def resolve_location(self, ctx, arg): log_util.setLoggerFilter(LOG, ctx.get('keyspace'), ctx.get('plan_id')) error = False resolved_location = None host_name = arg.get('host_name') clli_code = arg.get('clli_code') if host_name: results = self.ip_ext_manager.map_method('resolve_host_location', host_name) elif clli_code: results = self.ip_ext_manager.map_method('resolve_clli_location', clli_code) else: results = None # unknown location response LOG.error( _LE("Unknown location type from the input template." "Expected location types are host_name" " or clli_code.")) if results and len(results) > 0: resolved_location = results[0] else: error = True return { 'response': { 'resolved_location': resolved_location }, 'error': error }
def resolve_demands(self, ctx, arg): log_util.setLoggerFilter(LOG, ctx.get('keyspace'), ctx.get('plan_id')) error = False demands = arg.get('demands') plan_info = arg.get('plan_info') triage_translator_data = arg.get('triage_translator_data') resolved_demands = None results = self.ip_ext_manager.map_method('resolve_demands', demands, plan_info, triage_translator_data) if results and len(results) > 0: resolved_demands = results[0] if self.triage_data_trans['plan_id'] == None: self.triage_data_trans['plan_name'] = triage_translator_data[ 'plan_name'] self.triage_data_trans['plan_id'] = triage_translator_data[ 'plan_id'] self.triage_data_trans['translator_triage'].append( triage_translator_data['dropped_candidates']) elif (not self.triage_data_trans['plan_id'] == triage_translator_data['plan_id']): self.triage_data_trans = { 'plan_id': None, 'plan_name': None, 'translator_triage': [] } self.triage_data_trans['plan_name'] = triage_translator_data[ 'plan_name'] self.triage_data_trans['plan_id'] = triage_translator_data[ 'plan_id'] self.triage_data_trans['translator_triage'].append( triage_translator_data['dropped_candidates']) else: self.triage_data_trans['translator_triage'].append( triage_translator_data['dropped_candidates']) else: error = True return { 'response': { 'resolved_demands': resolved_demands, 'trans': self.triage_data_trans }, 'error': error }
def run(self): """Run""" LOG.debug("%s" % self.__class__.__name__) # TODO(snarayanan): This is really meant to be a control loop # As long as self.running is true, we process another request. while self.running: # Delay time (Seconds) for MUSIC requests. time.sleep(self.conf.delay_time) # plans = Plan.query().all() # Find the first plan with a status of SOLVED. # Change its status to RESERVING. solution = None translation = None p = None # requests_to_reserve = dict() # Instead of using the query.all() method, now creating an index for 'status' # field in conductor.plans table, and query plans by status columns solved_plans = self.Plan.query.get_plan_by_col( "status", self.Plan.SOLVED) reserving_plans = self.Plan.query.get_plan_by_col( "status", self.Plan.RESERVING) # combine the plans with status = 'solved' and 'reserving' together plans = solved_plans + reserving_plans found_solved_template = False for p in plans: # when a plan is in RESERVING status more than timeout value if p.status == self.Plan.RESERVING and \ (self.current_time_seconds() - self.millisec_to_sec(p.updated)) > self.conf.reservation.timeout: # change the plan status to SOLVED for another VM to reserve p.status = self.Plan.SOLVED p.update(condition=self.reservating_status_condition) break elif p.status == self.Plan.SOLVED: solution = p.solution translation = p.translation found_solved_template = True break if not solution: if found_solved_template: message = _LE("Plan {} status is solved, yet " "the solution wasn't found").format(p.id) LOG.error(message) p.status = self.Plan.ERROR p.message = message p.update(condition=self.solved_status_condition) continue if found_solved_template and p and p.reservation_counter >= self.conf.reservation.max_reservation_counter: message = _LE("Tried {} times. Plan {} is unable to reserve") \ .format(self.conf.reservation.max_reservation_counter, p.id) LOG.error(message) p.status = self.Plan.ERROR p.message = message p.update(condition=self.solved_status_condition) continue log_util.setLoggerFilter(LOG, self.conf.keyspace, p.id) # update status to reserving p.status = self.Plan.RESERVING p.reservation_counter += 1 p.reservation_owner = socket.gethostname() _is_updated = p.update(condition=self.solved_status_condition) if not _is_updated: continue if 'FAILURE' in _is_updated: continue LOG.info( _LI("Reservation starts, changing the template status from solved to reserving, " "atomic update response from MUSIC {}").format( _is_updated)) LOG.info( _LI("Plan {} with request id {} is reserving by machine {}. Tried to reserve it for {} times." ).format(p.id, p.name, p.reservation_owner, p.reservation_counter)) # begin reservations # if plan needs reservation proceed with reservation # else set status to done. reservations = None _is_success = "FAILURE" if translation: conductor_solver = translation.get("conductor_solver") if conductor_solver: reservations = conductor_solver.get("reservations") else: LOG.error("no conductor_solver in " "translation for Plan {}".format(p.id)) if reservations: recommendations = solution.get("recommendations") reservation_list = list() # TODO(larry) combine the two reservation logic as one, make the code service independent sdwan_candidate_list = list() service_model = reservations.get("service_model") for reservation, resource in reservations.get("demands", {}).items(): candidates = list() reservation_demand = resource.get("demands") reservation_name = resource.get("name") reservation_type = resource.get("type") reservation_properties = resource.get("properties") if reservation_properties: controller = reservation_properties.get("controller") request = reservation_properties.get("request") for recommendation in recommendations: for demand, r_resource in recommendation.items(): if demand == reservation_demand: # get selected candidate from translation selected_candidate_id = r_resource.get( "candidate").get("candidate_id") demands = translation.get( "conductor_solver").get("demands") for demand_name, d_resource in demands.items(): if demand_name == demand: for candidate in d_resource.get( "candidates"): if candidate.get( "candidate_id" ) == selected_candidate_id: candidate['request'] = request candidates.append(candidate) sdwan_candidate_list.append( candidate) #TODO(larry) combine the two reservation logic as one, make the code service independent if service_model == "ADIOD": is_success = self.try_reservation_call( method="reserve", candidate_list=candidates, reservation_type=service_model, controller=controller, request=request, reservation_name=None) # if reservation succeed continue with next candidate if is_success: curr_reservation = dict() curr_reservation['candidate_list'] = candidates curr_reservation[ 'reservation_name'] = reservation_name curr_reservation[ 'reservation_type'] = reservation_type curr_reservation['controller'] = controller curr_reservation['request'] = request reservation_list.append(curr_reservation) else: # begin roll back of all reserved resources on # the first failed reservation rollback_status = \ self.rollback_reservation(reservation_list) # order_lock spin-up rollback for decision in solution.get('recommendations'): candidate = list(decision.values())[0].get( 'candidate' ) # Python 3 Conversion -- dict object to list object if candidate.get('inventory_type') == 'cloud': # TODO(larry) change the code to get('conflict_id') instead of 'location_id' conflict_id = candidate.get('conflict_id') order_record = self.OrderLock.query.get_plan_by_col( "id", conflict_id)[0] if order_record: order_record.delete() # statuses if rollback_status: # released all reservations, # move plan to translated if p.reservation_counter >= self.conf.reservation.max_reservation_counter: p.status = self.Plan.ERROR p.message = _LE( "Tried {} times. Plan {} is unable to reserve" ).format( self.conf.reservation. max_reservation_counter, p.id) LOG.error(p.message) else: p.status = self.Plan.TRANSLATED # TODO(larry): Should be replaced by the new api from MUSIC while 'FAILURE' in _is_success: _is_success = p.update( condition=self. reservation_owner_condition) LOG.info( _LI("Rolling back the template from reserving to {} status, " "atomic update response from MUSIC {}" ).format(p.status, _is_success)) del reservation_list[:] else: LOG.error("Reservation rollback failed") p.status = self.Plan.ERROR p.message = "Reservation release failed" # TODO(larry): Should be replaced by the new api from MUSIC while 'FAILURE' in _is_success: _is_success = p.update( condition=self. reservation_owner_condition) LOG.info( _LI("Rollback Failed, Changing the template status from reserving to error, " "atomic update response from MUSIC {}" ).format(_is_success)) break # reservation failed continue # continue with reserving the next candidate # TODO(larry) combine the two reservation logic as one, make the code service independent if service_model == "DHV": is_success = self.try_reservation_call( method="reserve", candidate_list=sdwan_candidate_list, reservation_type=service_model, controller=controller, request=request, reservation_name=None) if not is_success: # order_lock spin-up rollback for decision in solution.get('recommendations'): candidate = list(decision.values())[0].get( 'candidate' ) # Python 3 Conversion -- dict object to list object if candidate.get('inventory_type') == 'cloud': conflict_id = candidate.get('conflict_id') order_record = self.OrderLock.query.get_plan_by_col( "id", conflict_id)[0] if order_record: order_record.delete() if p.reservation_counter >= self.conf.reservation.max_reservation_counter: p.status = self.Plan.ERROR p.message = _LE( "Tried {} times. Plan {} is unable to reserve" ).format( self.conf.reservation.max_reservation_counter, p.id) LOG.error(p.message) else: p.status = self.Plan.TRANSLATED # TODO(larry): Should be replaced by the new api from MUSIC while 'FAILURE' in _is_success: _is_success = p.update( condition=self.reservation_owner_condition) LOG.info( _LI("Rolling back the template from reserving to {} status, " "atomic update response from MUSIC {}"). format(p.status, _is_success)) del reservation_list[:] # verify if all candidates have been reserved if p.status == self.Plan.RESERVING: # all reservations succeeded. LOG.info( _LI("Plan {} with request id {} Reservation complete"). format(p.id, p.name)) LOG.debug("Plan {} Reservation complete".format(p.id)) p.status = self.Plan.DONE while 'FAILURE' in _is_success and ( self.current_time_seconds() - self.millisec_to_sec( p.updated)) <= self.conf.reservation.timeout: _is_success = p.update( condition=self.reservation_owner_condition) LOG.info( _LI("Reservation is complete, changing the template status from reserving to done, " "atomic update response from MUSIC {}").format( _is_success)) continue
def run(self): """Run""" LOG.debug("{}".format(self.__class__.__name__)) # TODO(snarayanan): This is really meant to be a control loop # As long as self.running is true, we process another request. while self.running: # Delay time (Seconds) for MUSIC requests. time.sleep(self.conf.delay_time) # plans = Plan.query().all() # Find the first plan with a status of TRANSLATED. # Change its status to SOLVING. # Then, read the "translated" field as "template". json_template = None p = None requests_to_solve = dict() regions_maps = dict() country_groups = list() # Instead of using the query.all() method, now creating an index for 'status' # field in conductor.plans table, and query plans by status columns translated_plans = self.Plan.query.get_plan_by_col("status", self.Plan.TRANSLATED) solving_plans = self.Plan.query.get_plan_by_col("status", self.Plan.SOLVING) # combine the plans with status = 'translated' and 'solving' together plans = translated_plans + solving_plans found_translated_template = False for p in plans: if p.status == self.Plan.TRANSLATED: json_template = p.translation found_translated_template = True break elif p.status == self.Plan.SOLVING and (self.current_time_seconds() - self.millisec_to_sec(p.updated)) > self.conf.solver.timeout: p.status = self.Plan.TRANSLATED p.update(condition=self.solving_status_condition) break if not json_template: if found_translated_template: message = _LE("Plan {} status is translated, yet " "the translation wasn't found").format(p.id) LOG.error(message) p.status = self.Plan.ERROR p.message = message p.update(condition=self.translated_status_condition) continue if found_translated_template and p and p.solver_counter >= self.conf.solver.max_solver_counter: message = _LE("Tried {} times. Plan {} is unable to solve").format(self.conf.solver.max_solver_counter, p.id) LOG.error(message) p.status = self.Plan.ERROR p.message = message p.update(condition=self.translated_status_condition) continue log_util.setLoggerFilter(LOG, self.conf.keyspace, p.id) p.status = self.Plan.SOLVING p.solver_counter += 1 p.solver_owner = socket.gethostname() _is_updated = p.update(condition=self.translated_status_condition) if not _is_updated: continue # other VMs have updated the status and start solving the plan if 'FAILURE' in _is_updated: continue LOG.info(_LI("Sovling starts, changing the template status from translated to solving, " "atomic update response from MUSIC {}").format(_is_updated)) LOG.info(_LI("Plan {} with request id {} is solving by machine {}. Tried to solve it for {} times."). format(p.id, p.name, p.solver_owner, p.solver_counter)) _is_success = "FAILURE" request = parser.Parser() request.cei = self.cei request.request_id = p.name request.plan_id = p.id # getting the number of solutions need to provide num_solution = getattr(p, 'recommend_max', '1') if num_solution.isdigit(): num_solution = int(num_solution) # TODO(inam/larry): move this part of logic inside of parser and don't apply it to distance_between try: # getting region placeholders from database and insert/put into regions_maps dictionary region_placeholders = self.RegionPlaceholders.query.all() for region in region_placeholders: regions_maps.update(region.countries) # getting country groups from database and insert into the country_groups list customer_loc = '' location_list = json_template["conductor_solver"]["locations"] for location_id, location_info in location_list.items(): customer_loc = location_info['country'] countries = self.CountryLatency.query.get_plan_by_col("country_name", customer_loc) LOG.info("Customer Location for Latency Reduction " + customer_loc) if len(countries) == 0: LOG.info("country is not present is country latency table, looking for * wildcard entry") countries = self.CountryLatency.query.get_plan_by_col("country_name", "*") if len(countries) != 0: LOG.info("Found '*' wild card entry in country latency table") else: msg = "No '*' wild card entry found in country latency table. No solution will be provided" LOG.info(msg) p.message = msg for country in countries: country_groups = country.groups LOG.info("Done getting Latency Country DB Groups ") except Exception as error_msg: LOG.error("Exception thrown while reading region_placeholders and country groups information " "from database. Exception message: {}".format(error_msg)) try: request.parse_template(json_template, country_groups, regions_maps) request.assgin_constraints_to_demands() requests_to_solve[p.id] = request opt = optimizer.Optimizer(self.conf, _requests=requests_to_solve) solution_list = opt.get_solution(num_solution) except Exception as err: message = _LE("Plan {} status encountered a " "parsing error: {}").format(p.id, err) LOG.error(traceback.print_exc()) p.status = self.Plan.ERROR p.message = message while 'FAILURE' in _is_success: _is_success = p.update(condition=self.solver_owner_condition) LOG.info(_LI("Encountered a parsing error, changing the template status from solving to error, " "atomic update response from MUSIC {}").format(_is_success)) continue LOG.info("Preparing the recommendations ") # checking if the order is 'initial' or 'speed changed' one is_speed_change = False if request and request.request_type == 'speed changed': is_speed_change = True recommendations = [] if not solution_list or len(solution_list) < 1: # when order takes too much time to solve if (int(round(time.time())) - self.millisec_to_sec(p.updated)) > self.conf.solver.solver_timeout: message = _LI("Plan {} is timed out, exceed the expected " "time {} seconds").format(p.id, self.conf.solver.timeout) # when no solution found else: message = _LI("Plan {} search failed, no " "recommendations found by machine {}").format(p.id, p.solver_owner) LOG.info(message) # Update the plan status p.status = self.Plan.NOT_FOUND p.message = message # Metrics to Prometheus m_svc_name = p.template.get('parameters', {}).get('service_name', 'N/A') PC.VNF_FAILURE.labels('ONAP', m_svc_name).inc() while 'FAILURE' in _is_success: _is_success = p.update(condition=self.solver_owner_condition) LOG.info(_LI("Plan serach failed, changing the template status from solving to not found, " "atomic update response from MUSIC {}").format(_is_success)) else: # Assemble recommendation result JSON for solution in solution_list: current_rec = dict() for demand_name in solution: resource = solution[demand_name] if not is_speed_change: is_rehome = "false" else: is_rehome = "false" if resource.get("existing_placement") == 'true' else "true" location_id = "" if resource.get("cloud_region_version") == '2.5' \ else resource.get("location_id") rec = { # FIXME(shankar) A&AI must not be hardcoded here. # Also, account for more than one Inventory Provider. "inventory_provider": "aai", "service_resource_id": resource.get("service_resource_id"), "candidate": { "candidate_id": resource.get("candidate_id"), "inventory_type": resource.get("inventory_type"), "cloud_owner": resource.get("cloud_owner"), "location_type": resource.get("location_type"), "location_id": location_id, "is_rehome": is_rehome}, "attributes": { "physical-location-id": resource.get("physical_location_id"), "cloud_owner": resource.get("cloud_owner"), 'aic_version': resource.get("cloud_region_version")}, } if rec["candidate"]["inventory_type"] in ["nssi", "nsi", "slice_profiles"]: rec["candidate"] = resource if resource.get('vim-id'): rec["candidate"]['vim-id'] = resource.get('vim-id') if rec["candidate"]["inventory_type"] == "service": rec["attributes"]["host_id"] = resource.get("host_id") rec["attributes"]["service_instance_id"] = resource.get("candidate_id") rec["candidate"]["host_id"] = resource.get("host_id") if resource.get('vlan_key'): rec["attributes"]['vlan_key'] = resource.get('vlan_key') if resource.get('port_key'): rec["attributes"]['port_key'] = resource.get('port_key') if rec["candidate"]["inventory_type"] == "vfmodule": rec["attributes"]["host_id"] = resource.get("host_id") rec["attributes"]["service_instance_id"] = resource.get("service_instance_id") rec["candidate"]["host_id"] = resource.get("host_id") if resource.get('vlan_key'): rec["attributes"]['vlan_key'] = resource.get('vlan_key') if resource.get('port_key'): rec["attributes"]['port_key'] = resource.get('port_key') vf_module_data = rec["attributes"] vf_module_data['nf-name'] = resource.get("nf-name") vf_module_data['nf-id'] = resource.get("nf-id") vf_module_data['nf-type'] = resource.get("nf-type") vf_module_data['vnf-type'] = resource.get("vnf-type") vf_module_data['vf-module-id'] = resource.get("vf-module-id") vf_module_data['vf-module-name'] = resource.get("vf-module-name") vf_module_data['ipv4-oam-address'] = resource.get("ipv4-oam-address") vf_module_data['ipv6-oam-address'] = resource.get("ipv6-oam-address") vf_module_data['vservers'] = resource.get("vservers") elif rec["candidate"]["inventory_type"] == "cloud": if resource.get("all_directives") and resource.get("flavor_map"): rec["attributes"]["directives"] = \ self.set_flavor_in_flavor_directives( resource.get("flavor_map"), resource.get("all_directives")) # Metrics to Prometheus m_vim_id = resource.get("vim-id") m_hpa_score = resource.get("hpa_score", 0) m_svc_name = p.template['parameters'].get( 'service_name', 'N/A') for vnfc, flavor in resource.get("flavor_map").items(): PC.VNF_COMPUTE_PROFILES.labels('ONAP', m_svc_name, demand_name, vnfc, flavor, m_vim_id).inc() PC.VNF_SCORE.labels('ONAP', m_svc_name, demand_name, m_hpa_score).inc() if resource.get('conflict_id'): rec["candidate"]["conflict_id"] = resource.get("conflict_id") if resource.get('passthrough_attributes'): for key, value in resource.get('passthrough_attributes').items(): if key in rec["attributes"]: LOG.error('Passthrough attribute {} in demand {} already exist for candidate {}'. format(key, demand_name, rec['candidate_id'])) else: rec["attributes"][key] = value # TODO(snarayanan): Add total value to recommendations? # msg = "--- total value of decision = {}" # LOG.debug(msg.format(_best_path.total_value)) # msg = "--- total cost of decision = {}" # LOG.debug(msg.format(_best_path.total_cost)) current_rec[demand_name] = rec recommendations.append(current_rec) # Update the plan with the solution p.solution = { "recommendations": recommendations } # multiple spin-ups logic ''' go through list of recommendations in the solution for cloud candidates, check if (cloud-region-id + e2evnfkey) is in the order_locks table if so, insert the row with status 'parked' in order_locks, changes plan status to 'pending' in plans table (or other status value) otherwise, insert the row with status 'locked' in order_locks, and change status to 'solved' in plans table - continue reservation ''' # clean up the data/record in order_locks table, deleting all records that failed from MSO order_locks = self.OrderLock.query.all() for order_lock_record in order_locks: plans = getattr(order_lock_record, 'plans') for plan_id, plan_attributes in plans.items(): plan_dict = json.loads(plan_attributes) if plan_dict.get('status', None) == OrderLock.FAILED: order_lock_record.delete() LOG.info(_LI("The order lock record {} with status {} is deleted (due to failure" " spinup from MSO) from order_locks table"). format(order_lock_record, plan_dict.get('status'))) break inserted_order_records_dict = dict() available_dependenies_set = set() is_inserted_to_order_locks = True is_conflict_id_missing = False is_order_translated_before_spinup = False for solution in solution_list: for demand_name, candidate in solution.items(): if candidate.get('inventory_type') == 'cloud': conflict_id = candidate.get('conflict_id') service_resource_id = candidate.get('service_resource_id') # TODO(larry): add more logic for missing conflict_id in template if not conflict_id: is_conflict_id_missing = True break available_dependenies_set.add(conflict_id) # check if conflict_id exists in order_locks table order_lock_record = self.OrderLock.query.get_plan_by_col("id", conflict_id) if order_lock_record: is_spinup_completed = getattr(order_lock_record[0], 'is_spinup_completed') spinup_completed_timestamp = getattr(order_lock_record[0], 'spinup_completed_timestamp') if is_spinup_completed and spinup_completed_timestamp > p.translation_begin_timestamp: is_order_translated_before_spinup = True break elif not is_spinup_completed: inserted_order_records_dict[conflict_id] = service_resource_id if is_conflict_id_missing: message = _LE("Missing conflict identifier field for cloud candidates in the template, " "could not insert into order_locks table") LOG.debug(message) p.status = self.Plan.SOLVED elif is_order_translated_before_spinup: message = _LE("Retriggering Plan {} due to the new order arrives before the " "spinup completion of the old order ").format(p.id) LOG.debug(message) p.rehome_plan() elif len(inserted_order_records_dict) > 0: new_dependenies_set = available_dependenies_set - set(inserted_order_records_dict.keys()) dependencies = ','.join(str(s) for s in new_dependenies_set) for conflict_id, service_resource_id in inserted_order_records_dict.items(): plan = { p.id: { "status": OrderLock.UNDER_SPIN_UP, "created": self.current_time_millis(), "updated": self.current_time_millis(), "service_resource_id": service_resource_id } } if dependencies: plan[p.id]['dependencies'] = dependencies order_lock_row = self.OrderLock(id=conflict_id, plans=plan) response = order_lock_row.insert() # TODO(larry): add more logs for inserting order lock record (insert/update) LOG.info(_LI("Inserting the order lock record to order_locks table in MUSIC, " "conditional insert operation response from MUSIC {}").format(response)) if response and response.status_code == 200: body = response.json() LOG.info("Succcessfully inserted the record in order_locks table with " "the following response message {}".format(body)) else: is_inserted_to_order_locks = False else: for solution in solution_list: for demand_name, candidate in solution.items(): if candidate.get('inventory_type') == 'cloud': conflict_id = candidate.get('conflict_id') service_resource_id = candidate.get('service_resource_id') order_lock_record = self.OrderLock.query.get_plan_by_col("id", conflict_id) if order_lock_record: deleting_record = order_lock_record[0] plans = getattr(deleting_record, 'plans') is_spinup_completed = getattr(deleting_record, 'is_spinup_completed') spinup_completed_timestamp = getattr(deleting_record, 'spinup_completed_timestamp') if is_spinup_completed: # persist the record in order_locks_history table order_lock_history_record = \ self.OrderLockHistory(conflict_id=conflict_id, plans=plans, is_spinup_completed=is_spinup_completed, spinup_completed_timestamp=spinup_completed_timestamp ) LOG.debug("Inserting the history record with conflict id {}" " to order_locks_history table".format(conflict_id)) order_lock_history_record.insert() # remove the older record LOG.debug("Deleting the order lock record {} from order_locks table" .format(deleting_record)) deleting_record.delete() plan = { p.id: { "status": OrderLock.UNDER_SPIN_UP, "created": self.current_time_millis(), "updated": self.current_time_millis(), "service_resource_id": service_resource_id } } order_lock_row = self.OrderLock(id=conflict_id, plans=plan) response = order_lock_row.insert() # TODO(larry): add more logs for inserting order lock record (insert/update) LOG.info(_LI("Inserting the order lock record to order_locks table in MUSIC, " "conditional insert operation response from MUSIC {}").format(response)) if response and response.status_code == 200: body = response.json() LOG.info("Succcessfully inserted the record in order_locks table " "with the following response message {}".format(body)) else: is_inserted_to_order_locks = False if not is_inserted_to_order_locks: message = _LE("Plan {} status encountered an " "error while inserting order lock message to MUSIC.").format(p.id) LOG.error(message) p.status = self.Plan.ERROR p.message = message elif p.status == self.Plan.SOLVING: if len(inserted_order_records_dict) > 0: LOG.info(_LI("The plan with id {} is parked in order_locks table," "waiting for MSO release calls").format(p.id)) p.status = self.Plan.WAITING_SPINUP else: LOG.info(_LI("The plan with id {} is inserted in order_locks table."). format(p.id)) p.status = self.Plan.SOLVED while 'FAILURE' in _is_success \ and (self.current_time_seconds() - self.millisec_to_sec(p.updated)) <= self.conf.solver.timeout: _is_success = p.update(condition=self.solver_owner_condition) LOG.info(_LI("Plan search complete, changing the template status from solving to {}, " "atomic update response from MUSIC {}").format(p.status, _is_success)) LOG.info(_LI("Plan {} search complete, {} solution(s) found by machine {}"). format(p.id, len(solution_list), p.solver_owner)) LOG.debug("Plan {} detailed solution: {}". format(p.id, p.solution)) LOG.info("Plan name: {}".format(p.name))
def __check_for_templates(self): """Wait for the polling interval, then do the real template check.""" # Wait for at least poll_interval sec polling_interval = self.conf.controller.polling_interval time.sleep(polling_interval) # Look for plans with the status set to TEMPLATE # Instead of using the query.all() method, now creating an index for 'status' # field in conductor.plans table, and query plans by status columns template_plans = self.Plan.query.get_plan_by_col( "status", self.Plan.TEMPLATE) translating_plans = self.Plan.query.get_plan_by_col( "status", self.Plan.TRANSLATING) # combine the plans with status = 'template' and 'translating' together plans = template_plans + translating_plans for plan in plans: # If there's a template to be translated, do it! if plan.status == self.Plan.TEMPLATE: if plan.translation_counter >= self.conf.controller.max_translation_counter: message = _LE("Tried {} times. Plan {} is unable to translate") \ .format(self.conf.controller.max_translation_counter, plan.id) plan.message = message plan.status = self.Plan.ERROR plan.update(condition=self.template_status_condition) LOG.error(message) break else: # change the plan status to "translating" and assign the current machine as translation owner plan.status = self.Plan.TRANSLATING plan.translation_counter += 1 plan.translation_owner = socket.gethostname() plan.translation_begin_timestamp = int( round(time.time() * 1000)) _is_updated = plan.update( condition=self.template_status_condition) log_util.setLoggerFilter(LOG, self.conf.keyspace, plan.id) LOG.info( _LE("Plan {} is trying to update the status from 'template' to 'translating'," " get {} response from MUSIC").format( plan.id, _is_updated)) if not _is_updated: break if _is_updated and 'SUCCESS' in _is_updated: self.translate(plan) break # TODO(larry): sychronized clock among Conducotr VMs, or use an offset elif plan.status == self.Plan.TRANSLATING and (self.current_time_seconds() - self.millisec_to_sec(plan.updated)) \ > self.conf.messaging_server.timeout: plan.status = self.Plan.TEMPLATE plan.update(condition=self.translating_status_condition) break elif plan.timedout: # TODO(jdandrea): How to tell all involved to stop working? # Not enough to just set status. continue