X-Git-Url: https://osm.etsi.org/gitweb/?a=blobdiff_plain;f=osm_lcm%2Fns.py;h=bcd675bca44059471bd27bd93da1b7479edbab38;hb=refs%2Fheads%2Fv7.0;hp=5d4f34d14ad0aa4ceb2f1da078942701aec16f27;hpb=18a6332d4127b708a898e9fcf9875bfac83b61f0;p=osm%2FLCM.git diff --git a/osm_lcm/ns.py b/osm_lcm/ns.py index 5d4f34d..bcd675b 100644 --- a/osm_lcm/ns.py +++ b/osm_lcm/ns.py @@ -33,12 +33,13 @@ from osm_common.dbbase import DbException from osm_common.fsbase import FsException from n2vc.n2vc_juju_conn import N2VCJujuConnector -from n2vc.exceptions import N2VCException +from n2vc.exceptions import N2VCException, N2VCNotFound, K8sException from copy import copy, deepcopy from http import HTTPStatus from time import time from uuid import uuid4 +from functools import partial __author__ = "Alfonso Tierno" @@ -48,8 +49,8 @@ class NsLcm(LcmBase): timeout_ns_deploy = 2 * 3600 # default global timeout for deployment a ns timeout_ns_terminate = 1800 # default global timeout for un deployment a ns timeout_charm_delete = 10 * 60 - timeout_primitive = 10 * 60 # timeout for primitive execution - timeout_progress_primitive = 2 * 60 # timeout for some progress in a primitive execution + timeout_primitive = 30 * 60 # timeout for primitive execution + timeout_progress_primitive = 10 * 60 # timeout for some progress in a primitive execution SUBOPERATION_STATUS_NOT_FOUND = -1 SUBOPERATION_STATUS_NEW = -2 @@ -214,11 +215,11 @@ class NsLcm(LcmBase): # write to database self.update_db_2("nsrs", nsr_id, db_dict) + except (asyncio.CancelledError, asyncio.TimeoutError): + raise except Exception as e: self.logger.warn('Error updating NS state for ns={}: {}'.format(nsr_id, e)) - return - def vnfd2RO(self, vnfd, new_id=None, additionalParams=None, nsrId=None): """ Converts creates a new vnfd descriptor for RO base on input OSM IM vnfd @@ -277,10 +278,12 @@ class NsLcm(LcmBase): raise LcmException("Error parsing Jinja2 to cloud-init content at vnfd[id={}]:vdu[id={}]: {}". format(vnfd["id"], vdu["id"], e)) - def ns_params_2_RO(self, ns_params, nsd, vnfd_dict, n2vc_key_list): + def _ns_params_2_RO(self, ns_params, nsd, vnfd_dict, db_vnfrs, n2vc_key_list): """ Creates a RO ns descriptor from OSM ns_instantiate params :param ns_params: OSM instantiate params + :param vnfd_dict: database content of vnfds, indexed by id (not _id). {id: {vnfd_object}, ...} + :param db_vnfrs: database content of vnfrs, indexed by member-vnf-index. {member-vnf-index: {vnfr_object}, ...} :return: The RO ns descriptor """ vim_2_RO = {} @@ -342,6 +345,13 @@ class NsLcm(LcmBase): "wim_account": wim_account_2_RO(ns_params.get("wimAccountId")), # "scenario": ns_params["nsdId"], } + # set vim_account of each vnf if different from general vim_account. + # Get this information from database content, key vim-account-id + # Vim account can be set by placement_engine and it may be different from + # the instantiate parameters (vnfs.member-vnf-index.datacenter). + for vnf_index, vnfr in db_vnfrs.items(): + if vnfr.get("vim-account-id") and vnfr["vim-account-id"] != ns_params["vimAccountId"]: + populate_dict(RO_ns_params, ("vnfs", vnf_index, "datacenter"), vim_account_2_RO(vnfr["vim-account-id"])) n2vc_key_list = n2vc_key_list or [] for vnfd_ref, vnfd in vnfd_dict.items(): @@ -390,9 +400,6 @@ class NsLcm(LcmBase): else: raise LcmException("Invalid instantiate parameter vnf:member-vnf-index={} is not present at nsd:" "constituent-vnfd".format(vnf_params["member-vnf-index"])) - if vnf_params.get("vimAccountId"): - populate_dict(RO_ns_params, ("vnfs", vnf_params["member-vnf-index"], "datacenter"), - vim_account_2_RO(vnf_params["vimAccountId"])) for vdu_params in get_iterable(vnf_params, "vdu"): # TODO feature 1417: check that this VDU exist and it is not a PDU @@ -751,6 +758,19 @@ class NsLcm(LcmBase): async def instantiate_RO(self, logging_text, nsr_id, nsd, db_nsr, db_nslcmop, db_vnfrs, db_vnfds_ref, n2vc_key_list, stage): + """ + Instantiate at RO + :param logging_text: preffix text to use at logging + :param nsr_id: nsr identity + :param nsd: database content of ns descriptor + :param db_nsr: database content of ns record + :param db_nslcmop: database content of ns operation, in this case, 'instantiate' + :param db_vnfrs: + :param db_vnfds_ref: database content of vnfds, indexed by id (not _id). {id: {vnfd_object}, ...} + :param n2vc_key_list: ssh-public-key list to be inserted to management vdus via cloud-init + :param stage: list with 3 items: [general stage, tasks, vim_specific]. This task will write over vim_specific + :return: None or exception + """ try: db_nsr_update = {} RO_descriptor_number = 0 # number of descriptors created at RO @@ -765,10 +785,15 @@ class NsLcm(LcmBase): # Check for and optionally request placement optimization. Database will be updated if placement activated stage[2] = "Waiting for Placement." - await self.do_placement(logging_text, db_nslcmop, db_vnfrs) + if await self._do_placement(logging_text, db_nslcmop, db_vnfrs): + # in case of placement change ns_params[vimAcountId) if not present at any vnfrs + for vnfr in db_vnfrs.values(): + if ns_params["vimAccountId"] == vnfr["vim-account-id"]: + break + else: + ns_params["vimAccountId"] == vnfr["vim-account-id"] # deploy RO - # get vnfds, instantiate at RO for c_vnf in nsd.get("constituent-vnfd", ()): member_vnf_index = c_vnf["member-vnf-index"] @@ -898,7 +923,7 @@ class NsLcm(LcmBase): await asyncio.wait(task_dependency, timeout=3600) stage[2] = "Checking instantiation parameters." - RO_ns_params = self.ns_params_2_RO(ns_params, nsd, db_vnfds_ref, n2vc_key_list) + RO_ns_params = self._ns_params_2_RO(ns_params, nsd, db_vnfds_ref, db_vnfrs, n2vc_key_list) stage[2] = "Deploying ns at VIM." db_nsr_update["detailed-status"] = " ".join(stage) self.update_db_2("nsrs", nsr_id, db_nsr_update) @@ -1137,9 +1162,12 @@ class NsLcm(LcmBase): namespace += ".{}-{}".format(vdu_id, vdu_index or 0) element_type = 'VDU' element_under_configuration = "{}-{}".format(vdu_id, vdu_index or 0) + elif kdu_name: + namespace += ".{}".format(kdu_name) + element_type = 'KDU' + element_under_configuration = kdu_name # Get artifact path - self.fs.sync() # Sync from FSMongo artifact_path = "{}/{}/charms/{}".format( base_folder["folder"], base_folder["pkg-dir"], @@ -1212,9 +1240,9 @@ class NsLcm(LcmBase): ee_id_parts = ee_id.split('.') model_name = ee_id_parts[0] application_name = ee_id_parts[1] - self.update_db_2("nsrs", nsr_id, {db_update_entry + "model": model_name, - db_update_entry + "application": application_name, - db_update_entry + "ee_id": ee_id}) + db_nsr_update = {db_update_entry + "model": model_name, + db_update_entry + "application": application_name, + db_update_entry + "ee_id": ee_id} # n2vc_redesign STEP 3.3 @@ -1225,7 +1253,8 @@ class NsLcm(LcmBase): vca_index=vca_index, status='INSTALLING SW', element_under_configuration=element_under_configuration, - element_type=element_type + element_type=element_type, + other_update=db_nsr_update ) # TODO check if already done @@ -1420,33 +1449,30 @@ class NsLcm(LcmBase): except DbException 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): + def _write_all_config_status(self, db_nsr: dict, status: str): try: - # nsrs record - db_nsr = self.db.get_one("nsrs", {"_id": nsr_id}) + nsr_id = db_nsr["_id"] # configurationStatus config_status = db_nsr.get('configurationStatus') if config_status: + db_nsr_update = {"configurationStatus.{}.status".format(index): status for index, v in + enumerate(config_status) if v} # 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) + self.update_db_2("nsrs", nsr_id, db_nsr_update) except DbException as e: self.logger.warn('Error writing all configuration status, ns={}: {}'.format(nsr_id, e)) def _write_configuration_status(self, nsr_id: str, vca_index: int, status: str = None, - element_under_configuration: str = None, element_type: str = None): + element_under_configuration: str = None, element_type: str = None, + other_update: dict = 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 = other_update or {} if status: db_dict[db_path + 'status'] = status if element_under_configuration: @@ -1458,29 +1484,44 @@ class NsLcm(LcmBase): self.logger.warn('Error writing configuration status={}, ns={}, vca_index={}: {}' .format(status, nsr_id, vca_index, e)) - async def do_placement(self, logging_text, db_nslcmop, db_vnfrs): + async def _do_placement(self, logging_text, db_nslcmop, db_vnfrs): + """ + Check and computes the placement, (vim account where to deploy). If it is decided by an external tool, it + sends the request via kafka and wait until the result is wrote at database (nslcmops _admin.plca). + Database is used because the result can be obtained from a different LCM worker in case of HA. + :param logging_text: contains the prefix for logging, with the ns and nslcmop identifiers + :param db_nslcmop: database content of nslcmop + :param db_vnfrs: database content of vnfrs, indexed by member-vnf-index. + :return: True if some modification is done. Modifies database vnfrs and parameter db_vnfr with the + computed 'vim-account-id' + """ + modified = False + nslcmop_id = db_nslcmop['_id'] placement_engine = deep_get(db_nslcmop, ('operationParams', 'placement-engine')) if placement_engine == "PLA": - self.logger.debug(logging_text + "Invoke placement optimization for nslcmopId={}".format(db_nslcmop['id'])) - await self.msg.aiowrite("pla", "get_placement", {'nslcmopId': db_nslcmop['_id']}, loop=self.loop) + self.logger.debug(logging_text + "Invoke and wait for placement optimization") + await self.msg.aiowrite("pla", "get_placement", {'nslcmopId': nslcmop_id}, loop=self.loop) db_poll_interval = 5 - wait = db_poll_interval * 4 + wait = db_poll_interval * 10 pla_result = None while not pla_result and wait >= 0: await asyncio.sleep(db_poll_interval) wait -= db_poll_interval - db_nslcmop = self.db.get_one("nslcmops", {"_id": db_nslcmop["_id"]}) + db_nslcmop = self.db.get_one("nslcmops", {"_id": nslcmop_id}) pla_result = deep_get(db_nslcmop, ('_admin', 'pla')) if not pla_result: - raise LcmException("Placement timeout for nslcmopId={}".format(db_nslcmop['id'])) + raise LcmException("Placement timeout for nslcmopId={}".format(nslcmop_id)) for pla_vnf in pla_result['vnf']: vnfr = db_vnfrs.get(pla_vnf['member-vnf-index']) if not pla_vnf.get('vimAccountId') or not vnfr: continue + modified = True self.db.set_one("vnfrs", {"_id": vnfr["_id"]}, {"vim-account-id": pla_vnf['vimAccountId']}) - return + # Modifies db_vnfrs + vnfr["vim-account-id"] = pla_vnf['vimAccountId'] + return modified def update_nsrs_with_pla_result(self, params): try: @@ -1506,6 +1547,9 @@ class NsLcm(LcmBase): logging_text = "Task ns={} instantiate={} ".format(nsr_id, nslcmop_id) self.logger.debug(logging_text + "Enter") + # Sync from FSMongo + self.fs.sync() + # get all needed from database # database nsrs record @@ -1561,8 +1605,9 @@ class NsLcm(LcmBase): # read from db: ns stage[1] = "Getting nsr={} from db".format(nsr_id) db_nsr = self.db.get_one("nsrs", {"_id": nsr_id}) - # nsd is replicated into ns (no db read) - nsd = db_nsr["nsd"] + stage[1] = "Getting nsd={} from db".format(db_nsr["nsd-id"]) + nsd = self.db.get_one("nsds", {"_id": db_nsr["nsd-id"]}) + db_nsr["nsd"] = nsd # nsr_name = db_nsr["name"] # TODO short-name?? # read from db: vnf's of this ns @@ -1890,14 +1935,13 @@ class NsLcm(LcmBase): error_detail=error_detail, other_update=db_nsr_update ) - if db_nslcmop: - self._write_op_status( - op_id=nslcmop_id, - stage="", - error_message=error_description_nslcmop, - operation_state=nslcmop_operation_state, - other_update=db_nslcmop_update, - ) + self._write_op_status( + op_id=nslcmop_id, + stage="", + error_message=error_description_nslcmop, + operation_state=nslcmop_operation_state, + other_update=db_nslcmop_update, + ) if nslcmop_operation_state: try: @@ -1923,13 +1967,14 @@ class NsLcm(LcmBase): # read nsr record db_nsr = self.db.get_one("nsrs", {"_id": nsr_id}) + nsd = self.db.get_one("nsds", {"_id": db_nsr["nsd-id"]}) # this VCA data my_vca = deep_get(db_nsr, ('_admin', 'deployed', 'VCA'))[vca_index] # read all ns-configuration relations ns_relations = list() - db_ns_relations = deep_get(db_nsr, ('nsd', 'ns-configuration', 'relation')) + db_ns_relations = deep_get(nsd, ('ns-configuration', 'relation')) if db_ns_relations: for r in db_ns_relations: # check if this VCA is in the relation @@ -2022,10 +2067,13 @@ class NsLcm(LcmBase): to_vca_endpoint = None vca_list = deep_get(db_nsr, ('_admin', 'deployed', 'VCA')) for vca in vca_list: - if vca.get('vdu_id') == r.get('entities')[0].get('id') and vca.get('config_sw_installed'): + key_to_check = "vdu_id" + if vca.get("vdu_id") is None: + key_to_check = "vnfd_id" + if vca.get(key_to_check) == r.get('entities')[0].get('id') and vca.get('config_sw_installed'): from_vca_ee_id = vca.get('ee_id') from_vca_endpoint = r.get('entities')[0].get('endpoint') - if vca.get('vdu_id') == r.get('entities')[1].get('id') and vca.get('config_sw_installed'): + if vca.get(key_to_check) == r.get('entities')[1].get('id') and vca.get('config_sw_installed'): to_vca_ee_id = vca.get('ee_id') to_vca_endpoint = r.get('entities')[1].get('endpoint') if from_vca_ee_id and to_vca_ee_id: @@ -2070,6 +2118,25 @@ class NsLcm(LcmBase): self.logger.warn(logging_text + ' ERROR adding relations: {}'.format(e)) return False + def _write_db_callback(self, task, item, _id, on_done=None, on_exc=None): + """ + callback for kdu install intended to store the returned kdu_instance at database + :return: None + """ + db_update = {} + try: + result = task.result() + if on_done: + db_update[on_done] = str(result) + except Exception as e: + if on_exc: + db_update[on_exc] = str(e) + if db_update: + try: + self.update_db_2(item, _id, db_update) + except Exception: + pass + async def deploy_kdus(self, logging_text, nsr_id, nslcmop_id, db_vnfrs, db_vnfds, task_instantiation_info): # Launch kdus if present in the descriptor @@ -2102,7 +2169,7 @@ class NsLcm(LcmBase): for kdur in get_iterable(vnfr_data, "kdur"): desc_params = self._format_additional_params(kdur.get("additionalParams")) vnfd_id = vnfr_data.get('vnfd-id') - pkgdir = deep_get(db_vnfds.get(vnfd_id), ('_admin', 'storage', 'pkg-dir')) + namespace = kdur.get("k8s-namespace") if kdur.get("helm-chart"): kdumodel = kdur["helm-chart"] k8sclustertype = "helm-chart" @@ -2115,11 +2182,14 @@ class NsLcm(LcmBase): format(vnfr_data["member-vnf-index-ref"], kdur["kdu-name"])) # check if kdumodel is a file and exists try: - # path format: /vnfdid/pkkdir/kdumodel - filename = '{}/{}/{}s/{}'.format(vnfd_id, pkgdir, k8sclustertype, kdumodel) - if self.fs.file_exists(filename, mode='file') or self.fs.file_exists(filename, mode='dir'): - kdumodel = self.fs.path + filename - except asyncio.CancelledError: + storage = deep_get(db_vnfds.get(vnfd_id), ('_admin', 'storage')) + if storage and storage.get('pkg-dir'): # may be not present if vnfd has not artifacts + # path format: /vnfdid/pkkdir/helm-charts|juju-bundles/kdumodel + filename = '{}/{}/{}s/{}'.format(storage["folder"], storage["pkg-dir"], k8sclustertype, + kdumodel) + if self.fs.file_exists(filename, mode='file') or self.fs.file_exists(filename, mode='dir'): + kdumodel = self.fs.path + filename + except (asyncio.TimeoutError, asyncio.CancelledError): raise except Exception: # it is not a file pass @@ -2149,20 +2219,25 @@ class NsLcm(LcmBase): "k8scluster-type": k8sclustertype, "member-vnf-index": vnfr_data["member-vnf-index-ref"], "kdu-name": kdur["kdu-name"], - "kdu-model": kdumodel} - db_nsr_update["_admin.deployed.K8s.{}".format(index)] = k8s_instace_info + "kdu-model": kdumodel, + "namespace": namespace} + db_path = "_admin.deployed.K8s.{}".format(index) + db_nsr_update[db_path] = k8s_instace_info self.update_db_2("nsrs", nsr_id, db_nsr_update) db_dict = {"collection": "nsrs", "filter": {"_id": nsr_id}, - "path": "_admin.deployed.K8s.{}".format(index)} + "path": db_path} task = asyncio.ensure_future( self.k8scluster_map[k8sclustertype].install(cluster_uuid=cluster_uuid, kdu_model=kdumodel, atomic=True, params=desc_params, db_dict=db_dict, timeout=600, - kdu_name=kdur["kdu-name"])) + kdu_name=kdur["kdu-name"], namespace=namespace)) + task.add_done_callback(partial(self._write_db_callback, item="nsrs", _id=nsr_id, + on_done=db_path + ".kdu-instance", + on_exc=db_path + ".detailed-status")) self.lcm_tasks.register("ns", nsr_id, nslcmop_id, "instantiate_KDU-{}".format(index), task) task_instantiation_info[task] = "Deploying KDU {}".format(kdur["kdu-name"]) @@ -2316,16 +2391,16 @@ class NsLcm(LcmBase): # sub-operations - def _reintent_or_skip_suboperation(self, db_nslcmop, op_index): - op = db_nslcmop.get('_admin', {}).get('operations', [])[op_index] - if (op.get('operationState') == 'COMPLETED'): + def _retry_or_skip_suboperation(self, db_nslcmop, op_index): + op = deep_get(db_nslcmop, ('_admin', 'operations'), [])[op_index] + if op.get('operationState') == 'COMPLETED': # b. Skip sub-operation # _ns_execute_primitive() or RO.create_action() will NOT be executed return self.SUBOPERATION_STATUS_SKIP else: - # c. Reintent executing sub-operation + # c. retry executing sub-operation # The sub-operation exists, and operationState != 'COMPLETED' - # Update operationState = 'PROCESSING' to indicate a reintent. + # Update operationState = 'PROCESSING' to indicate a retry. operationState = 'PROCESSING' detailed_status = 'In progress' self._update_suboperation_status( @@ -2338,7 +2413,7 @@ class NsLcm(LcmBase): # Find a sub-operation where all keys in a matching dictionary must match # Returns the index of the matching sub-operation, or SUBOPERATION_STATUS_NOT_FOUND if no match def _find_suboperation(self, db_nslcmop, match): - if (db_nslcmop and match): + if db_nslcmop and match: op_list = db_nslcmop.get('_admin', {}).get('operations', []) for i, op in enumerate(op_list): if all(op.get(k) == match[k] for k in match): @@ -2406,11 +2481,11 @@ class NsLcm(LcmBase): # Check for 3 different cases: # a. New: First time execution, return SUBOPERATION_STATUS_NEW # b. Skip: Existing sub-operation exists, operationState == 'COMPLETED', return SUBOPERATION_STATUS_SKIP - # c. Reintent: Existing sub-operation exists, operationState != 'COMPLETED', return op_index to re-execute + # c. retry: Existing sub-operation exists, operationState != 'COMPLETED', return op_index to re-execute def _check_or_add_scale_suboperation(self, db_nslcmop, vnf_index, vnf_config_primitive, primitive_params, operationType, RO_nsr_id=None, RO_scaling_info=None): # Find this sub-operation - if (RO_nsr_id and RO_scaling_info): + if RO_nsr_id and RO_scaling_info: operationType = 'SCALE-RO' match = { 'member_vnf_index': vnf_index, @@ -2425,7 +2500,7 @@ class NsLcm(LcmBase): 'lcmOperationType': operationType } op_index = self._find_suboperation(db_nslcmop, match) - if (op_index == self.SUBOPERATION_STATUS_NOT_FOUND): + if op_index == self.SUBOPERATION_STATUS_NOT_FOUND: # a. New sub-operation # The sub-operation does not exist, add it. # _ns_execute_primitive() will be called from scale() as usual, with non-modified arguments @@ -2433,7 +2508,7 @@ class NsLcm(LcmBase): vdu_id = None vdu_count_index = None vdu_name = None - if (RO_nsr_id and RO_scaling_info): + if RO_nsr_id and RO_scaling_info: vnf_config_primitive = None primitive_params = None else: @@ -2459,7 +2534,7 @@ class NsLcm(LcmBase): else: # Return either SUBOPERATION_STATUS_SKIP (operationState == 'COMPLETED'), # or op_index (operationState != 'COMPLETED') - return self._reintent_or_skip_suboperation(db_nslcmop, op_index) + return self._retry_or_skip_suboperation(db_nslcmop, op_index) # Function to return execution_environment id @@ -2526,11 +2601,14 @@ class NsLcm(LcmBase): if destroy_ee: await self.n2vc.delete_execution_environment(vca_deployed["ee_id"]) - 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, total_timeout=self.timeout_charm_delete) - self._write_all_config_status(nsr_id=nsr_id, status='DELETED') + async def _delete_all_N2VC(self, db_nsr: dict): + self._write_all_config_status(db_nsr=db_nsr, status='TERMINATING') + namespace = "." + db_nsr["_id"] + try: + await self.n2vc.delete_namespace(namespace=namespace, total_timeout=self.timeout_charm_delete) + except N2VCNotFound: # already deleted. Skip + pass + self._write_all_config_status(db_nsr=db_nsr, status='DELETED') async def _terminate_RO(self, logging_text, nsr_deployed, nsr_id, nslcmop_id, stage): """ @@ -2686,6 +2764,7 @@ class NsLcm(LcmBase): timeout_ns_terminate = self.timeout_ns_terminate db_nsr = None db_nslcmop = None + operation_params = None exc = None error_list = [] # annotates all failed error messages db_nslcmop_update = {} @@ -2777,13 +2856,14 @@ class NsLcm(LcmBase): # remove All execution environments at once stage[0] = "Stage 3/3 delete all." - stage[1] = "Deleting all execution environments." - self.logger.debug(logging_text + stage[1]) - task_delete_ee = asyncio.ensure_future(asyncio.wait_for(self._delete_N2VC(nsr_id=nsr_id), - timeout=self.timeout_charm_delete)) - # task_delete_ee = asyncio.ensure_future(self.n2vc.delete_namespace(namespace="." + nsr_id)) - tasks_dict_info[task_delete_ee] = "Terminating all VCA" + if nsr_deployed.get("VCA"): + stage[1] = "Deleting all execution environments." + self.logger.debug(logging_text + stage[1]) + task_delete_ee = asyncio.ensure_future(asyncio.wait_for(self._delete_all_N2VC(db_nsr=db_nsr), + timeout=self.timeout_charm_delete)) + # task_delete_ee = asyncio.ensure_future(self.n2vc.delete_namespace(namespace="." + nsr_id)) + tasks_dict_info[task_delete_ee] = "Terminating all VCA" # Delete from k8scluster stage[1] = "Deleting KDUs." @@ -2869,14 +2949,14 @@ class NsLcm(LcmBase): error_detail=error_detail, other_update=db_nsr_update ) - if db_nslcmop: - self._write_op_status( - op_id=nslcmop_id, - stage="", - error_message=error_description_nslcmop, - operation_state=nslcmop_operation_state, - other_update=db_nslcmop_update, - ) + self._write_op_status( + op_id=nslcmop_id, + stage="", + error_message=error_description_nslcmop, + operation_state=nslcmop_operation_state, + other_update=db_nslcmop_update, + ) + if operation_params: autoremove = operation_params.get("autoremove", False) if nslcmop_operation_state: try: @@ -2922,7 +3002,8 @@ class NsLcm(LcmBase): new_error = created_tasks_info[task] + ": {}".format(exc) error_list.append(created_tasks_info[task]) error_detail_list.append(new_error) - if isinstance(exc, (str, DbException, N2VCException, ROclient.ROClientException, LcmException)): + if isinstance(exc, (str, DbException, N2VCException, ROclient.ROClientException, LcmException, + K8sException)): self.logger.error(logging_text + new_error) else: exc_traceback = "".join(traceback.format_exception(None, exc, exc.__traceback__)) @@ -3146,8 +3227,11 @@ class NsLcm(LcmBase): else: desc_params = self._format_additional_params(db_nsr.get("additionalParamsForNs")) + if kdu_name: + kdu_action = True if not deep_get(kdu, ("kdu-configuration", "juju")) else False + # TODO check if ns is in a proper status - if kdu_name and primitive in ("upgrade", "rollback", "status"): + if kdu_name and (primitive in ("upgrade", "rollback", "status") or kdu_action): # kdur and desc_params already set from before if primitive_params: desc_params.update(primitive_params) @@ -3199,13 +3283,24 @@ class NsLcm(LcmBase): cluster_uuid=kdu.get("k8scluster-uuid"), kdu_instance=kdu.get("kdu-instance")), timeout=timeout_ns_action) + else: + kdu_instance = kdu.get("kdu-instance") or "{}-{}".format(kdu["kdu-name"], nsr_id) + params = self._map_primitive_params(config_primitive_desc, primitive_params, desc_params) + + detailed_status = await asyncio.wait_for( + self.k8scluster_map[kdu["k8scluster-type"]].exec_primitive( + cluster_uuid=kdu.get("k8scluster-uuid"), + kdu_instance=kdu_instance, + primitive_name=primitive, + params=params, db_dict=db_dict, + timeout=timeout_ns_action), + timeout=timeout_ns_action) if detailed_status: nslcmop_operation_state = 'COMPLETED' else: detailed_status = '' nslcmop_operation_state = 'FAILED' - else: nslcmop_operation_state, detailed_status = await self._ns_execute_primitive( self._look_for_deployed_vca(nsr_deployed["VCA"], @@ -3222,7 +3317,7 @@ class NsLcm(LcmBase): detailed_status)) return # database update is called inside finally - except (DbException, LcmException, N2VCException) as e: + except (DbException, LcmException, N2VCException, K8sException) as e: self.logger.error(logging_text + "Exit Exception {}".format(e)) exc = e except asyncio.CancelledError: @@ -3250,14 +3345,13 @@ class NsLcm(LcmBase): other_update=db_nsr_update ) - if db_nslcmop: - self._write_op_status( - op_id=nslcmop_id, - stage="", - error_message=error_description_nslcmop, - operation_state=nslcmop_operation_state, - other_update=db_nslcmop_update, - ) + self._write_op_status( + op_id=nslcmop_id, + stage="", + error_message=error_description_nslcmop, + operation_state=nslcmop_operation_state, + other_update=db_nslcmop_update, + ) if nslcmop_operation_state: try: @@ -3453,10 +3547,10 @@ class NsLcm(LcmBase): db_nsr_update["config-status"] = "configuring pre-scaling" primitive_params = self._map_primitive_params(config_primitive, {}, vnfr_params) - # Pre-scale reintent check: Check if this sub-operation has been executed before + # Pre-scale retry check: Check if this sub-operation has been executed before op_index = self._check_or_add_scale_suboperation( db_nslcmop, nslcmop_id, vnf_index, vnf_config_primitive, primitive_params, 'PRE-SCALE') - if (op_index == self.SUBOPERATION_STATUS_SKIP): + if op_index == self.SUBOPERATION_STATUS_SKIP: # Skip sub-operation result = 'COMPLETED' result_detail = 'Done' @@ -3464,20 +3558,20 @@ class NsLcm(LcmBase): "vnf_config_primitive={} Skipped sub-operation, result {} {}".format( vnf_config_primitive, result, result_detail)) else: - if (op_index == self.SUBOPERATION_STATUS_NEW): + if op_index == self.SUBOPERATION_STATUS_NEW: # New sub-operation: Get index of this sub-operation op_index = len(db_nslcmop.get('_admin', {}).get('operations')) - 1 self.logger.debug(logging_text + "vnf_config_primitive={} New sub-operation". format(vnf_config_primitive)) else: - # Reintent: Get registered params for this existing sub-operation + # retry: Get registered params for this existing sub-operation op = db_nslcmop.get('_admin', {}).get('operations', [])[op_index] vnf_index = op.get('member_vnf_index') vnf_config_primitive = op.get('primitive') primitive_params = op.get('primitive_params') - self.logger.debug(logging_text + "vnf_config_primitive={} Sub-operation reintent". + self.logger.debug(logging_text + "vnf_config_primitive={} Sub-operation retry". format(vnf_config_primitive)) - # Execute the primitive, either with new (first-time) or registered (reintent) args + # Execute the primitive, either with new (first-time) or registered (retry) args result, result_detail = await self._ns_execute_primitive( self._look_for_deployed_vca(nsr_deployed["VCA"], member_vnf_index=vnf_index, @@ -3501,26 +3595,26 @@ class NsLcm(LcmBase): # if (RO_nsr_id and RO_scaling_info): if RO_scaling_info: scale_process = "RO" - # Scale RO reintent check: Check if this sub-operation has been executed before + # Scale RO retry check: Check if this sub-operation has been executed before op_index = self._check_or_add_scale_suboperation( db_nslcmop, vnf_index, None, None, 'SCALE-RO', RO_nsr_id, RO_scaling_info) - if (op_index == self.SUBOPERATION_STATUS_SKIP): + if op_index == self.SUBOPERATION_STATUS_SKIP: # Skip sub-operation result = 'COMPLETED' result_detail = 'Done' self.logger.debug(logging_text + "Skipped sub-operation RO, result {} {}".format( result, result_detail)) else: - if (op_index == self.SUBOPERATION_STATUS_NEW): + if op_index == self.SUBOPERATION_STATUS_NEW: # New sub-operation: Get index of this sub-operation op_index = len(db_nslcmop.get('_admin', {}).get('operations')) - 1 self.logger.debug(logging_text + "New sub-operation RO") else: - # Reintent: Get registered params for this existing sub-operation + # retry: Get registered params for this existing sub-operation op = db_nslcmop.get('_admin', {}).get('operations', [])[op_index] RO_nsr_id = op.get('RO_nsr_id') RO_scaling_info = op.get('RO_scaling_info') - self.logger.debug(logging_text + "Sub-operation RO reintent".format( + self.logger.debug(logging_text + "Sub-operation RO retry for primitive {}".format( vnf_config_primitive)) RO_desc = await self.RO.create_action("ns", RO_nsr_id, {"vdu-scaling": RO_scaling_info}) @@ -3646,7 +3740,7 @@ class NsLcm(LcmBase): db_nsr_update["config-status"] = "configuring post-scaling" primitive_params = self._map_primitive_params(config_primitive, {}, vnfr_params) - # Post-scale reintent check: Check if this sub-operation has been executed before + # Post-scale retry check: Check if this sub-operation has been executed before op_index = self._check_or_add_scale_suboperation( db_nslcmop, nslcmop_id, vnf_index, vnf_config_primitive, primitive_params, 'POST-SCALE') if op_index == self.SUBOPERATION_STATUS_SKIP: @@ -3663,14 +3757,14 @@ class NsLcm(LcmBase): self.logger.debug(logging_text + "vnf_config_primitive={} New sub-operation". format(vnf_config_primitive)) else: - # Reintent: Get registered params for this existing sub-operation + # retry: Get registered params for this existing sub-operation op = db_nslcmop.get('_admin', {}).get('operations', [])[op_index] vnf_index = op.get('member_vnf_index') vnf_config_primitive = op.get('primitive') primitive_params = op.get('primitive_params') - self.logger.debug(logging_text + "vnf_config_primitive={} Sub-operation reintent". + self.logger.debug(logging_text + "vnf_config_primitive={} Sub-operation retry". format(vnf_config_primitive)) - # Execute the primitive, either with new (first-time) or registered (reintent) args + # Execute the primitive, either with new (first-time) or registered (retry) args result, result_detail = await self._ns_execute_primitive( self._look_for_deployed_vca(nsr_deployed["VCA"], member_vnf_index=vnf_index, @@ -3689,9 +3783,6 @@ class NsLcm(LcmBase): scale_process = None # POST-SCALE END - db_nslcmop_update["operationState"] = nslcmop_operation_state = "COMPLETED" - db_nslcmop_update["statusEnteredTime"] = time() - db_nslcmop_update["detailed-status"] = "done" db_nsr_update["detailed-status"] = "" # "scaled {} {}".format(scaling_group, scaling_type) db_nsr_update["operational-status"] = "running" if old_operational_status == "failed" \ else old_operational_status @@ -3714,10 +3805,8 @@ class NsLcm(LcmBase): current_operation_id=None ) if exc: - if db_nslcmop: - db_nslcmop_update["detailed-status"] = "FAILED {}: {}".format(step, exc) - db_nslcmop_update["operationState"] = nslcmop_operation_state = "FAILED" - db_nslcmop_update["statusEnteredTime"] = time() + db_nslcmop_update["detailed-status"] = error_description_nslcmop = "FAILED {}: {}".format(step, exc) + nslcmop_operation_state = "FAILED" if db_nsr: db_nsr_update["operational-status"] = old_operational_status db_nsr_update["config-status"] = old_config_status @@ -3729,20 +3818,27 @@ class NsLcm(LcmBase): db_nsr_update["operational-status"] = "failed" db_nsr_update["detailed-status"] = "FAILED scaling nslcmop={} {}: {}".format(nslcmop_id, step, exc) - try: - if db_nslcmop and db_nslcmop_update: - self.update_db_2("nslcmops", nslcmop_id, db_nslcmop_update) - if db_nsr: - self._write_ns_status( - nsr_id=nsr_id, - ns_state=None, - current_operation="IDLE", - current_operation_id=None, - other_update=db_nsr_update - ) + else: + error_description_nslcmop = None + nslcmop_operation_state = "COMPLETED" + db_nslcmop_update["detailed-status"] = "Done" + + self._write_op_status( + op_id=nslcmop_id, + stage="", + error_message=error_description_nslcmop, + operation_state=nslcmop_operation_state, + other_update=db_nslcmop_update, + ) + if db_nsr: + self._write_ns_status( + nsr_id=nsr_id, + ns_state=None, + current_operation="IDLE", + current_operation_id=None, + other_update=db_nsr_update + ) - except DbException as e: - self.logger.error(logging_text + "Cannot update database: {}".format(e)) if nslcmop_operation_state: try: await self.msg.aiowrite("ns", "scaled", {"nsr_id": nsr_id, "nslcmop_id": nslcmop_id,