users_by_hrn.keys()).difference(slice_persons_by_hrn.keys())
- persons_to_delete = set(slice_persons_by_hrn.keys()).difference(users_by_hrn.keys())
- persons_to_keep = set(users_by_hrn.keys()).intersection(slice_persons_by_hrn.keys())
-
-
+ # locate the site object
+ # due to a limitation in PLCAPI, we have to specify 'hrn' as part of the return fields
+ site = self.driver.shell.GetSites ({'peer_id':None, 'hrn':site_hrn}, ['site_id','hrn'])[0]
+ site_id = site['site_id']
+
+ # locate the slice object
+ slice = self.driver.shell.GetSlices ({'peer_id':None, 'hrn':slice_hrn}, ['slice_id','hrn','person_ids'])[0]
+ slice_id = slice['slice_id']
+ slice_person_ids = slice['person_ids']
+
+ # the common set of attributes for our calls to GetPersons
+ person_fields = ['person_id','email','hrn']
+
+ # for the intended set of hrns, locate existing persons
+ target_hrns = [ user['hrn'] for user in users ]
+ target_existing_persons = self.driver.shell.GetPersons ({'peer_id':None, 'hrn': target_hrns}, person_fields)
+ target_existing_person_ids = [ person ['person_id'] for person in target_existing_persons ]
+ # find out the hrns that *do not* have a corresponding person
+ existing_hrns = [ person['hrn'] for person in target_existing_persons ]
+ tocreate_hrns = set (target_hrns) - set (existing_hrns)
+ # create these
+ target_created_person_ids = [ self.create_person_from_user (users_by_hrn[hrn], site_id) for hrn in tocreate_hrns ]
+
+ # we can partition the persons of interest into one of these 3 classes
+ add_person_ids = set(target_created_person_ids) | set(target_existing_person_ids) - set(slice_person_ids)
+ keep_person_ids = set(target_existing_person_ids) & set(slice_person_ids)
+ del_person_ids = set(slice_person_ids) - set(target_existing_person_ids)
+
+ # delete
+ for person_id in del_person_ids:
+ self.driver.shell.DeletePersonFromSlice (person_id, slice_id)
+
+ # about the last 2 sets, for managing keys, we need to trace back person_id -> user
+ # and for this we need all the Person objects; we already have the target_existing ones
+ # also we avoid issuing a call if possible
+ target_created_persons = [] if not target_created_person_ids \
+ else self.driver.shell.GetPersons \
+ ({'peer_id':None, 'person_id':target_created_person_ids}, person_fields)
+ persons_by_person_id = { person['person_id'] : person \
+ for person in target_existing_persons + target_created_persons }
+
+ def user_by_person_id (person_id):
+ person = persons_by_person_id [person_id]
+ hrn = person ['hrn']
+ return users_by_hrn [hrn]
+
persons_to_verify_keys = {}
-
- # Add persons or add persons to slice
- for person_hrn in persons_to_add:
- person_email = users_by_hrn[person_hrn].get('email', "%s@geni.net"%person_hrn.split('.')[-1])
- if person_email and person_email in persons_by_email.keys():
- # check if the user already exist in PL
- person_id = persons_by_email[person_email]['person_id']
- self.driver.shell.AddPersonToSlice(person_id, slice['slice_id'])
- persons_to_verify_keys[person_id] = users_by_hrn[person_hrn]
-
- else:
- person = {
- 'first_name': person_hrn,
- 'last_name': person_hrn,
- 'email': users_by_hrn[person_hrn].get('email', "%s@geni.net"%person_hrn.split('.')[-1]),
- }
-
- person_id = self.driver.shell.AddPerson(person)
- self.driver.shell.AddRoleToPerson('user', int(person_id))
- # enable the account
- self.driver.shell.UpdatePerson(int(person_id), {'enabled': True})
- self.driver.shell.SetPersonSfaCreated(int(person_id), 'True')
- self.driver.shell.AddPersonToSite(int(person_id), site['site_id'])
- self.driver.shell.AddPersonToSlice(int(person_id), slice['slice_id'])
- self.driver.shell.SetPersonHrn(int(person_id), person_hrn)
-
- # Add keys
- for key in users_by_hrn[person_hrn].get('keys', []):
- key = {'key':key, 'key_type':'ssh'}
- self.driver.shell.AddPersonKey(person_id, key)
-
-
- # Delete persons from slice
- for person_hrn in persons_to_delete:
- person_id = slice_persons_by_hrn[person_hrn].get('person_id')
- slice_id = slice['slice_id']
- self.driver.shell.DeletePersonFromSlice(person_id, slice_id)
-
-
+ # add
+ for person_id in add_person_ids:
+ self.driver.shell.AddPersonToSlice(person_id, slice_id)
+ persons_to_verify_keys[person_id] = user_by_person_id(person_id)
# Update kept persons
- for person_hrn in persons_to_keep:
- person_id = slice_persons_by_hrn[person_hrn].get('person_id')
- persons_to_verify_keys[person_id] = users_by_hrn[person_hrn]
-
+ for person_id in keep_person_ids:
+ persons_to_verify_keys[person_id] = user_by_person_id(person_id)
self.verify_keys(persons_to_verify_keys, options)
- return persons_to_add
+ # return hrns of the newly added persons
+ return [ persons_by_person_id[person_id]['hrn'] for person_id in add_person_ids ]
- def verify_keys(self, persons_to_verify_keys, options={}):
+ def verify_keys(self, persons_to_verify_keys, options=None):
+ if options is None: options={}
# we only add keys that comes from sfa to persons in PL
for person_id in persons_to_verify_keys:
person_sfa_keys = persons_to_verify_keys[person_id].get('keys', [])
self.driver.shell.AddPersonKey(int(person_id), key)
- def verify_slice_attributes(self, slice, requested_slice_attributes, options={}, admin=False):
- append = options.get('append', True)
- # get list of attributes users ar able to manage
+ def verify_slice_tags(self, slice, requested_slice_attributes, options=None, admin=False):
+ """
+ This function deals with slice tags, and supports 3 modes described
+ in the 'pltags' option that can be either
+ (*) 'ignore' (default) - do nothing
+ (*) 'append' - only add incoming tags, that do not match an existing tag
+ (*) 'sync' - tries to do the plain wholesale thing,
+ i.e. to leave the db in sync with incoming tags
+ """
+ if options is None: options={}
+
+ # lookup 'pltags' in options to find out which mode is requested here
+ pltags = options.get('pltags', 'ignore')
+ # make sure the default is 'ignore'
+ if pltags not in ('ignore', 'append', 'sync'):
+ pltags = 'ignore'
+
+ if pltags == 'ignore':
+ logger.info('verify_slice_tags in ignore mode - leaving slice tags as-is')
+ return
+
+ # incoming data (attributes) have a (name, value) pair
+ # while PLC data (tags) have a (tagname, value) pair
+ # we must be careful not to mix these up
+
+ # get list of tags users are able to manage - based on category
filter = {'category': '*slice*'}
if not admin:
filter['|roles'] = ['user']
- slice_attributes = self.driver.shell.GetTagTypes(filter)
- valid_slice_attribute_names = [attribute['tagname'] for attribute in slice_attributes]
+ valid_tag_types = self.driver.shell.GetTagTypes(filter)
+ valid_tag_names = [ tag_type['tagname'] for tag_type in valid_tag_types ]
+ logger.debug("verify_slice_attributes: valid names={}".format(valid_tag_names))
- # get sliver attributes
- added_slice_attributes = []
- removed_slice_attributes = []
+ # get slice tags
+ slice_attributes_to_add = []
+ slice_tags_to_remove = []
# we need to keep the slice hrn anyway
- ignored_slice_attribute_names = ['hrn']
- existing_slice_attributes = self.driver.shell.GetSliceTags({'slice_id': slice['slice_id']})
+ ignored_slice_tag_names = ['hrn']
+ existing_slice_tags = self.driver.shell.GetSliceTags({'slice_id': slice['slice_id']})
- # get attributes that should be removed
- for slice_tag in existing_slice_attributes:
- if slice_tag['tagname'] in ignored_slice_attribute_names:
+ # get tags that should be removed
+ for slice_tag in existing_slice_tags:
+ if slice_tag['tagname'] in ignored_slice_tag_names:
# If a slice already has a admin only role it was probably given to them by an
# admin, so we should ignore it.
- ignored_slice_attribute_names.append(slice_tag['tagname'])
- attribute_found=True
+ ignored_slice_tag_names.append(slice_tag['tagname'])
+ tag_found = True
else:
- # If an existing slice attribute was not found in the request it should
+ # If an existing slice tag was not found in the request it should
# be removed
- attribute_found=False
+ tag_found = False
for requested_attribute in requested_slice_attributes:
if requested_attribute['name'] == slice_tag['tagname'] and \
requested_attribute['value'] == slice_tag['value']:
- attribute_found=True
+ tag_found = True
break
+ # remove tags only if not in append mode
+ if not tag_found and pltags != 'append':
+ slice_tags_to_remove.append(slice_tag)
- if not attribute_found and not append:
- removed_slice_attributes.append(slice_tag)
-
- # get attributes that should be added:
+ # get tags that should be added:
for requested_attribute in requested_slice_attributes:
# if the requested attribute wasn't found we should add it
- if requested_attribute['name'] in valid_slice_attribute_names:
- attribute_found = False
- for existing_attribute in existing_slice_attributes:
+ if requested_attribute['name'] in valid_tag_names:
+ tag_found = False
+ for existing_attribute in existing_slice_tags:
if requested_attribute['name'] == existing_attribute['tagname'] and \
requested_attribute['value'] == existing_attribute['value']:
- attribute_found=True
+ tag_found = True
break
- if not attribute_found:
- added_slice_attributes.append(requested_attribute)
-
-
- # remove stale attributes
- for attribute in removed_slice_attributes:
+ if not tag_found:
+ slice_attributes_to_add.append(requested_attribute)
+
+ def friendly_message (tag_or_att):
+ name = tag_or_att['tagname'] if 'tagname' in tag_or_att else tag_or_att['name']
+ return "SliceTag slice={}, tagname={} value={}, node_id={}"\
+ .format(slice['name'], tag_or_att['name'], tag_or_att['value'], tag_or_att.get('node_id'))
+
+ # remove stale tags
+ for tag in slice_tags_to_remove:
try:
- self.driver.shell.DeleteSliceTag(attribute['slice_tag_id'])
- except Exception, e:
- logger.warn('Failed to remove sliver attribute. name: %s, value: %s, node_id: %s\nCause:%s'\
- % (slice['name'], attribute['value'], attribute.get('node_id'), str(e)))
-
- # add requested_attributes
- for attribute in added_slice_attributes:
+ logger.info("Removing Slice Tag {}".format(friendly_message(tag)))
+ self.driver.shell.DeleteSliceTag(tag['slice_tag_id'])
+ except Exception as e:
+ logger.warn("Failed to remove slice tag {}\nCause:{}"\
+ .format(friendly_message(tag), e))
+
+ # add requested_tags
+ for attribute in slice_attributes_to_add:
try:
- self.driver.shell.AddSliceTag(slice['name'], attribute['name'], attribute['value'], attribute.get('node_id', None))
- except Exception, e:
- logger.warn('Failed to add sliver attribute. name: %s, value: %s, node_id: %s\nCause:%s'\
- % (slice['name'], attribute['value'], attribute.get('node_id'), str(e)))
-
+ logger.info("Adding Slice Tag {}".format(friendly_message(attribute)))
+ self.driver.shell.AddSliceTag(slice['name'], attribute['name'],
+ attribute['value'], attribute.get('node_id', None))
+ except Exception as e:
+ logger.warn("Failed to add slice tag {}\nCause:{}"\
+ .format(friendly_message(attribute), e))