X-Git-Url: https://osm.etsi.org/gitweb/?a=blobdiff_plain;f=osm_lcm%2Fns.py;h=76f0bf5f8f4d93be4ed7b08ac3a95d63e2657811;hb=2c8ab4dcb82ecbc207f2e2bf5dd37707bc4be73e;hp=63ae2a409915d3c826b36cd7b8ee15ab74bf87ce;hpb=9e9a923448e6673fc0396265732115b55f1df9da;p=osm%2FLCM.git diff --git a/osm_lcm/ns.py b/osm_lcm/ns.py index 63ae2a4..76f0bf5 100644 --- a/osm_lcm/ns.py +++ b/osm_lcm/ns.py @@ -97,7 +97,6 @@ from osm_lcm.data_utils.vnfr import ( from osm_lcm.data_utils.dict_utils import parse_yaml_strings from osm_lcm.data_utils.database.vim_account import VimAccountDB from n2vc.definitions import RelationEndpoint -from n2vc.k8s_helm_conn import K8sHelmConnector from n2vc.k8s_helm3_conn import K8sHelm3Connector from n2vc.k8s_juju_conn import K8sJujuConnector @@ -172,15 +171,6 @@ class NsLcm(LcmBase): on_update_db=self._on_update_n2vc_db, ) - self.k8sclusterhelm2 = K8sHelmConnector( - kubectl_command=self.vca_config.kubectlpath, - helm_command=self.vca_config.helmpath, - log=self.logger, - on_update_db=None, - fs=self.fs, - db=self.db, - ) - self.k8sclusterhelm3 = K8sHelm3Connector( kubectl_command=self.vca_config.kubectlpath, helm_command=self.vca_config.helm3path, @@ -200,7 +190,6 @@ class NsLcm(LcmBase): ) self.k8scluster_map = { - "helm-chart": self.k8sclusterhelm2, "helm-chart-v3": self.k8sclusterhelm3, "chart": self.k8sclusterhelm3, "juju-bundle": self.k8sclusterjuju, @@ -1647,7 +1636,7 @@ class NsLcm(LcmBase): vca_id = self.get_vca_id(db_vnfr, db_nsr) # create or register execution environment in VCA - if vca_type in ("lxc_proxy_charm", "k8s_proxy_charm", "helm", "helm-v3"): + if vca_type in ("lxc_proxy_charm", "k8s_proxy_charm", "helm-v3"): self._write_configuration_status( nsr_id=nsr_id, vca_index=vca_index, @@ -1669,7 +1658,7 @@ class NsLcm(LcmBase): db_dict=db_dict, vca_id=vca_id, ) - elif vca_type == "helm" or vca_type == "helm-v3": + elif vca_type == "helm-v3": ee_id, credentials = await self.vca_map[ vca_type ].create_execution_environment( @@ -1815,7 +1804,7 @@ class NsLcm(LcmBase): # if SSH access is required, then get execution environment SSH public # if native charm we have waited already to VM be UP - if vca_type in ("k8s_proxy_charm", "lxc_proxy_charm", "helm", "helm-v3"): + if vca_type in ("k8s_proxy_charm", "lxc_proxy_charm", "helm-v3"): pub_key = None user = None # self.logger.debug("get ssh key block") @@ -1957,7 +1946,7 @@ class NsLcm(LcmBase): # TODO register in database that primitive is done # STEP 7 Configure metrics - if vca_type == "helm" or vca_type == "helm-v3": + if vca_type == "helm-v3": # TODO: review for those cases where the helm chart is a reference and # is not part of the NF package prometheus_jobs = await self.extract_prometheus_scrape_jobs( @@ -2965,7 +2954,15 @@ class NsLcm(LcmBase): stage[1] = stage[2] = "" except asyncio.CancelledError: error_list.append("Cancelled") - # TODO cancel all tasks + await self._cancel_pending_tasks(logging_text, tasks_dict_info) + await self._wait_for_tasks( + logging_text, + tasks_dict_info, + timeout_ns_deploy, + stage, + nslcmop_id, + nsr_id=nsr_id, + ) except Exception as exc: error_list.append(str(exc)) @@ -3628,9 +3625,11 @@ class NsLcm(LcmBase): vnfr_data.get("_id"), {"kdur.{}.status".format(kdu_index): "ERROR"}, ) - except Exception: + except Exception as error: # ignore to keep original exception - pass + self.logger.warning( + f"An exception occurred while updating DB: {str(error)}" + ) # reraise original error raise @@ -3649,7 +3648,6 @@ class NsLcm(LcmBase): k8scluster_id_2_uuic = { "helm-chart-v3": {}, - "helm-chart": {}, "juju-bundle": {}, } @@ -3748,11 +3746,6 @@ class NsLcm(LcmBase): # Default version: helm3, if helm-version is v2 assign v2 k8sclustertype = "helm-chart-v3" self.logger.debug("kdur: {}".format(kdur)) - if ( - kdur.get("helm-version") - and kdur.get("helm-version") == "v2" - ): - k8sclustertype = "helm-chart" elif kdur.get("juju-bundle"): kdumodel = kdur["juju-bundle"] k8sclustertype = "juju-bundle" @@ -3790,8 +3783,8 @@ class NsLcm(LcmBase): kdumodel = self.fs.path + filename except (asyncio.TimeoutError, asyncio.CancelledError): raise - except Exception: # it is not a file - pass + except Exception as e: # it is not a file + self.logger.warning(f"An exception occurred: {str(e)}") k8s_cluster_id = kdur["k8s-cluster"]["id"] step = "Synchronize repos for k8s cluster '{}'".format( @@ -3973,10 +3966,7 @@ class NsLcm(LcmBase): vca_type = "native_charm" elif ee_item.get("helm-chart"): vca_name = ee_item["helm-chart"] - if ee_item.get("helm-version") and ee_item.get("helm-version") == "v2": - vca_type = "helm" - else: - vca_type = "helm-v3" + vca_type = "helm-v3" else: self.logger.debug( logging_text + "skipping non juju neither charm configuration" @@ -4492,9 +4482,7 @@ class NsLcm(LcmBase): ) and vca.get("needed_terminate") # For helm we must destroy_ee. Also for native_charm, as juju_model cannot be deleted if there are # pending native charms - destroy_ee = ( - True if vca_type in ("helm", "helm-v3", "native_charm") else False - ) + destroy_ee = True if vca_type in ("helm-v3", "native_charm") else False # self.logger.debug(logging_text + "vca_index: {}, ee_id: {}, vca_type: {} destroy_ee: {}".format( # vca_index, vca.get("ee_id"), vca_type, destroy_ee)) task = asyncio.ensure_future( @@ -4634,7 +4622,14 @@ class NsLcm(LcmBase): stage[1] = stage[2] = "" except asyncio.CancelledError: error_list.append("Cancelled") - # TODO cancell all tasks + await self._cancel_pending_tasks(logging_text, tasks_dict_info) + await self._wait_for_tasks( + logging_text, + tasks_dict_info, + timeout_ns_terminate, + stage, + nslcmop_id, + ) except Exception as exc: error_list.append(str(exc)) # update status at database @@ -4797,6 +4792,11 @@ class NsLcm(LcmBase): self._write_op_status(nslcmop_id, stage) return error_detail_list + async def _cancel_pending_tasks(self, logging_text, created_tasks_info): + for task, name in created_tasks_info.items(): + self.logger.debug(logging_text + "Cancelling task: " + name) + task.cancel() + @staticmethod def _map_primitive_params(primitive_desc, params, instantiation_params): """ @@ -5161,7 +5161,7 @@ class NsLcm(LcmBase): kdu_action = ( True if primitive_name in actions - and kdu["k8scluster-type"] not in ("helm-chart", "helm-chart-v3") + and kdu["k8scluster-type"] != "helm-chart-v3" else False ) @@ -5851,13 +5851,7 @@ class NsLcm(LcmBase): # add chart to list and all parameters step = "Getting helm chart name" chart_name = ee_item.get("helm-chart") - if ( - ee_item.get("helm-version") - and ee_item.get("helm-version") == "v2" - ): - vca_type = "helm" - else: - vca_type = "helm-v3" + vca_type = "helm-v3" step = "Setting Helm chart artifact paths" helm_artifacts.append( @@ -5909,7 +5903,25 @@ class NsLcm(LcmBase): if result == "FAILED": nslcmop_operation_state = result error_description_nslcmop = detailed_status + old_operational_status = "failed" db_nslcmop_update["detailed-status"] = detailed_status + db_nsr_update["detailed-status"] = detailed_status + scaling_aspect = get_scaling_aspect(latest_vnfd) + scaling_group_desc = db_nsr.get("_admin").get( + "scaling-group", None + ) + if scaling_group_desc: + for aspect in scaling_aspect: + scaling_group_id = aspect.get("id") + for scale_index, scaling_group in enumerate( + scaling_group_desc + ): + if scaling_group.get("name") == scaling_group_id: + db_nsr_update[ + "_admin.scaling-group.{}.nb-scale-op".format( + scale_index + ) + ] = 0 self.logger.debug( logging_text + " step {} Done with result {} {}".format( @@ -6238,6 +6250,7 @@ class NsLcm(LcmBase): old_operational_status = "" old_config_status = "" nsi_id = None + prom_job_name = "" try: # wait for any previous tasks in process step = "Waiting for previous operations to terminate" @@ -6348,6 +6361,15 @@ class NsLcm(LcmBase): vdud = get_vdu(db_vnfd, vdu_delta["id"]) # vdu_index also provides the number of instance of the targeted vdu vdu_count = vdu_index = get_vdur_index(db_vnfr, vdu_delta) + if vdu_index <= len(db_vnfr["vdur"]): + vdu_name_id = db_vnfr["vdur"][vdu_index - 1]["vdu-name"] + prom_job_name = ( + db_vnfr["_id"] + vdu_name_id + str(vdu_index - 1) + ) + prom_job_name = prom_job_name.replace("_", "") + prom_job_name = prom_job_name.replace("-", "") + else: + prom_job_name = None cloud_init_text = self._get_vdu_cloud_init_content( vdud, db_vnfd ) @@ -6424,11 +6446,6 @@ class NsLcm(LcmBase): if kdur.get("helm-chart"): k8s_cluster_type = "helm-chart-v3" self.logger.debug("kdur: {}".format(kdur)) - if ( - kdur.get("helm-version") - and kdur.get("helm-version") == "v2" - ): - k8s_cluster_type = "helm-chart" elif kdur.get("juju-bundle"): k8s_cluster_type = "juju-bundle" else: @@ -6557,11 +6574,6 @@ class NsLcm(LcmBase): if kdur.get("helm-chart"): k8s_cluster_type = "helm-chart-v3" self.logger.debug("kdur: {}".format(kdur)) - if ( - kdur.get("helm-version") - and kdur.get("helm-version") == "v2" - ): - k8s_cluster_type = "helm-chart" elif kdur.get("juju-bundle"): k8s_cluster_type = "juju-bundle" else: @@ -7146,7 +7158,69 @@ class NsLcm(LcmBase): db_nsr_update["config-status"] = old_config_status scale_process = None # POST-SCALE END + # Check if each vnf has exporter for metric collection if so update prometheus job records + if scaling_type == "SCALE_OUT": + if "exporters-endpoints" in db_vnfd.get("df")[0]: + vnfr_id = db_vnfr["id"] + db_vnfr = self.db.get_one("vnfrs", {"_id": vnfr_id}) + exporter_config = db_vnfd.get("df")[0].get("exporters-endpoints") + self.logger.debug("exporter config :{}".format(exporter_config)) + artifact_path = "{}/{}/{}".format( + base_folder["folder"], + base_folder["pkg-dir"], + "exporter-endpoint", + ) + ee_id = None + ee_config_descriptor = exporter_config + rw_mgmt_ip = await self.wait_vm_up_insert_key_ro( + logging_text, + nsr_id, + vnfr_id, + vdu_id=db_vnfr["vdur"][-1]["vdu-id-ref"], + vdu_index=db_vnfr["vdur"][-1]["count-index"], + user=None, + pub_key=None, + ) + self.logger.debug("rw_mgmt_ip:{}".format(rw_mgmt_ip)) + self.logger.debug("Artifact_path:{}".format(artifact_path)) + vdu_id_for_prom = None + vdu_index_for_prom = None + for x in get_iterable(db_vnfr, "vdur"): + vdu_id_for_prom = x.get("vdu-id-ref") + vdu_index_for_prom = x.get("count-index") + vnfr_id = vnfr_id + vdu_id + str(vdu_index) + vnfr_id = vnfr_id.replace("_", "") + prometheus_jobs = await self.extract_prometheus_scrape_jobs( + ee_id=ee_id, + artifact_path=artifact_path, + ee_config_descriptor=ee_config_descriptor, + vnfr_id=vnfr_id, + nsr_id=nsr_id, + target_ip=rw_mgmt_ip, + element_type="VDU", + vdu_id=vdu_id_for_prom, + vdu_index=vdu_index_for_prom, + ) + self.logger.debug("Prometheus job:{}".format(prometheus_jobs)) + if prometheus_jobs: + db_nsr_update[ + "_admin.deployed.prometheus_jobs" + ] = prometheus_jobs + self.update_db_2( + "nsrs", + nsr_id, + db_nsr_update, + ) + + for job in prometheus_jobs: + self.db.set_one( + "prometheus_jobs", + {"job_name": ""}, + job, + upsert=True, + fail_on_empty=False, + ) db_nsr_update[ "detailed-status" ] = "" # "scaled {} {}".format(scaling_group, scaling_type) @@ -7177,16 +7251,31 @@ class NsLcm(LcmBase): exc_info=True, ) finally: + error_list = list() + if exc: + error_list.append(str(exc)) self._write_ns_status( nsr_id=nsr_id, ns_state=None, current_operation="IDLE", current_operation_id=None, ) - if tasks_dict_info: - stage[1] = "Waiting for instantiate pending tasks." - self.logger.debug(logging_text + stage[1]) - exc = await self._wait_for_tasks( + try: + if tasks_dict_info: + stage[1] = "Waiting for instantiate pending tasks." + self.logger.debug(logging_text + stage[1]) + exc = await self._wait_for_tasks( + logging_text, + tasks_dict_info, + self.timeout.ns_deploy, + stage, + nslcmop_id, + nsr_id=nsr_id, + ) + except asyncio.CancelledError: + error_list.append("Cancelled") + await self._cancel_pending_tasks(logging_text, tasks_dict_info) + await self._wait_for_tasks( logging_text, tasks_dict_info, self.timeout.ns_deploy, @@ -7194,10 +7283,13 @@ class NsLcm(LcmBase): nslcmop_id, nsr_id=nsr_id, ) - if exc: + if error_list: + error_detail = "; ".join(error_list) db_nslcmop_update[ "detailed-status" - ] = error_description_nslcmop = "FAILED {}: {}".format(step, exc) + ] = error_description_nslcmop = "FAILED {}: {}".format( + step, error_detail + ) nslcmop_operation_state = "FAILED" if db_nsr: db_nsr_update["operational-status"] = old_operational_status @@ -7211,12 +7303,18 @@ class NsLcm(LcmBase): db_nsr_update[ "detailed-status" ] = "FAILED scaling nslcmop={} {}: {}".format( - nslcmop_id, step, exc + nslcmop_id, step, error_detail ) else: error_description_nslcmop = None nslcmop_operation_state = "COMPLETED" db_nslcmop_update["detailed-status"] = "Done" + if scaling_type == "SCALE_IN" and prom_job_name is not None: + self.db.del_one( + "prometheus_jobs", + {"job_name": prom_job_name}, + fail_on_empty=False, + ) self._write_op_status( op_id=nslcmop_id, @@ -7960,10 +8058,25 @@ class NsLcm(LcmBase): exc_info=True, ) finally: - if tasks_dict_info: - stage[1] = "Waiting for healing pending tasks." - self.logger.debug(logging_text + stage[1]) - exc = await self._wait_for_tasks( + error_list = list() + if exc: + error_list.append(str(exc)) + try: + if tasks_dict_info: + stage[1] = "Waiting for healing pending tasks." + self.logger.debug(logging_text + stage[1]) + exc = await self._wait_for_tasks( + logging_text, + tasks_dict_info, + self.timeout.ns_deploy, + stage, + nslcmop_id, + nsr_id=nsr_id, + ) + except asyncio.CancelledError: + error_list.append("Cancelled") + await self._cancel_pending_tasks(logging_text, tasks_dict_info) + await self._wait_for_tasks( logging_text, tasks_dict_info, self.timeout.ns_deploy, @@ -7971,17 +8084,22 @@ class NsLcm(LcmBase): nslcmop_id, nsr_id=nsr_id, ) - if exc: + if error_list: + error_detail = "; ".join(error_list) db_nslcmop_update[ "detailed-status" - ] = error_description_nslcmop = "FAILED {}: {}".format(step, exc) + ] = error_description_nslcmop = "FAILED {}: {}".format( + step, error_detail + ) nslcmop_operation_state = "FAILED" if db_nsr: db_nsr_update["operational-status"] = old_operational_status db_nsr_update["config-status"] = old_config_status db_nsr_update[ "detailed-status" - ] = "FAILED healing nslcmop={} {}: {}".format(nslcmop_id, step, exc) + ] = "FAILED healing nslcmop={} {}: {}".format( + nslcmop_id, step, error_detail + ) for task, task_name in tasks_dict_info.items(): if not task.done() or task.cancelled() or task.exception(): if task_name.startswith(self.task_name_deploy_vca): @@ -8181,10 +8299,7 @@ class NsLcm(LcmBase): vca_type = "native_charm" elif ee_item.get("helm-chart"): vca_name = ee_item["helm-chart"] - if ee_item.get("helm-version") and ee_item.get("helm-version") == "v2": - vca_type = "helm" - else: - vca_type = "helm-v3" + vca_type = "helm-v3" else: self.logger.debug( logging_text + "skipping non juju neither charm configuration" @@ -8516,7 +8631,7 @@ class NsLcm(LcmBase): # if SSH access is required, then get execution environment SSH public # if native charm we have waited already to VM be UP - if vca_type in ("k8s_proxy_charm", "lxc_proxy_charm", "helm", "helm-v3"): + if vca_type in ("k8s_proxy_charm", "lxc_proxy_charm", "helm-v3"): pub_key = None user = None # self.logger.debug("get ssh key block") @@ -8724,6 +8839,10 @@ class NsLcm(LcmBase): db_nslcmop = None db_nslcmop_update = {} nslcmop_operation_state = None + old_db_update = {} + q_filter = {} + old_vdu_index = None + old_flavor_id = None db_nsr_update = {} target = {} exc = None @@ -8747,6 +8866,59 @@ class NsLcm(LcmBase): ) db_nslcmop = self.db.get_one("nslcmops", {"_id": nslcmop_id}) operationParams = db_nslcmop.get("operationParams") + # Update the VNFRS and NSRS with the requested flavour detail, So that ro tasks can function properly + db_nsr = self.db.get_one("nsrs", {"_id": nsr_id}) + db_flavor = db_nsr.get("flavor") + db_flavor_index = str(len(db_flavor)) + change_vnf_flavor_data = operationParams["changeVnfFlavorData"] + flavor_dict = change_vnf_flavor_data["additionalParams"] + count_index = flavor_dict["vduCountIndex"] + vdu_id_ref = flavor_dict["vduid"] + flavor_dict_update = { + "id": db_flavor_index, + "memory-mb": flavor_dict["virtualMemory"], + "name": f"{vdu_id_ref}-{count_index}-flv", + "storage-gb": flavor_dict["sizeOfStorage"], + "vcpu-count": flavor_dict["numVirtualCpu"], + } + db_flavor.append(flavor_dict_update) + db_update = {} + db_update["flavor"] = db_flavor + ns_q_filter = { + "_id": nsr_id, + } + self.db.set_one( + "nsrs", + q_filter=ns_q_filter, + update_dict=db_update, + fail_on_empty=True, + ) + db_vnfr = self.db.get_one( + "vnfrs", {"_id": change_vnf_flavor_data["vnfInstanceId"]} + ) + for vdu_index, vdur in enumerate(db_vnfr.get("vdur", ())): + if ( + vdur.get("count-index") == count_index + and vdur.get("vdu-id-ref") == vdu_id_ref + ): + old_flavor_id = vdur.get("ns-flavor-id", 0) + old_vdu_index = vdu_index + filter_text = { + "_id": change_vnf_flavor_data["vnfInstanceId"], + "vdur.count-index": count_index, + "vdur.vdu-id-ref": vdu_id_ref, + } + q_filter.update(filter_text) + db_update = {} + db_update[ + "vdur.{}.ns-flavor-id".format(vdu_index) + ] = db_flavor_index + self.db.set_one( + "vnfrs", + q_filter=q_filter, + update_dict=db_update, + fail_on_empty=True, + ) target = {} target.update(operationParams) desc = await self.RO.vertical_scale(nsr_id, target) @@ -8760,7 +8932,12 @@ class NsLcm(LcmBase): self.timeout.verticalscale, operation="verticalscale", ) - except (ROclient.ROClientException, DbException, LcmException) as e: + except ( + NgRoException, + ROclient.ROClientException, + DbException, + LcmException, + ) as e: self.logger.error("Exit Exception {}".format(e)) exc = e except asyncio.CancelledError: @@ -8781,6 +8958,9 @@ class NsLcm(LcmBase): if exc: db_nslcmop_update["detailed-status"] = "FAILED {}: {}".format(step, exc) nslcmop_operation_state = "FAILED" + old_db_update[ + "vdur.{}.ns-flavor-id".format(old_vdu_index) + ] = old_flavor_id else: nslcmop_operation_state = "COMPLETED" db_nslcmop_update["detailed-status"] = "Done" @@ -8793,6 +8973,16 @@ class NsLcm(LcmBase): operation_state=nslcmop_operation_state, other_update=db_nslcmop_update, ) + if old_vdu_index and old_db_update != {}: + self.logger.critical( + "Reverting Old Flavor -- : {}".format(old_db_update) + ) + self.db.set_one( + "vnfrs", + q_filter=q_filter, + update_dict=old_db_update, + fail_on_empty=True, + ) if nslcmop_operation_state: try: msg = {