X-Git-Url: https://osm.etsi.org/gitweb/?a=blobdiff_plain;ds=sidebyside;f=osm_lcm%2Fns.py;h=b7805d74a631f31b3c69b509448d2d0ed27dfd6e;hb=b901815760ff325029fe86bc8324f22c1eabdcb4;hp=5d4f34d14ad0aa4ceb2f1da078942701aec16f27;hpb=18a6332d4127b708a898e9fcf9875bfac83b61f0;p=osm%2FLCM.git diff --git a/osm_lcm/ns.py b/osm_lcm/ns.py index 5d4f34d..b7805d7 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" @@ -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 @@ -277,7 +278,7 @@ 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, n2vc_key_list): """ Creates a RO ns descriptor from OSM ns_instantiate params :param ns_params: OSM instantiate params @@ -765,7 +766,7 @@ 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) + await self._do_placement(logging_text, db_nslcmop, db_vnfrs) # deploy RO @@ -898,7 +899,7 @@ 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, 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) @@ -1137,6 +1138,10 @@ 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 @@ -1212,9 +1217,9 @@ 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 +1230,8 @@ 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 +1426,30 @@ 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: @@ -1458,28 +1461,40 @@ 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: None. Modifies database vnfrs and parameter db_vnfr with the computed 'vim-account-id' + """ + 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 self.db.set_one("vnfrs", {"_id": vnfr["_id"]}, {"vim-account-id": pla_vnf['vimAccountId']}) + # Modifies db_vnfrs + vnfr["vim-account-id"] = pla_vnf['vimAccountId'] return def update_nsrs_with_pla_result(self, params): @@ -2070,6 +2085,25 @@ 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 +2136,7 @@ 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 +2149,14 @@ 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 +2186,25 @@ 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"]) @@ -2316,9 +2358,9 @@ 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 @@ -2425,7 +2467,7 @@ 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 +2475,7 @@ 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 +2501,7 @@ 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 +2568,14 @@ 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): """ @@ -2777,13 +2822,14 @@ 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." @@ -3146,8 +3192,11 @@ 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 +3248,24 @@ 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 +3282,7 @@ 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: