From: quilesj Date: Thu, 12 Dec 2019 16:08:35 +0000 (+0000) Subject: Fix bug 970. Changes in NS and operation status X-Git-Tag: v7.0.1rc1~6 X-Git-Url: https://osm.etsi.org/gitweb/?a=commitdiff_plain;h=3655ae05bbe111d9f03c20e1944750cba7a80af6;p=osm%2FLCM.git Fix bug 970. Changes in NS and operation status Change-Id: I0af317244a98d703d0086b652b199d1036cf64e4 Signed-off-by: quilesj --- diff --git a/osm_lcm/lcm_utils.py b/osm_lcm/lcm_utils.py index a83e788..0178921 100644 --- a/osm_lcm/lcm_utils.py +++ b/osm_lcm/lcm_utils.py @@ -433,7 +433,7 @@ class TaskRegistry(LcmBase): q_filter = {'_id': _id} # NS/NSI if self._is_service_type_HA(topic): - update_dict = {'detailed-status': step} + update_dict = {'detailed-status': step, 'queuePosition': new_num_related_tasks} # VIM/WIM/SDN elif self._is_account_type_HA(topic): _, op_index = self._get_account_and_op_HA(op_id) diff --git a/osm_lcm/ns.py b/osm_lcm/ns.py index 18bd6f9..9a001b1 100644 --- a/osm_lcm/ns.py +++ b/osm_lcm/ns.py @@ -121,9 +121,7 @@ class NsLcm(LcmBase): url='{}:{}'.format(self.vca_config['host'], self.vca_config['port']), username=self.vca_config.get('user', None), vca_config=self.vca_config, - on_update_db=self._on_update_n2vc_db, - # ca_cert=self.vca_config.get('cacert'), - # api_proxy=self.vca_config.get('apiproxy'), + on_update_db=self._on_update_n2vc_db ) self.k8sclusterhelm = K8sHelmConnector( @@ -147,32 +145,108 @@ class NsLcm(LcmBase): # create RO client self.RO = ROclient.ROClient(self.loop, **self.ro_config) - def _on_update_n2vc_db(self, table, filter, path, updated_data): + def _on_update_ro_db(self, nsrs_id, ro_descriptor): - self.logger.debug('_on_update_n2vc_db(table={}, filter={}, path={}, updated_data={}' - .format(table, filter, path, updated_data)) + # self.logger.debug('_on_update_ro_db(nsrs_id={}'.format(nsrs_id)) + + try: + # TODO filter RO descriptor fields... + + # write to database + db_dict = dict() + # db_dict['deploymentStatus'] = yaml.dump(ro_descriptor, default_flow_style=False, indent=2) + db_dict['deploymentStatus'] = ro_descriptor + self.update_db_2("nsrs", nsrs_id, db_dict) + + except Exception as e: + self.logger.warn('Cannot write database RO deployment for ns={} -> {}'.format(nsrs_id, e)) + + async def _on_update_n2vc_db(self, table, filter, path, updated_data): + + # self.logger.debug('_on_update_n2vc_db(table={}, filter={}, path={}, updated_data={}' + # .format(table, filter, path, updated_data)) + + try: + + nsr_id = filter.get('_id') + + # read ns record from database + nsr = self.db.get_one(table='nsrs', q_filter=filter) + current_ns_status = nsr.get('nsState') + + # get vca status for NS + # status_dict = await self.n2vc.get_status(namespace='.' + nsr_id, yaml_format=False) + # TEMPORAL + status_dict = str(await self.n2vc.get_status(namespace='.' + nsr_id)) + + # vcaStatus + db_dict = dict() + db_dict['vcaStatus'] = status_dict + + # update configurationStatus for this VCA + try: + vca_index = int(path[path.rfind(".")+1:]) + + vca_list = deep_get(target_dict=nsr, key_list=('_admin', 'deployed', 'VCA')) + vca_status = vca_list[vca_index].get('status') + + configuration_status_list = nsr.get('configurationStatus') + config_status = configuration_status_list[vca_index].get('status') + + if config_status == 'BROKEN' and vca_status != 'failed': + db_dict['configurationStatus'][vca_index] = 'READY' + elif config_status != 'BROKEN' and vca_status == 'failed': + db_dict['configurationStatus'][vca_index] = 'BROKEN' + except Exception as e: + # not update configurationStatus + self.logger.debug('Error updating vca_index (ignore): {}'.format(e)) + + # if nsState = 'READY' check if juju is reporting some error => nsState = 'DEGRADED' + # if nsState = 'DEGRADED' check if all is OK + is_degraded = False + if current_ns_status in ('READY', 'DEGRADED'): + error_description = '' + # check machines + if status_dict.get('machines'): + for machine_id in status_dict.get('machines'): + machine = status_dict.get('machines').get(machine_id) + # check machine agent-status + if machine.get('agent-status'): + s = machine.get('agent-status').get('status') + if s != 'started': + is_degraded = True + error_description += 'machine {} agent-status={} ; '.format(machine_id, s) + # check machine instance status + if machine.get('instance-status'): + s = machine.get('instance-status').get('status') + if s != 'running': + is_degraded = True + error_description += 'machine {} instance-status={} ; '.format(machine_id, s) + # check applications + if status_dict.get('applications'): + for app_id in status_dict.get('applications'): + app = status_dict.get('applications').get(app_id) + # check application status + if app.get('status'): + s = app.get('status').get('status') + if s != 'active': + is_degraded = True + error_description += 'application {} status={} ; '.format(app_id, s) + + if error_description: + db_dict['errorDescription'] = error_description + if current_ns_status == 'READY' and is_degraded: + db_dict['nsState'] = 'DEGRADED' + if current_ns_status == 'DEGRADED' and not is_degraded: + db_dict['nsState'] = 'READY' + + # write to database + self.update_db_2("nsrs", nsr_id, db_dict) + + except Exception as e: + self.logger.warn('Error updating NS state for ns={}: {}'.format(nsr_id, e)) return - # write NS status to database - # try: - # # nsrs_id = filter.get('_id') - # # print(nsrs_id) - # # get ns record - # nsr = self.db.get_one(table=table, q_filter=filter) - # # get VCA deployed list - # vca_list = deep_get(target_dict=nsr, key_list=('_admin', 'deployed', 'VCA')) - # # get RO deployed - # # ro_list = deep_get(target_dict=nsr, key_list=('_admin', 'deployed', 'RO')) - # for vca in vca_list: - # # status = vca.get('status') - # # print(status) - # # detailed_status = vca.get('detailed-status') - # # print(detailed_status) - # # for ro in ro_list: - # # print(ro) - # - # except Exception as e: - # self.logger.error('Error writing NS status to db: {}'.format(e)) def vnfd2RO(self, vnfd, new_id=None, additionalParams=None, nsrId=None): """ @@ -740,7 +814,6 @@ class NsLcm(LcmBase): db_nsr_update["_admin.deployed.RO.vnfd.{}".format(index)] = RO_update db_nsr["_admin"]["deployed"]["RO"]["vnfd"][index] = RO_update self.update_db_2("nsrs", nsr_id, db_nsr_update) - self._on_update_n2vc_db("nsrs", {"_id": nsr_id}, "_admin.deployed", db_nsr_update) # create nsd at RO nsd_ref = nsd["id"] @@ -772,7 +845,6 @@ class NsLcm(LcmBase): db_nsr_update["_admin.deployed.RO.nsd_id"] = RO_nsd_uuid = desc["uuid"] self.logger.debug(logging_text + "nsd={} created at RO. RO_id={}".format(nsd_ref, RO_nsd_uuid)) self.update_db_2("nsrs", nsr_id, db_nsr_update) - self._on_update_n2vc_db("nsrs", {"_id": nsr_id}, "_admin.deployed", db_nsr_update) # Crate ns at RO # if present use it unless in error status @@ -782,6 +854,7 @@ class NsLcm(LcmBase): step = db_nsr_update["_admin.deployed.RO.detailed-status"] = "Looking for existing ns at RO" # self.logger.debug(logging_text + step + " RO_ns_id={}".format(RO_nsr_id)) desc = await self.RO.show("ns", RO_nsr_id) + except ROclient.ROClientException as e: if e.http_code != HTTPStatus.NOT_FOUND: raise @@ -827,36 +900,42 @@ class NsLcm(LcmBase): db_nsr_update["_admin.deployed.RO.nsr_status"] = "BUILD" self.logger.debug(logging_text + "ns created at RO. RO_id={}".format(desc["uuid"])) self.update_db_2("nsrs", nsr_id, db_nsr_update) - self._on_update_n2vc_db("nsrs", {"_id": nsr_id}, "_admin.deployed", db_nsr_update) # wait until NS is ready step = ns_status_detailed = detailed_status = "Waiting VIM to deploy ns. RO_ns_id={}".format(RO_nsr_id) detailed_status_old = None self.logger.debug(logging_text + step) + old_desc = None while time() <= start_deploy + self.total_deploy_timeout: desc = await self.RO.show("ns", RO_nsr_id) - ns_status, ns_status_info = self.RO.check_ns_status(desc) - db_nsr_update["_admin.deployed.RO.nsr_status"] = ns_status - if ns_status == "ERROR": - raise ROclient.ROClientException(ns_status_info) - elif ns_status == "BUILD": - detailed_status = ns_status_detailed + "; {}".format(ns_status_info) - elif ns_status == "ACTIVE": - step = detailed_status = "Waiting for management IP address reported by the VIM. Updating VNFRs" - try: - if vdu_flag: - self.ns_update_vnfr(db_vnfrs, desc) - break - except LcmExceptionNoMgmtIP: - pass - else: - assert False, "ROclient.check_ns_status returns unknown {}".format(ns_status) - if detailed_status != detailed_status_old: - detailed_status_old = db_nsr_update["_admin.deployed.RO.detailed-status"] = detailed_status - self.update_db_2("nsrs", nsr_id, db_nsr_update) - self._on_update_n2vc_db("nsrs", {"_id": nsr_id}, "_admin.deployed", db_nsr_update) - await asyncio.sleep(5, loop=self.loop) + + # deploymentStatus + if desc != old_desc: + # desc has changed => update db + self._on_update_ro_db(nsrs_id=nsr_id, ro_descriptor=desc) + old_desc = desc + + ns_status, ns_status_info = self.RO.check_ns_status(desc) + db_nsr_update["_admin.deployed.RO.nsr_status"] = ns_status + if ns_status == "ERROR": + raise ROclient.ROClientException(ns_status_info) + elif ns_status == "BUILD": + detailed_status = ns_status_detailed + "; {}".format(ns_status_info) + elif ns_status == "ACTIVE": + step = detailed_status = "Waiting for management IP address reported by the VIM. Updating VNFRs" + try: + if vdu_flag: + self.ns_update_vnfr(db_vnfrs, desc) + break + except LcmExceptionNoMgmtIP: + pass + else: + assert False, "ROclient.check_ns_status returns unknown {}".format(ns_status) + if detailed_status != detailed_status_old: + detailed_status_old = db_nsr_update["_admin.deployed.RO.detailed-status"] = detailed_status + self.update_db_2("nsrs", nsr_id, db_nsr_update) + await asyncio.sleep(5, loop=self.loop) else: # total_deploy_timeout raise ROclient.ROClientException("Timeout waiting ns to be ready") @@ -867,7 +946,7 @@ class NsLcm(LcmBase): db_nsr["_admin.deployed.RO.detailed-status"] = "Deployed at VIM" db_nsr_update["_admin.deployed.RO.detailed-status"] = "Deployed at VIM" self.update_db_2("nsrs", nsr_id, db_nsr_update) - self._on_update_n2vc_db("nsrs", {"_id": nsr_id}, "_admin.deployed", db_nsr_update) + # await self._on_update_n2vc_db("nsrs", {"_id": nsr_id}, "_admin.deployed", db_nsr_update) step = "Deployed at VIM" self.logger.debug(logging_text + step) @@ -975,29 +1054,36 @@ class NsLcm(LcmBase): """ my_vca = vca_deployed_list[vca_index] if my_vca.get("vdu_id") or my_vca.get("kdu_name"): + # vdu or kdu: no dependencies return timeout = 300 while timeout >= 0: - for index, vca_deployed in enumerate(vca_deployed_list): + db_nsr = self.db.get_one("nsrs", {"_id": nsr_id}) + vca_deployed_list = db_nsr["_admin"]["deployed"]["VCA"] + configuration_status_list = db_nsr["configurationStatus"] + for index, vca_deployed in enumerate(configuration_status_list): if index == vca_index: + # myself continue if not my_vca.get("member-vnf-index") or \ (vca_deployed.get("member-vnf-index") == my_vca.get("member-vnf-index")): - if not vca_deployed.get("instantiation"): - break # wait - if vca_deployed["instantiation"] == "FAILED": + internal_status = configuration_status_list[index].get("status") + if internal_status == 'READY': + continue + elif internal_status == 'BROKEN': raise LcmException("Configuration aborted because dependent charm/s has failed") + else: + break else: + # no dependencies, return return await asyncio.sleep(10) timeout -= 1 - db_nsr = self.db.get_one("nsrs", {"_id": nsr_id}) - vca_deployed_list = db_nsr["_admin"]["deployed"]["VCA"] raise LcmException("Configuration aborted because dependent charm/s timeout") - async def instantiate_N2VC(self, logging_text, vca_index, nsi_id, db_nsr, db_vnfr, vdu_id, - kdu_name, vdu_index, config_descriptor, deploy_params, base_folder): + async def instantiate_N2VC(self, logging_text, vca_index, nsi_id, db_nsr, db_vnfr, vdu_id, kdu_name, + vdu_index, config_descriptor, deploy_params, base_folder, nslcmop_id): nsr_id = db_nsr["_id"] db_update_entry = "_admin.deployed.VCA.{}.".format(vca_index) vca_deployed_list = db_nsr["_admin"]["deployed"]["VCA"] @@ -1009,6 +1095,10 @@ class NsLcm(LcmBase): } step = "" try: + + element_type = 'NS' + element_under_configuration = nsr_id + vnfr_id = None if db_vnfr: vnfr_id = db_vnfr["_id"] @@ -1016,10 +1106,15 @@ class NsLcm(LcmBase): namespace = "{nsi}.{ns}".format( nsi=nsi_id if nsi_id else "", ns=nsr_id) + if vnfr_id: + element_type = 'VNF' + element_under_configuration = vnfr_id namespace += "." + vnfr_id if vdu_id: namespace += ".{}-{}".format(vdu_id, vdu_index or 0) + element_type = 'VDU' + element_under_configuration = vdu_id + '-' + vdu_index # Get artifact path artifact_path = "{}/{}/charms/{}".format( @@ -1039,11 +1134,21 @@ class NsLcm(LcmBase): # create or register execution environment in VCA if is_proxy_charm: + + await self._write_configuration_status( + nsr_id=nsr_id, + vca_index=vca_index, + status='CREATING', + element_under_configuration=element_under_configuration, + element_type=element_type + ) + step = "create execution environment" self.logger.debug(logging_text + step) ee_id, credentials = await self.n2vc.create_execution_environment(namespace=namespace, reuse_ee_id=ee_id, db_dict=db_dict) + else: step = "Waiting to VM being up and getting IP address" self.logger.debug(logging_text + step) @@ -1066,6 +1171,14 @@ class NsLcm(LcmBase): credentials["username"] = username # n2vc_redesign STEP 3.2 + await self._write_configuration_status( + nsr_id=nsr_id, + vca_index=vca_index, + status='REGISTERING', + element_under_configuration=element_under_configuration, + element_type=element_type + ) + step = "register execution environment {}".format(credentials) self.logger.debug(logging_text + step) ee_id = await self.n2vc.register_execution_environment(credentials=credentials, namespace=namespace, @@ -1083,6 +1196,15 @@ class NsLcm(LcmBase): # n2vc_redesign STEP 3.3 step = "Install configuration Software" + + await self._write_configuration_status( + nsr_id=nsr_id, + vca_index=vca_index, + status='INSTALLING SW', + element_under_configuration=element_under_configuration, + element_type=element_type + ) + # TODO check if already done self.logger.debug(logging_text + step) await self.n2vc.install_configuration_sw(ee_id=ee_id, artifact_path=artifact_path, db_dict=db_dict) @@ -1128,8 +1250,34 @@ class NsLcm(LcmBase): # add config if not present for NS charm initial_config_primitive_list = self._get_initial_config_primitive_list(initial_config_primitive_list, vca_deployed) + + # wait for dependent primitives execution (NS -> VNF -> VDU) if initial_config_primitive_list: await self._wait_dependent_n2vc(nsr_id, vca_deployed_list, vca_index) + + # stage, in function of element type: vdu, kdu, vnf or ns + my_vca = vca_deployed_list[vca_index] + if my_vca.get("vdu_id") or my_vca.get("kdu_name"): + # VDU or KDU + stage = 'Stage 3/5: running Day-1 primitives for VDU' + elif my_vca.get("member-vnf-index"): + # VNF + stage = 'Stage 4/5: running Day-1 primitives for VNF' + else: + # NS + stage = 'Stage 5/5: running Day-1 primitives for NS' + + await self._write_configuration_status( + nsr_id=nsr_id, + vca_index=vca_index, + status='EXECUTING PRIMITIVE' + ) + + self._write_op_status( + op_id=nslcmop_id, + stage=stage + ) + for initial_config_primitive in initial_config_primitive_list: # adding information on the vca_deployed if it is a NS execution environment if not vca_deployed["member-vnf-index"]: @@ -1145,19 +1293,30 @@ class NsLcm(LcmBase): params_dict=primitive_params_, db_dict=db_dict ) + # TODO register in database that primitive is done step = "instantiated at VCA" - self.update_db_2("nsrs", nsr_id, {db_update_entry + "instantiation": "COMPLETED"}) self.logger.debug(logging_text + step) + await self._write_configuration_status( + nsr_id=nsr_id, + vca_index=vca_index, + status='READY' + ) + except Exception as e: # TODO not use Exception but N2VC exception - self.update_db_2("nsrs", nsr_id, {db_update_entry + "instantiation": "FAILED"}) + # self.update_db_2("nsrs", nsr_id, {db_update_entry + "instantiation": "FAILED"}) + await self._write_configuration_status( + nsr_id=nsr_id, + vca_index=vca_index, + status='BROKEN' + ) raise Exception("{} {}".format(step, e)) from e # TODO raise N2VC exception with 'step' extra information def _write_ns_status(self, nsr_id: str, ns_state: str, current_operation: str, current_operation_id: str, - error_description: str = None, error_detail: str = None): + error_description: str = None): try: db_dict = dict() if ns_state: @@ -1165,10 +1324,57 @@ class NsLcm(LcmBase): db_dict["currentOperation"] = current_operation db_dict["currentOperationID"] = current_operation_id db_dict["errorDescription"] = error_description - db_dict["errorDetail"] = error_detail self.update_db_2("nsrs", nsr_id, db_dict) except Exception as e: - self.logger.warn('Error writing NS status: {}'.format(e)) + self.logger.warn('Error writing NS status, ns={}: {}'.format(nsr_id, e)) + + def _write_op_status(self, op_id: str, stage: str = None, error_message: str = None, queuePosition: int = 0): + try: + db_dict = dict() + db_dict['queuePosition'] = queuePosition + db_dict['stage'] = stage + if error_message: + db_dict['errorMessage'] = error_message + self.update_db_2("nslcmops", op_id, db_dict) + except Exception as e: + self.logger.warn('Error writing OPERATION status for op_id: {} -> {}'.format(op_id, e)) + + def _write_all_config_status(self, nsr_id: str, status: str): + try: + # nsrs record + db_nsr = self.db.get_one("nsrs", {"_id": nsr_id}) + # configurationStatus + config_status = db_nsr.get('configurationStatus') + if config_status: + # update status + db_dict = dict() + db_dict['configurationStatus'] = list() + for c in config_status: + c['status'] = status + db_dict['configurationStatus'].append(c) + self.update_db_2("nsrs", nsr_id, db_dict) + + except Exception as e: + self.logger.warn('Error writing all configuration status, ns={}: {}'.format(nsr_id, e)) + + async def _write_configuration_status(self, nsr_id: str, vca_index: int, status: str, + element_under_configuration: str = None, element_type: str = None): + + # self.logger.debug('_write_configuration_status(): vca_index={}, status={}' + # .format(vca_index, status)) + + try: + db_path = 'configurationStatus.{}.'.format(vca_index) + db_dict = dict() + db_dict[db_path + 'status'] = status + if element_under_configuration: + db_dict[db_path + 'elementUnderConfiguration'] = element_under_configuration + if element_type: + db_dict[db_path + 'elementType'] = element_type + self.update_db_2("nsrs", nsr_id, db_dict) + except Exception as e: + self.logger.warn('Error writing configuration status={}, ns={}, vca_index={}: {}' + .format(status, nsr_id, vca_index, e)) async def instantiate(self, nsr_id, nslcmop_id): """ @@ -1181,7 +1387,7 @@ class NsLcm(LcmBase): # Try to lock HA task here task_is_locked_by_me = self.lcm_tasks.lock_HA('ns', 'nslcmops', nslcmop_id) if not task_is_locked_by_me: - self.logger.debug('instantiate() task is not locked by me') + self.logger.debug('instantiate() task is not locked by me, ns={}'.format(nsr_id)) return logging_text = "Task ns={} instantiate={} ".format(nsr_id, nslcmop_id) @@ -1246,6 +1452,12 @@ class NsLcm(LcmBase): db_vnfds = {} # every vnfd data indexed by vnf id db_vnfds_index = {} # every vnfd data indexed by vnf member-index + self._write_op_status( + op_id=nslcmop_id, + stage='Stage 1/5: preparation of the environment', + queuePosition=0 + ) + # for each vnf in ns, read vnfd for vnfr in db_vnfrs_list: db_vnfrs[vnfr["member-vnf-index-ref"]] = vnfr # vnf's dict indexed by member-index: '1', '2', etc @@ -1269,7 +1481,9 @@ class NsLcm(LcmBase): vca_deployed_list = db_nsr["_admin"]["deployed"].get("VCA") if vca_deployed_list is None: vca_deployed_list = [] + configuration_status_list = [] db_nsr_update["_admin.deployed.VCA"] = vca_deployed_list + db_nsr_update["configurationStatus"] = configuration_status_list # add _admin.deployed.VCA to db_nsr dictionary, value=vca_deployed_list populate_dict(db_nsr, ("_admin", "deployed", "VCA"), vca_deployed_list) elif isinstance(vca_deployed_list, dict): @@ -1288,6 +1502,14 @@ class NsLcm(LcmBase): # set state to INSTANTIATED. When instantiated NBI will not delete directly db_nsr_update["_admin.nsState"] = "INSTANTIATED" self.update_db_2("nsrs", nsr_id, db_nsr_update) + + # n2vc_redesign STEP 2 Deploy Network Scenario + + self._write_op_status( + op_id=nslcmop_id, + stage='Stage 2/5: deployment of VMs and execution environments' + ) + self.logger.debug(logging_text + "Before deploy_kdus") # Call to deploy_kdus in case exists the "vdu:kdu" param task_kdu = asyncio.ensure_future( @@ -1308,7 +1530,6 @@ class NsLcm(LcmBase): if self.vca_config.get("public_key"): n2vc_key_list.append(self.vca_config["public_key"]) - # n2vc_redesign STEP 2 Deploy Network Scenario task_ro = asyncio.ensure_future( self.instantiate_RO( logging_text=logging_text, @@ -1504,8 +1725,13 @@ class NsLcm(LcmBase): for task in pending: instantiated_ok = False if task == task_ro: + # RO task is pending + db_nsr_update["operational-status"] = "failed" + elif task == task_kdu: + # KDU task is pending db_nsr_update["operational-status"] = "failed" else: + # A N2VC task is pending db_nsr_update["config-status"] = "failed" self.logger.error(logging_text + task_instantiation_info[task] + ": Timeout") error_text_list.append(task_instantiation_info[task] + ": Timeout") @@ -1513,8 +1739,13 @@ class NsLcm(LcmBase): if task.cancelled(): instantiated_ok = False if task == task_ro: + # RO task was cancelled + db_nsr_update["operational-status"] = "failed" + elif task == task_kdu: + # KDU task was cancelled db_nsr_update["operational-status"] = "failed" else: + # A N2VC was cancelled db_nsr_update["config-status"] = "failed" self.logger.warn(logging_text + task_instantiation_info[task] + ": Cancelled") error_text_list.append(task_instantiation_info[task] + ": Cancelled") @@ -1523,10 +1754,16 @@ class NsLcm(LcmBase): if exc: instantiated_ok = False if task == task_ro: + # RO task raised an exception + db_nsr_update["operational-status"] = "failed" + elif task == task_kdu: + # KDU task raised an exception db_nsr_update["operational-status"] = "failed" else: + # A N2VC task raised an exception db_nsr_update["config-status"] = "failed" self.logger.error(logging_text + task_instantiation_info[task] + ": Failed") + if isinstance(exc, (N2VCException, ROclient.ROClientException)): error_text_list.append(task_instantiation_info[task] + ": {}".format(exc)) else: @@ -1579,26 +1816,33 @@ class NsLcm(LcmBase): # nsState="READY/BROKEN", currentOperation="IDLE", currentOperationID=None ns_state = None error_description = None - error_detail = None if instantiated_ok: ns_state = "READY" else: ns_state = "BROKEN" error_description = 'Operation: INSTANTIATING.{}, step: {}'.format(nslcmop_id, step) - error_detail = error_text + self._write_ns_status( nsr_id=nsr_id, ns_state=ns_state, current_operation="IDLE", current_operation_id=None, - error_description=error_description, - error_detail=error_detail + error_description=error_description + ) + + self._write_op_status( + op_id=nslcmop_id, + error_message=error_description ) if db_nslcmop_update: self.update_db_2("nslcmops", nslcmop_id, db_nslcmop_update) + + self.logger.debug(logging_text + 'End of instantiation: {}'.format(instantiated_ok)) + except DbException as e: self.logger.error(logging_text + "Cannot update database: {}".format(e)) + if nslcmop_operation_state: try: await self.msg.aiowrite("ns", "instantiated", {"nsr_id": nsr_id, "nslcmop_id": nslcmop_id, @@ -1613,6 +1857,8 @@ class NsLcm(LcmBase): async def deploy_kdus(self, logging_text, nsr_id, db_nsr, db_vnfrs): # Launch kdus if present in the descriptor + deployed_ok = True + k8scluster_id_2_uuic = {"helm-chart": {}, "juju-bundle": {}} def _get_cluster_id(cluster_id, cluster_type): @@ -1653,13 +1899,15 @@ class NsLcm(LcmBase): k8sclustertype = "juju" k8sclustertype_full = "juju-bundle" else: - error_text = "kdu type is neither helm-chart not juju-bundle. Maybe an old NBI version is" \ + error_text = "kdu type is neither helm-chart nor juju-bundle. Maybe an old NBI version is" \ " running" try: if not error_text: cluster_uuid = _get_cluster_id(kdur["k8s-cluster"]["id"], k8sclustertype_full) except LcmException as e: error_text = str(e) + deployed_ok = False + step = "Instantiate KDU {} in k8s cluster {}".format(kdur["kdu-name"], cluster_uuid) k8s_instace_info = {"kdu-instance": None, "k8scluster-uuid": cluster_uuid, @@ -1698,19 +1946,23 @@ class NsLcm(LcmBase): if not done_list: # timeout for task in pending_list: db_nsr_update[pending_tasks(task) + "detailed-status"] = "Timeout" + deployed_ok = False break for task in done_list: exc = task.exception() if exc: db_nsr_update[pending_tasks[task] + "detailed-status"] = "{}".format(exc) + deployed_ok = False else: db_nsr_update[pending_tasks[task] + "kdu-instance"] = task.result() + if not deployed_ok: + raise LcmException('Cannot deploy KDUs') + except Exception as e: self.logger.critical(logging_text + "Exit Exception {} while '{}': {}".format(type(e).__name__, step, e)) raise LcmException("{} Exit Exception {} while '{}': {}".format(logging_text, type(e).__name__, step, e)) finally: - # TODO Write in data base if db_nsr_update: self.update_db_2("nsrs", nsr_id, db_nsr_update) @@ -1745,7 +1997,14 @@ class NsLcm(LcmBase): "vdu_name": vdu_name, } vca_index += 1 - self.update_db_2("nsrs", nsr_id, {"_admin.deployed.VCA.{}".format(vca_index): vca_deployed}) + + # create VCA and configurationStatus in db + db_dict = { + "_admin.deployed.VCA.{}".format(vca_index): vca_deployed, + "configurationStatus.{}".format(vca_index): dict() + } + self.update_db_2("nsrs", nsr_id, db_dict) + db_nsr["_admin"]["deployed"]["VCA"].append(vca_deployed) # Launch task @@ -1762,6 +2021,7 @@ class NsLcm(LcmBase): deploy_params=deploy_params, config_descriptor=descriptor_config, base_folder=base_folder, + nslcmop_id=nslcmop_id ) ) self.lcm_tasks.register("ns", nsr_id, nslcmop_id, "instantiate_N2VC-{}".format(vca_index), task_n2vc) @@ -2066,6 +2326,12 @@ class NsLcm(LcmBase): .format(vnf_index, seq.get("name"), e), ) + async def _delete_N2VC(self, nsr_id: str): + self._write_all_config_status(nsr_id=nsr_id, status='TERMINATING') + namespace = "." + nsr_id + await self.n2vc.delete_namespace(namespace=namespace) + self._write_all_config_status(nsr_id=nsr_id, status='DELETED') + async def terminate(self, nsr_id, nslcmop_id): # Try to lock HA task here @@ -2098,6 +2364,10 @@ class NsLcm(LcmBase): current_operation="TERMINATING", current_operation_id=nslcmop_id ) + self._write_op_status( + op_id=nslcmop_id, + queuePosition=0 + ) step = "Getting nslcmop={} from db".format(nslcmop_id) db_nslcmop = self.db.get_one("nslcmops", {"_id": nslcmop_id}) @@ -2122,10 +2392,12 @@ class NsLcm(LcmBase): step = "delete execution environment" self.logger.debug(logging_text + step) - task_delete_ee = asyncio.ensure_future(self.n2vc.delete_namespace(namespace="." + nsr_id)) + task_delete_ee = asyncio.ensure_future(self._delete_N2VC(nsr_id=nsr_id)) + # task_delete_ee = asyncio.ensure_future(self.n2vc.delete_namespace(namespace="." + nsr_id)) + pending_tasks.append(task_delete_ee) except Exception as e: - msg = "Failed while deleting NS in VCA: {}".format(e) + msg = "Failed while deleting ns={} in VCA: {}".format(nsr_id, e) self.logger.error(msg) failed_detail.append(msg) @@ -2153,7 +2425,7 @@ class NsLcm(LcmBase): continue pending_tasks.append(task_delete_kdu_instance) except LcmException as e: - msg = "Failed while deleting KDUs from NS: {}".format(e) + msg = "Failed while deleting KDUs from ns={}: {}".format(nsr_id, e) self.logger.error(msg) failed_detail.append(msg) @@ -2191,6 +2463,10 @@ class NsLcm(LcmBase): item_id_name=RO_nsr_id, extra_item="action", extra_item_id=RO_delete_action) + + # deploymentStatus + self._on_update_ro_db(nsrs_id=nsr_id, ro_descriptor=desc) + ns_status, ns_status_info = self.RO.check_action_status(desc) if ns_status == "ERROR": raise ROclient.ROClientException(ns_status_info) @@ -2320,16 +2596,21 @@ class NsLcm(LcmBase): error_detail = None else: ns_state = "BROKEN" - error_description = 'Operation: TERMINATING.{}, step: {}'.format(nslcmop_id, step) error_detail = "; ".join(failed_detail) + error_description = 'Operation: TERMINATING.{}, step: {}. Detail: {}'\ + .format(nslcmop_id, step, error_detail) self._write_ns_status( nsr_id=nsr_id, ns_state=ns_state, current_operation="IDLE", current_operation_id=None, - error_description=error_description, - error_detail=error_detail + error_description=error_description + ) + + self._write_op_status( + op_id=nslcmop_id, + error_message=error_description ) except DbException as e: @@ -2683,6 +2964,11 @@ class NsLcm(LcmBase): current_operation="IDLE", current_operation_id=None ) + if exc: + self._write_op_status( + op_id=nslcmop_id, + error_message=nslcmop_operation_state_detail + ) except DbException as e: self.logger.error(logging_text + "Cannot update database: {}".format(e)) self.logger.debug(logging_text + "Exit") @@ -2966,6 +3252,10 @@ class NsLcm(LcmBase): if not RO_task_done: desc = await self.RO.show("ns", item_id_name=RO_nsr_id, extra_item="action", extra_item_id=RO_nslcmop_id) + + # deploymentStatus + self._on_update_ro_db(nsrs_id=nsr_id, ro_descriptor=desc) + ns_status, ns_status_info = self.RO.check_action_status(desc) if ns_status == "ERROR": raise ROclient.ROClientException(ns_status_info) @@ -2991,6 +3281,10 @@ class NsLcm(LcmBase): vnfr_scaled = True try: desc = await self.RO.show("ns", RO_nsr_id) + + # deploymentStatus + self._on_update_ro_db(nsrs_id=nsr_id, ro_descriptor=desc) + # nsr_deployed["nsr_ip"] = RO.get_ns_vnf_info(desc) self.ns_update_vnfr({db_vnfr["member-vnf-index-ref"]: db_vnfr}, desc) break @@ -3121,6 +3415,12 @@ class NsLcm(LcmBase): exc = traceback.format_exc() self.logger.critical(logging_text + "Exit Exception {} {}".format(type(e).__name__, e), exc_info=True) finally: + self._write_ns_status( + nsr_id=nsr_id, + ns_state=None, + current_operation="IDLE", + current_operation_id=None + ) if exc: if db_nslcmop: db_nslcmop_update["detailed-status"] = "FAILED {}: {}".format(step, exc)