X-Git-Url: https://osm.etsi.org/gitweb/?a=blobdiff_plain;f=osm_lcm%2Fns.py;h=5c2d4c3a3ddaf1a3b4b37ef89e339067467197ab;hb=e1ea267ea9baf033221e5d2f665cc6f4bbdc3028;hp=613163b598e77f76e8c1779c96635d30ad86508a;hpb=8d625f13e98a5ac75f8dde3c5449c31d3f36ac92;p=osm%2FLCM.git diff --git a/osm_lcm/ns.py b/osm_lcm/ns.py index 613163b..5c2d4c3 100644 --- a/osm_lcm/ns.py +++ b/osm_lcm/ns.py @@ -17,6 +17,7 @@ ## import asyncio +import shutil from typing import Any, Dict, List import yaml import logging @@ -55,6 +56,8 @@ from osm_lcm.lcm_utils import ( deep_get, get_iterable, populate_dict, + check_juju_bundle_existence, + get_charm_artifact_path, ) from osm_lcm.data_utils.nsd import ( get_ns_configuration_relation_list, @@ -62,6 +65,8 @@ from osm_lcm.data_utils.nsd import ( get_vnf_profiles, ) from osm_lcm.data_utils.vnfd import ( + get_kdu, + get_kdu_services, get_relation_list, get_vdu_list, get_vdu_profile, @@ -76,6 +81,7 @@ from osm_lcm.data_utils.vnfd import ( get_number_of_instances, get_juju_ee_ref, get_kdu_resource_profile, + find_software_version, ) from osm_lcm.data_utils.list_utils import find_in_list from osm_lcm.data_utils.vnfr import get_osm_params, get_vdur_index, get_kdur @@ -96,6 +102,7 @@ from n2vc.n2vc_juju_conn import N2VCJujuConnector from n2vc.exceptions import N2VCException, N2VCNotFound, K8sException from osm_lcm.lcm_helm_conn import LCMHelmConn +from osm_lcm.osm_config import OsmConfigBuilder from osm_lcm.prometheus import parse_job from copy import copy, deepcopy @@ -115,9 +122,11 @@ class NsLcm(LcmBase): timeout_ns_terminate = 1800 # default global timeout for un deployment a ns timeout_charm_delete = 10 * 60 timeout_primitive = 30 * 60 # timeout for primitive execution + timeout_ns_update = 30 * 60 # timeout for ns update timeout_progress_primitive = ( 10 * 60 ) # timeout for some progress in a primitive execution + timeout_migrate = 1800 # default global timeout for migrating vnfs SUBOPERATION_STATUS_NOT_FOUND = -1 SUBOPERATION_STATUS_NEW = -2 @@ -348,43 +357,49 @@ class NsLcm(LcmBase): self.logger.warn("Error updating NS state for ns={}: {}".format(nsr_id, e)) async def _on_update_k8s_db( - self, cluster_uuid, kdu_instance, filter=None, vca_id=None + self, cluster_uuid, kdu_instance, filter=None, vca_id=None, cluster_type="juju" ): """ Updating vca status in NSR record :param cluster_uuid: UUID of a k8s cluster :param kdu_instance: The unique name of the KDU instance :param filter: To get nsr_id + :cluster_type: The cluster type (juju, k8s) :return: none """ # self.logger.debug("_on_update_k8s_db(cluster_uuid={}, kdu_instance={}, filter={}" # .format(cluster_uuid, kdu_instance, filter)) + nsr_id = filter.get("_id") try: - nsr_id = filter.get("_id") - - # get vca status for NS - vca_status = await self.k8sclusterjuju.status_kdu( - cluster_uuid, - kdu_instance, - complete_status=True, + vca_status = await self.k8scluster_map[cluster_type].status_kdu( + cluster_uuid=cluster_uuid, + kdu_instance=kdu_instance, yaml_format=False, + complete_status=True, vca_id=vca_id, ) + # vcaStatus db_dict = dict() db_dict["vcaStatus"] = {nsr_id: vca_status} - await self.k8sclusterjuju.update_vca_status( - db_dict["vcaStatus"], - kdu_instance, - vca_id=vca_id, + if cluster_type in ("juju-bundle", "juju"): + # TODO -> this should be done in a more uniform way, I think in N2VC, in order to update the K8s VCA + # status in a similar way between Juju Bundles and Helm Charts on this side + await self.k8sclusterjuju.update_vca_status( + db_dict["vcaStatus"], + kdu_instance, + vca_id=vca_id, + ) + + self.logger.debug( + f"Obtained VCA status for cluster type '{cluster_type}': {vca_status}" ) # write to database self.update_db_2("nsrs", nsr_id, db_dict) - except (asyncio.CancelledError, asyncio.TimeoutError): raise except Exception as e: @@ -892,8 +907,12 @@ class NsLcm(LcmBase): for flavor in target["flavor"]: flavor["vim_info"] = {} if db_nsr.get("affinity-or-anti-affinity-group"): - target["affinity-or-anti-affinity-group"] = deepcopy(db_nsr["affinity-or-anti-affinity-group"]) - for affinity_or_anti_affinity_group in target["affinity-or-anti-affinity-group"]: + target["affinity-or-anti-affinity-group"] = deepcopy( + db_nsr["affinity-or-anti-affinity-group"] + ) + for affinity_or_anti_affinity_group in target[ + "affinity-or-anti-affinity-group" + ]: affinity_or_anti_affinity_group["vim_info"] = {} if db_nslcmop.get("lcmOperationType") != "instantiate": @@ -1409,7 +1428,7 @@ class NsLcm(LcmBase): :param nsr_id: :param vnfr_id: :param kdu_name: - :return: IP address + :return: IP address, K8s services """ # self.logger.debug(logging_text + "Starting wait_kdu_up") @@ -1431,7 +1450,7 @@ class NsLcm(LcmBase): ) if kdur.get("status"): if kdur["status"] in ("READY", "ENABLED"): - return kdur.get("ip-address") + return kdur.get("ip-address"), kdur.get("services") else: raise LcmException( "target KDU={} is in error state".format(kdu_name) @@ -1974,9 +1993,33 @@ class NsLcm(LcmBase): # wait for RO (ip-address) Insert pub_key into VM if vnfr_id: if kdu_name: - rw_mgmt_ip = await self.wait_kdu_up( + rw_mgmt_ip, services = await self.wait_kdu_up( logging_text, nsr_id, vnfr_id, kdu_name ) + vnfd = self.db.get_one( + "vnfds_revisions", + {"_id": f'{db_vnfr["vnfd-id"]}:{db_vnfr["revision"]}'}, + ) + kdu = get_kdu(vnfd, kdu_name) + kdu_services = [ + service["name"] for service in get_kdu_services(kdu) + ] + exposed_services = [] + for service in services: + if any(s in service["name"] for s in kdu_services): + exposed_services.append(service) + await self.vca_map[vca_type].exec_primitive( + ee_id=ee_id, + primitive_name="config", + params_dict={ + "osm-config": json.dumps( + OsmConfigBuilder( + k8s={"services": exposed_services} + ).build() + ) + }, + vca_id=vca_id, + ) else: rw_mgmt_ip = await self.wait_vm_up_insert_key_ro( logging_text, @@ -1987,6 +2030,7 @@ class NsLcm(LcmBase): user=user, pub_key=pub_key, ) + else: rw_mgmt_ip = None # This is for a NS configuration @@ -2381,7 +2425,9 @@ class NsLcm(LcmBase): kdur_list = [] for kdur in vnfr["kdur"]: if kdur.get("additionalParams"): - kdur["additionalParams"] = json.loads(kdur["additionalParams"]) + kdur["additionalParams"] = json.loads( + kdur["additionalParams"] + ) kdur_list.append(kdur) vnfr["kdur"] = kdur_list @@ -2579,8 +2625,8 @@ class NsLcm(LcmBase): ) deploy_params_kdu = {"OSM": get_osm_params(db_vnfr)} if kdur.get("additionalParams"): - deploy_params_kdu = parse_yaml_strings( - kdur["additionalParams"] + deploy_params_kdu.update( + parse_yaml_strings(kdur["additionalParams"].copy()) ) self._deploy_n2vc( @@ -4441,6 +4487,7 @@ class NsLcm(LcmBase): cluster_uuid=kdu.get("k8scluster-uuid"), kdu_instance=kdu_instance, vca_id=vca_id, + namespace=kdu.get("namespace"), ) ) else: @@ -4866,10 +4913,18 @@ class NsLcm(LcmBase): db_nsr = self.db.get_one("nsrs", {"_id": nsr_id}) vca_id = self.get_vca_id({}, db_nsr) if db_nsr["_admin"]["deployed"]["K8s"]: - for k8s_index, k8s in enumerate(db_nsr["_admin"]["deployed"]["K8s"]): - cluster_uuid, kdu_instance = k8s["k8scluster-uuid"], k8s["kdu-instance"] + for _, k8s in enumerate(db_nsr["_admin"]["deployed"]["K8s"]): + cluster_uuid, kdu_instance, cluster_type = ( + k8s["k8scluster-uuid"], + k8s["kdu-instance"], + k8s["k8scluster-type"], + ) await self._on_update_k8s_db( - cluster_uuid, kdu_instance, filter={"_id": nsr_id}, vca_id=vca_id + cluster_uuid=cluster_uuid, + kdu_instance=kdu_instance, + filter={"_id": nsr_id}, + vca_id=vca_id, + cluster_type=cluster_type, ) else: for vca_index, _ in enumerate(db_nsr["_admin"]["deployed"]["VCA"]): @@ -4936,7 +4991,9 @@ class NsLcm(LcmBase): kdur_list = [] for kdur in db_vnfr["kdur"]: if kdur.get("additionalParams"): - kdur["additionalParams"] = json.loads(kdur["additionalParams"]) + kdur["additionalParams"] = json.loads( + kdur["additionalParams"] + ) kdur_list.append(kdur) db_vnfr["kdur"] = kdur_list step = "Getting vnfd from database" @@ -5015,7 +5072,17 @@ class NsLcm(LcmBase): actions.add(primitive["name"]) for primitive in kdu_configuration.get("config-primitive", []): actions.add(primitive["name"]) - kdu_action = True if primitive_name in actions else False + kdu = find_in_list( + nsr_deployed["K8s"], + lambda kdu: kdu_name == kdu["kdu-name"] + and kdu["member-vnf-index"] == vnf_index, + ) + kdu_action = ( + True + if primitive_name in actions + and kdu["k8scluster-type"] not in ("helm-chart", "helm-chart-v3") + else False + ) # TODO check if ns is in a proper status if kdu_name and ( @@ -5232,6 +5299,375 @@ class NsLcm(LcmBase): self.lcm_tasks.remove("ns", nsr_id, nslcmop_id, "ns_action") return nslcmop_operation_state, detailed_status + async def _ns_charm_upgrade( + self, + ee_id, + charm_id, + charm_type, + path, + timeout: float = None, + ) -> (str, str): + """This method upgrade charms in VNF instances + + Args: + ee_id: Execution environment id + path: Local path to the charm + charm_id: charm-id + charm_type: Charm type can be lxc-proxy-charm, native-charm or k8s-proxy-charm + timeout: (Float) Timeout for the ns update operation + + Returns: + result: (str, str) COMPLETED/FAILED, details + """ + try: + charm_type = charm_type or "lxc_proxy_charm" + output = await self.vca_map[charm_type].upgrade_charm( + ee_id=ee_id, + path=path, + charm_id=charm_id, + charm_type=charm_type, + timeout=timeout or self.timeout_ns_update, + ) + + if output: + return "COMPLETED", output + + except (LcmException, asyncio.CancelledError): + raise + + except Exception as e: + + self.logger.debug("Error upgrading charm {}".format(path)) + + return "FAILED", "Error upgrading charm {}: {}".format(path, e) + + async def update(self, nsr_id, nslcmop_id): + """Update NS according to different update types + + This method performs upgrade of VNF instances then updates the revision + number in VNF record + + Args: + nsr_id: Network service will be updated + nslcmop_id: ns lcm operation id + + Returns: + It may raise DbException, LcmException, N2VCException, K8sException + + """ + # 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: + return + + logging_text = "Task ns={} update={} ".format(nsr_id, nslcmop_id) + self.logger.debug(logging_text + "Enter") + + # Set the required variables to be filled up later + db_nsr = None + db_nslcmop_update = {} + vnfr_update = {} + nslcmop_operation_state = None + db_nsr_update = {} + error_description_nslcmop = "" + exc = None + change_type = "" + detailed_status = "" + + try: + # wait for any previous tasks in process + step = "Waiting for previous operations to terminate" + await self.lcm_tasks.waitfor_related_HA("ns", "nslcmops", nslcmop_id) + self._write_ns_status( + nsr_id=nsr_id, + ns_state=None, + current_operation="UPDATING", + current_operation_id=nslcmop_id, + ) + + step = "Getting nslcmop from database" + db_nslcmop = self.db.get_one( + "nslcmops", {"_id": nslcmop_id}, fail_on_empty=False + ) + update_type = db_nslcmop["operationParams"]["updateType"] + + step = "Getting nsr from database" + db_nsr = self.db.get_one("nsrs", {"_id": nsr_id}) + old_operational_status = db_nsr["operational-status"] + db_nsr_update["operational-status"] = "updating" + self.update_db_2("nsrs", nsr_id, db_nsr_update) + nsr_deployed = db_nsr["_admin"].get("deployed") + + if update_type == "CHANGE_VNFPKG": + + # Get the input parameters given through update request + vnf_instance_id = db_nslcmop["operationParams"][ + "changeVnfPackageData" + ].get("vnfInstanceId") + + vnfd_id = db_nslcmop["operationParams"]["changeVnfPackageData"].get( + "vnfdId" + ) + timeout_seconds = db_nslcmop["operationParams"].get("timeout_ns_update") + + step = "Getting vnfr from database" + db_vnfr = self.db.get_one( + "vnfrs", {"_id": vnf_instance_id}, fail_on_empty=False + ) + + step = "Getting vnfds from database" + # Latest VNFD + latest_vnfd = self.db.get_one( + "vnfds", {"_id": vnfd_id}, fail_on_empty=False + ) + latest_vnfd_revision = latest_vnfd["_admin"].get("revision") + + # Current VNFD + current_vnf_revision = db_vnfr.get("revision", 1) + current_vnfd = self.db.get_one( + "vnfds_revisions", + {"_id": vnfd_id + ":" + str(current_vnf_revision)}, + fail_on_empty=False, + ) + # Charm artifact paths will be filled up later + ( + current_charm_artifact_path, + target_charm_artifact_path, + charm_artifact_paths, + ) = ([], [], []) + + step = "Checking if revision has changed in VNFD" + if current_vnf_revision != latest_vnfd_revision: + + # There is new revision of VNFD, update operation is required + current_vnfd_path = vnfd_id + ":" + str(current_vnf_revision) + latest_vnfd_path = vnfd_id + + step = "Removing the VNFD packages if they exist in the local path" + shutil.rmtree(self.fs.path + current_vnfd_path, ignore_errors=True) + shutil.rmtree(self.fs.path + latest_vnfd_path, ignore_errors=True) + + step = "Get the VNFD packages from FSMongo" + self.fs.sync(from_path=latest_vnfd_path) + self.fs.sync(from_path=current_vnfd_path) + + step = ( + "Get the charm-type, charm-id, ee-id if there is deployed VCA" + ) + base_folder = latest_vnfd["_admin"]["storage"] + + for charm_index, charm_deployed in enumerate( + get_iterable(nsr_deployed, "VCA") + ): + vnf_index = db_vnfr.get("member-vnf-index-ref") + + # Getting charm-id and charm-type + if charm_deployed.get("member-vnf-index") == vnf_index: + charm_id = self.get_vca_id(db_vnfr, db_nsr) + charm_type = charm_deployed.get("type") + + # Getting ee-id + ee_id = charm_deployed.get("ee_id") + + step = "Getting descriptor config" + descriptor_config = get_configuration( + current_vnfd, current_vnfd["id"] + ) + + if "execution-environment-list" in descriptor_config: + ee_list = descriptor_config.get( + "execution-environment-list", [] + ) + else: + ee_list = [] + + # There could be several charm used in the same VNF + for ee_item in ee_list: + if ee_item.get("juju"): + + step = "Getting charm name" + charm_name = ee_item["juju"].get("charm") + + step = "Setting Charm artifact paths" + current_charm_artifact_path.append( + get_charm_artifact_path( + base_folder, + charm_name, + charm_type, + current_vnf_revision, + ) + ) + target_charm_artifact_path.append( + get_charm_artifact_path( + base_folder, + charm_name, + charm_type, + ) + ) + + charm_artifact_paths = zip( + current_charm_artifact_path, target_charm_artifact_path + ) + + step = "Checking if software version has changed in VNFD" + if find_software_version(current_vnfd) != find_software_version( + latest_vnfd + ): + + step = "Checking if existing VNF has charm" + for current_charm_path, target_charm_path in list( + charm_artifact_paths + ): + if current_charm_path: + raise LcmException( + "Software version change is not supported as VNF instance {} has charm.".format( + vnf_instance_id + ) + ) + + # There is no change in the charm package, then redeploy the VNF + # based on new descriptor + step = "Redeploying VNF" + # This part is in https://osm.etsi.org/gerrit/11943 + + else: + step = "Checking if any charm package has changed or not" + for current_charm_path, target_charm_path in list( + charm_artifact_paths + ): + if ( + current_charm_path + and target_charm_path + and self.check_charm_hash_changed( + current_charm_path, target_charm_path + ) + ): + + step = "Checking whether VNF uses juju bundle" + if check_juju_bundle_existence(current_vnfd): + + raise LcmException( + "Charm upgrade is not supported for the instance which" + " uses juju-bundle: {}".format( + check_juju_bundle_existence(current_vnfd) + ) + ) + + step = "Upgrading Charm" + ( + result, + detailed_status, + ) = await self._ns_charm_upgrade( + ee_id=ee_id, + charm_id=charm_id, + charm_type=charm_type, + path=self.fs.path + target_charm_path, + timeout=timeout_seconds, + ) + + if result == "FAILED": + nslcmop_operation_state = result + error_description_nslcmop = detailed_status + + db_nslcmop_update["detailed-status"] = detailed_status + self.logger.debug( + logging_text + + " step {} Done with result {} {}".format( + step, nslcmop_operation_state, detailed_status + ) + ) + + step = "Updating policies" + # This part is in https://osm.etsi.org/gerrit/11943 + + # If nslcmop_operation_state is None, so any operation is not failed. + if not nslcmop_operation_state: + nslcmop_operation_state = "COMPLETED" + + # If update CHANGE_VNFPKG nslcmop_operation is successful + # vnf revision need to be updated + vnfr_update["revision"] = latest_vnfd_revision + self.update_db_2("vnfrs", db_vnfr["_id"], vnfr_update) + + self.logger.debug( + logging_text + + " task Done with result {} {}".format( + nslcmop_operation_state, detailed_status + ) + ) + elif update_type == "REMOVE_VNF": + # This part is included in https://osm.etsi.org/gerrit/11876 + pass + + # If nslcmop_operation_state is None, so any operation is not failed. + # All operations are executed in overall. + if not nslcmop_operation_state: + nslcmop_operation_state = "COMPLETED" + db_nsr_update["operational-status"] = old_operational_status + + except (DbException, LcmException, N2VCException, K8sException) as e: + self.logger.error(logging_text + "Exit Exception {}".format(e)) + exc = e + except asyncio.CancelledError: + self.logger.error( + logging_text + "Cancelled Exception while '{}'".format(step) + ) + exc = "Operation was cancelled" + except asyncio.TimeoutError: + self.logger.error(logging_text + "Timeout while '{}'".format(step)) + exc = "Timeout" + except Exception as e: + exc = traceback.format_exc() + self.logger.critical( + logging_text + "Exit Exception {} {}".format(type(e).__name__, e), + exc_info=True, + ) + finally: + if exc: + db_nslcmop_update[ + "detailed-status" + ] = ( + detailed_status + ) = error_description_nslcmop = "FAILED {}: {}".format(step, exc) + nslcmop_operation_state = "FAILED" + db_nsr_update["operational-status"] = old_operational_status + if db_nsr: + self._write_ns_status( + nsr_id=nsr_id, + ns_state=db_nsr["nsState"], + current_operation="IDLE", + current_operation_id=None, + other_update=db_nsr_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: + await self.msg.aiowrite( + "ns", + "updated", + { + "nsr_id": nsr_id, + "nslcmop_id": nslcmop_id, + "operationState": nslcmop_operation_state, + }, + loop=self.loop, + ) + except Exception as e: + self.logger.error( + logging_text + "kafka_write notification Exception {}".format(e) + ) + self.logger.debug(logging_text + "Exit") + self.lcm_tasks.remove("ns", nsr_id, nslcmop_id, "ns_update") + return nslcmop_operation_state, detailed_status + async def scale(self, nsr_id, nslcmop_id): # Try to lock HA task here task_is_locked_by_me = self.lcm_tasks.lock_HA("ns", "nslcmops", nslcmop_id) @@ -6475,3 +6911,99 @@ class NsLcm(LcmBase): """ config = VimAccountDB.get_vim_account_with_id(vim_account_id).get("config", {}) return config.get("vca_k8s_cloud"), config.get("vca_k8s_cloud_credential") + + async def migrate(self, nsr_id, nslcmop_id): + """ + Migrate VNFs and VDUs instances in a NS + + :param: nsr_id: NS Instance ID + :param: nslcmop_id: nslcmop ID of migrate + + """ + # 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: + return + logging_text = "Task ns={} migrate ".format(nsr_id) + self.logger.debug(logging_text + "Enter") + # get all needed from database + db_nslcmop = None + db_nslcmop_update = {} + nslcmop_operation_state = None + db_nsr_update = {} + target = {} + exc = None + # in case of error, indicates what part of scale was failed to put nsr at error status + start_deploy = time() + + try: + # wait for any previous tasks in process + step = "Waiting for previous operations to terminate" + await self.lcm_tasks.waitfor_related_HA('ns', 'nslcmops', nslcmop_id) + + self._write_ns_status( + nsr_id=nsr_id, + ns_state=None, + current_operation="MIGRATING", + current_operation_id=nslcmop_id + ) + step = "Getting nslcmop from database" + self.logger.debug(step + " after having waited for previous tasks to be completed") + db_nslcmop = self.db.get_one("nslcmops", {"_id": nslcmop_id}) + migrate_params = db_nslcmop.get("operationParams") + + target = {} + target.update(migrate_params) + desc = await self.RO.migrate(nsr_id, target) + self.logger.debug("RO return > {}".format(desc)) + action_id = desc["action_id"] + await self._wait_ng_ro( + nsr_id, action_id, nslcmop_id, start_deploy, self.timeout_migrate + ) + except (ROclient.ROClientException, DbException, LcmException) as e: + self.logger.error("Exit Exception {}".format(e)) + exc = e + except asyncio.CancelledError: + self.logger.error("Cancelled Exception while '{}'".format(step)) + exc = "Operation was cancelled" + except Exception as e: + exc = traceback.format_exc() + self.logger.critical("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: + db_nslcmop_update[ + "detailed-status" + ] = "FAILED {}: {}".format(step, exc) + nslcmop_operation_state = "FAILED" + else: + nslcmop_operation_state = "COMPLETED" + db_nslcmop_update["detailed-status"] = "Done" + db_nsr_update["detailed-status"] = "Done" + + self._write_op_status( + op_id=nslcmop_id, + stage="", + error_message="", + operation_state=nslcmop_operation_state, + other_update=db_nslcmop_update, + ) + if nslcmop_operation_state: + try: + msg = { + "nsr_id": nsr_id, + "nslcmop_id": nslcmop_id, + "operationState": nslcmop_operation_state, + } + await self.msg.aiowrite("ns", "migrated", msg, loop=self.loop) + except Exception as e: + self.logger.error( + logging_text + "kafka_write notification Exception {}".format(e) + ) + self.logger.debug(logging_text + "Exit") + self.lcm_tasks.remove("ns", nsr_id, nslcmop_id, "ns_migrate")