X-Git-Url: https://osm.etsi.org/gitweb/?a=blobdiff_plain;f=osm_lcm%2Fns.py;h=198def1010c17c0a872e1ad8f7a750e265e37b0d;hb=aae391fc83f3b20e058769ff3356ca4d9965a3b8;hp=21c3d57a85daf15f6726c0819d4d0add841b1774;hpb=d6b2b51993c9d983c20e893036f307648ae177c8;p=osm%2FLCM.git diff --git a/osm_lcm/ns.py b/osm_lcm/ns.py index 21c3d57..198def1 100644 --- a/osm_lcm/ns.py +++ b/osm_lcm/ns.py @@ -21,65 +21,48 @@ import yaml import logging import logging.handlers import traceback -from jinja2 import Environment, Template, meta, TemplateError, TemplateNotFound, TemplateSyntaxError +import json +from jinja2 import Environment, TemplateError, TemplateNotFound, StrictUndefined, UndefinedError from osm_lcm import ROclient -from osm_lcm.lcm_utils import LcmException, LcmExceptionNoMgmtIP, LcmBase, deep_get +from osm_lcm.ng_ro import NgRoClient, NgRoException +from osm_lcm.lcm_utils import LcmException, LcmExceptionNoMgmtIP, LcmBase, deep_get, get_iterable, populate_dict from n2vc.k8s_helm_conn import K8sHelmConnector +from n2vc.k8s_helm3_conn import K8sHelm3Connector from n2vc.k8s_juju_conn import K8sJujuConnector from osm_common.dbbase import DbException from osm_common.fsbase import FsException from n2vc.n2vc_juju_conn import N2VCJujuConnector +from n2vc.exceptions import N2VCException, N2VCNotFound, K8sException + +from osm_lcm.lcm_helm_conn import LCMHelmConn from copy import copy, deepcopy from http import HTTPStatus from time import time from uuid import uuid4 -__author__ = "Alfonso Tierno" - - -def get_iterable(in_dict, in_key): - """ - Similar to .get(), but if value is None, False, ..., An empty tuple is returned instead - :param in_dict: a dictionary - :param in_key: the key to look for at in_dict - :return: in_dict[in_var] or () if it is None or not present - """ - if not in_dict.get(in_key): - return () - return in_dict[in_key] - - -def populate_dict(target_dict, key_list, value): - """ - Update target_dict creating nested dictionaries with the key_list. Last key_list item is asigned the value. - Example target_dict={K: J}; key_list=[a,b,c]; target_dict will be {K: J, a: {b: {c: value}}} - :param target_dict: dictionary to be changed - :param key_list: list of keys to insert at target_dict - :param value: - :return: None - """ - for key in key_list[0:-1]: - if key not in target_dict: - target_dict[key] = {} - target_dict = target_dict[key] - target_dict[key_list[-1]] = value +from random import randint + +__author__ = "Alfonso Tierno " class NsLcm(LcmBase): timeout_vca_on_error = 5 * 60 # Time for charm from first time at blocked,error status to mark as failed - total_deploy_timeout = 2 * 3600 # global timeout for deployment + 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_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 SUBOPERATION_STATUS_SKIP = -3 + task_name_deploy_vca = "Deploying VCA" - def __init__(self, db, msg, fs, lcm_tasks, ro_config, vca_config, loop): + def __init__(self, db, msg, fs, lcm_tasks, config, loop, prometheus=None): """ Init, Connect to database, filesystem storage, and messaging :param config: two level dictionary with configuration. Top level should contain 'database', 'storage', @@ -94,14 +77,10 @@ class NsLcm(LcmBase): self.loop = loop self.lcm_tasks = lcm_tasks - self.ro_config = ro_config - self.vca_config = vca_config - if 'pubkey' in self.vca_config: - self.vca_config['public_key'] = self.vca_config['pubkey'] - if 'cacert' in self.vca_config: - self.vca_config['ca_cert'] = self.vca_config['cacert'] - if 'apiproxy' in self.vca_config: - self.vca_config['api_proxy'] = self.vca_config['apiproxy'] + self.timeout = config["timeout"] + self.ro_config = config["ro_config"] + self.ng_ro = config["ro_config"].get("ng") + self.vca_config = config["VCA"].copy() # create N2VC connector self.n2vc = N2VCJujuConnector( @@ -112,12 +91,21 @@ 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.conn_helm_ee = LCMHelmConn( + db=self.db, + fs=self.fs, + log=self.logger, + loop=self.loop, + url=None, + username=None, + vca_config=self.vca_config, + on_update_db=self._on_update_n2vc_db ) - self.k8sclusterhelm = K8sHelmConnector( + self.k8sclusterhelm2 = K8sHelmConnector( kubectl_command=self.vca_config.get("kubectlpath"), helm_command=self.vca_config.get("helmpath"), fs=self.fs, @@ -126,44 +114,217 @@ class NsLcm(LcmBase): on_update_db=None, ) + self.k8sclusterhelm3 = K8sHelm3Connector( + kubectl_command=self.vca_config.get("kubectlpath"), + helm_command=self.vca_config.get("helm3path"), + fs=self.fs, + log=self.logger, + db=self.db, + on_update_db=None, + ) + self.k8sclusterjuju = K8sJujuConnector( kubectl_command=self.vca_config.get("kubectlpath"), juju_command=self.vca_config.get("jujupath"), fs=self.fs, log=self.logger, db=self.db, + loop=self.loop, on_update_db=None, + vca_config=self.vca_config, ) + self.k8scluster_map = { + "helm-chart": self.k8sclusterhelm2, + "helm-chart-v3": self.k8sclusterhelm3, + "chart": self.k8sclusterhelm3, + "juju-bundle": self.k8sclusterjuju, + "juju": self.k8sclusterjuju, + } + + self.vca_map = { + "lxc_proxy_charm": self.n2vc, + "native_charm": self.n2vc, + "k8s_proxy_charm": self.n2vc, + "helm": self.conn_helm_ee, + "helm-v3": self.conn_helm_ee + } + + self.prometheus = prometheus + # create RO client - self.RO = ROclient.ROClient(self.loop, **self.ro_config) + if self.ng_ro: + self.RO = NgRoClient(self.loop, **self.ro_config) + else: + 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)) - 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)) + 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): + + # remove last dot from path (if exists) + if path.endswith('.'): + path = path[:-1] + + # 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) + + # 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 (asyncio.CancelledError, asyncio.TimeoutError): + raise + except Exception as e: + self.logger.warn('Error updating NS state for ns={}: {}'.format(nsr_id, e)) + + @staticmethod + def _parse_cloud_init(cloud_init_text, additional_params, vnfd_id, vdu_id): + try: + env = Environment(undefined=StrictUndefined) + template = env.from_string(cloud_init_text) + return template.render(additional_params or {}) + except UndefinedError as e: + raise LcmException("Variable {} at vnfd[id={}]:vdu[id={}]:cloud-init/cloud-init-" + "file, must be provided in the instantiation parameters inside the " + "'additionalParamsForVnf/Vdu' block".format(e, vnfd_id, vdu_id)) + except (TemplateError, TemplateNotFound) as e: + raise LcmException("Error parsing Jinja2 to cloud-init content at vnfd[id={}]:vdu[id={}]: {}". + format(vnfd_id, vdu_id, e)) + + def _get_cloud_init(self, vdu, vnfd): + try: + cloud_init_content = cloud_init_file = None + if vdu.get("cloud-init-file"): + base_folder = vnfd["_admin"]["storage"] + cloud_init_file = "{}/{}/cloud_init/{}".format(base_folder["folder"], base_folder["pkg-dir"], + vdu["cloud-init-file"]) + with self.fs.file_open(cloud_init_file, "r") as ci_file: + cloud_init_content = ci_file.read() + elif vdu.get("cloud-init"): + cloud_init_content = vdu["cloud-init"] + + return cloud_init_content + except FsException as e: + raise LcmException("Error reading vnfd[id={}]:vdu[id={}]:cloud-init-file={}: {}". + format(vnfd["id"], vdu["id"], cloud_init_file, e)) + + def _get_osm_params(self, db_vnfr, vdu_id=None, vdu_count_index=0): + osm_params = {x.replace("-", "_"): db_vnfr[x] for x in ("ip-address", "vim-account-id", "vnfd-id", "vnfd-ref") + if db_vnfr.get(x) is not None} + osm_params["ns_id"] = db_vnfr["nsr-id-ref"] + osm_params["vnf_id"] = db_vnfr["_id"] + osm_params["member_vnf_index"] = db_vnfr["member-vnf-index-ref"] + if db_vnfr.get("vdur"): + osm_params["vdu"] = {} + for vdur in db_vnfr["vdur"]: + vdu = { + "count_index": vdur["count-index"], + "vdu_id": vdur["vdu-id-ref"], + "interfaces": {} + } + if vdur.get("ip-address"): + vdu["ip_address"] = vdur["ip-address"] + for iface in vdur["interfaces"]: + vdu["interfaces"][iface["name"]] = \ + {x.replace("-", "_"): iface[x] for x in ("mac-address", "ip-address", "vnf-vld-id", "name") + if iface.get(x) is not None} + vdu_id_index = "{}-{}".format(vdur["vdu-id-ref"], vdur["count-index"]) + osm_params["vdu"][vdu_id_index] = vdu + if vdu_id: + osm_params["vdu_id"] = vdu_id + osm_params["count_index"] = vdu_count_index + return osm_params + + def _get_vdu_additional_params(self, db_vnfr, vdu_id): + vdur = next(vdur for vdur in db_vnfr.get("vdur") if vdu_id == vdur["vdu-id-ref"]) + additional_params = vdur.get("additionalParams") + return self._format_additional_params(additional_params) def vnfd2RO(self, vnfd, new_id=None, additionalParams=None, nsrId=None): """ @@ -174,59 +335,30 @@ class NsLcm(LcmBase): :param nsrId: Id of the NSR :return: copy of vnfd """ - try: - vnfd_RO = deepcopy(vnfd) - # remove unused by RO configuration, monitoring, scaling and internal keys - vnfd_RO.pop("_id", None) - vnfd_RO.pop("_admin", None) - vnfd_RO.pop("vnf-configuration", None) - vnfd_RO.pop("monitoring-param", None) - vnfd_RO.pop("scaling-group-descriptor", None) - vnfd_RO.pop("kdu", None) - vnfd_RO.pop("k8s-cluster", None) - if new_id: - vnfd_RO["id"] = new_id - - # parse cloud-init or cloud-init-file with the provided variables using Jinja2 - for vdu in get_iterable(vnfd_RO, "vdu"): - cloud_init_file = None - if vdu.get("cloud-init-file"): - base_folder = vnfd["_admin"]["storage"] - cloud_init_file = "{}/{}/cloud_init/{}".format(base_folder["folder"], base_folder["pkg-dir"], - vdu["cloud-init-file"]) - with self.fs.file_open(cloud_init_file, "r") as ci_file: - cloud_init_content = ci_file.read() - vdu.pop("cloud-init-file", None) - elif vdu.get("cloud-init"): - cloud_init_content = vdu["cloud-init"] - else: - continue - - env = Environment() - ast = env.parse(cloud_init_content) - mandatory_vars = meta.find_undeclared_variables(ast) - if mandatory_vars: - for var in mandatory_vars: - if not additionalParams or var not in additionalParams.keys(): - raise LcmException("Variable '{}' defined at vnfd[id={}]:vdu[id={}]:cloud-init/cloud-init-" - "file, must be provided in the instantiation parameters inside the " - "'additionalParamsForVnf' block".format(var, vnfd["id"], vdu["id"])) - template = Template(cloud_init_content) - cloud_init_content = template.render(additionalParams or {}) - vdu["cloud-init"] = cloud_init_content - - return vnfd_RO - except FsException as e: - raise LcmException("Error reading vnfd[id={}]:vdu[id={}]:cloud-init-file={}: {}". - format(vnfd["id"], vdu["id"], cloud_init_file, e)) - except (TemplateError, TemplateNotFound, TemplateSyntaxError) as e: - 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): + vnfd_RO = deepcopy(vnfd) + # remove unused by RO configuration, monitoring, scaling and internal keys + vnfd_RO.pop("_id", None) + vnfd_RO.pop("_admin", None) + vnfd_RO.pop("vnf-configuration", None) + vnfd_RO.pop("monitoring-param", None) + vnfd_RO.pop("scaling-group-descriptor", None) + vnfd_RO.pop("kdu", None) + vnfd_RO.pop("k8s-cluster", None) + if new_id: + vnfd_RO["id"] = new_id + + # parse cloud-init or cloud-init-file with the provided variables using Jinja2 + for vdu in get_iterable(vnfd_RO, "vdu"): + vdu.pop("cloud-init-file", None) + vdu.pop("cloud-init", None) + return vnfd_RO + + 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 = {} @@ -288,6 +420,13 @@ 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(): @@ -322,6 +461,25 @@ class NsLcm(LcmBase): populate_dict(RO_ns_params, ("vnfs", vnf_member["member-vnf-index"], "vdus", vdu, "mgmt_keys"), n2vc_key_list) + # cloud init + for vdu in get_iterable(vnfd, "vdu"): + cloud_init_text = self._get_cloud_init(vdu, vnfd) + if not cloud_init_text: + continue + for vnf_member in nsd.get("constituent-vnfd"): + if vnf_member["vnfd-id-ref"] != vnfd_ref: + continue + db_vnfr = db_vnfrs[vnf_member["member-vnf-index"]] + additional_params = self._get_vdu_additional_params(db_vnfr, vdu["id"]) or {} + + cloud_init_list = [] + for vdu_index in range(0, int(vdu.get("count", 1))): + additional_params["OSM"] = self._get_osm_params(db_vnfr, vdu["id"], vdu_index) + cloud_init_list.append(self._parse_cloud_init(cloud_init_text, additional_params, vnfd["id"], + vdu["id"])) + populate_dict(RO_ns_params, + ("vnfs", vnf_member["member-vnf-index"], "vdus", vdu["id"], "cloud_init"), + cloud_init_list) if ns_params.get("vduImage"): RO_ns_params["vduImage"] = ns_params["vduImage"] @@ -336,9 +494,6 @@ 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 @@ -454,8 +609,7 @@ class NsLcm(LcmBase): if isinstance(vld_params["ns-net"], dict): for vld_id, instance_scenario_id in vld_params["ns-net"].items(): RO_vld_ns_net = {"instance_scenario_id": instance_scenario_id, "osm_id": vld_id} - if RO_vld_ns_net: - populate_dict(RO_ns_params, ("networks", vld_params["name"], "use-network"), RO_vld_ns_net) + populate_dict(RO_ns_params, ("networks", vld_params["name"], "use-network"), RO_vld_ns_net) if "vnfd-connection-point-ref" in vld_params: for cp_params in vld_params["vnfd-connection-point-ref"]: # look for interface @@ -511,11 +665,21 @@ class NsLcm(LcmBase): continue vdu_id_ref = vdur["vdu-id-ref"] if vdu_create and vdu_create.get(vdu_id_ref): + vdur_copy = deepcopy(vdur) + vdur_copy["status"] = "BUILD" + vdur_copy["status-detailed"] = None + vdur_copy["ip_address"]: None + for iface in vdur_copy["interfaces"]: + iface["ip-address"] = None + iface["mac-address"] = None + iface.pop("mgmt_vnf", None) # only first vdu can be managment of vnf # TODO ALF for index in range(0, vdu_create[vdu_id_ref]): - vdur = deepcopy(vdur) - vdur["_id"] = str(uuid4()) - vdur["count-index"] += 1 - vdurs.insert(vdu_index+1+index, vdur) + vdur_copy["_id"] = str(uuid4()) + vdur_copy["count-index"] += 1 + vdurs.insert(vdu_index+1+index, vdur_copy) + self.logger.debug("scale out, adding vdu={}".format(vdur_copy)) + vdur_copy = deepcopy(vdur_copy) + del vdu_create[vdu_id_ref] if vdu_delete and vdu_delete.get(vdu_id_ref): del vdurs[vdu_index] @@ -556,6 +720,21 @@ class NsLcm(LcmBase): else: raise LcmException("ns_update_nsr: Not found vld={} at RO info".format(vld["id"])) + def set_vnfr_at_error(self, db_vnfrs, error_text): + try: + for db_vnfr in db_vnfrs.values(): + vnfr_update = {"status": "ERROR"} + for vdu_index, vdur in enumerate(get_iterable(db_vnfr, "vdur")): + if "status" not in vdur: + vdur["status"] = "ERROR" + vnfr_update["vdur.{}.status".format(vdu_index)] = "ERROR" + if error_text: + vdur["status-detailed"] = str(error_text) + vnfr_update["vdur.{}.status-detailed".format(vdu_index)] = "ERROR" + self.update_db_2("vnfrs", db_vnfr["_id"], vnfr_update) + except DbException as e: + self.logger.error("Cannot update vnf. {}".format(e)) + def ns_update_vnfr(self, db_vnfrs, nsr_desc_RO): """ Updates database vnfr with the RO info, e.g. ip_address, vim_id... Descriptor db_vnfrs is also updated @@ -571,7 +750,8 @@ class NsLcm(LcmBase): if vnf_RO.get("ip_address"): db_vnfr["ip-address"] = vnfr_update["ip-address"] = vnf_RO["ip_address"].split(";")[0] elif not db_vnfr.get("ip-address"): - raise LcmExceptionNoMgmtIP("ns member_vnf_index '{}' has no IP address".format(vnf_index)) + if db_vnfr.get("vdur"): # if not VDUs, there is not ip_address + raise LcmExceptionNoMgmtIP("ns member_vnf_index '{}' has no IP address".format(vnf_index)) for vdu_index, vdur in enumerate(get_iterable(db_vnfr, "vdur")): vdur_RO_count_index = 0 @@ -628,15 +808,16 @@ class NsLcm(LcmBase): else: raise LcmException("ns_update_vnfr: Not found member_vnf_index={} from VIM info".format(vnf_index)) - @staticmethod - def _get_ns_config_info(vca_deployed_list): + def _get_ns_config_info(self, nsr_id): """ Generates a mapping between vnf,vdu elements and the N2VC id - :param vca_deployed_list: List of database _admin.deploy.VCA that contains this list + :param nsr_id: id of nsr to get last database _admin.deployed.VCA that contains this list :return: a dictionary with {osm-config-mapping: {}} where its element contains: "": for a vnf configuration, or "..": for a vdu configuration """ + db_nsr = self.db.get_one("nsrs", {"_id": nsr_id}) + vca_deployed_list = db_nsr["_admin"]["deployed"]["VCA"] mapping = {} ns_config_info = {"osm-config-mapping": mapping} for vca in vca_deployed_list: @@ -650,19 +831,27 @@ class NsLcm(LcmBase): return ns_config_info @staticmethod - def _get_initial_config_primitive_list(desc_primitive_list, vca_deployed): + def _get_initial_config_primitive_list(desc_primitive_list, vca_deployed, ee_descriptor_id): """ Generates a list of initial-config-primitive based on the list provided by the descriptor. It includes internal primitives as verify-ssh-credentials, or config when needed :param desc_primitive_list: information of the descriptor :param vca_deployed: information of the deployed, needed for known if it is related to an NS, VNF, VDU and if this element contains a ssh public key + :param ee_descriptor_id: execution environment descriptor id. It is the value of + XXX_configuration.execution-environment-list.INDEX.id; it can be None :return: The modified list. Can ba an empty list, but always a list """ - if desc_primitive_list: - primitive_list = desc_primitive_list.copy() - else: - primitive_list = [] + + primitive_list = desc_primitive_list or [] + + # filter primitives by ee_id + primitive_list = [p for p in primitive_list if p.get("execution-environment-ref") == ee_descriptor_id] + + # sort by 'seq' + if primitive_list: + primitive_list.sort(key=lambda val: int(val['seq'])) + # look for primitive config, and get the position. None if not present config_position = None for index, primitive in enumerate(primitive_list): @@ -674,194 +863,452 @@ class NsLcm(LcmBase): if not vca_deployed["member-vnf-index"] and config_position is None: primitive_list.insert(0, {"name": "config", "parameter": []}) config_position = 0 - # for VNF/VDU add verify-ssh-credentials after config + # TODO revise if needed: for VNF/VDU add verify-ssh-credentials after config if vca_deployed["member-vnf-index"] and config_position is not None and vca_deployed.get("ssh-public-key"): primitive_list.insert(config_position + 1, {"name": "verify-ssh-credentials", "parameter": []}) return primitive_list - async def instantiate_RO(self, logging_text, nsr_id, nsd, db_nsr, - db_nslcmop, db_vnfrs, db_vnfds_ref, n2vc_key_list): + async def _instantiate_ng_ro(self, logging_text, nsr_id, nsd, db_nsr, db_nslcmop, db_vnfrs, db_vnfds_ref, + n2vc_key_list, stage, start_deploy, timeout_ns_deploy): + nslcmop_id = db_nslcmop["_id"] + target = { + "name": db_nsr["name"], + "ns": {"vld": []}, + "vnf": [], + "image": deepcopy(db_nsr["image"]), + "flavor": deepcopy(db_nsr["flavor"]), + "action_id": nslcmop_id, + } + for image in target["image"]: + image["vim_info"] = [] + for flavor in target["flavor"]: + flavor["vim_info"] = [] + + ns_params = db_nslcmop.get("operationParams") + ssh_keys = [] + if ns_params.get("ssh_keys"): + ssh_keys += ns_params.get("ssh_keys") + if n2vc_key_list: + ssh_keys += n2vc_key_list + + cp2target = {} + for vld_index, vld in enumerate(nsd.get("vld")): + target_vld = {"id": vld["id"], + "name": vld["name"], + "mgmt-network": vld.get("mgmt-network", False), + "type": vld.get("type"), + "vim_info": [{"vim-network-name": vld.get("vim-network-name"), + "vim_account_id": ns_params["vimAccountId"]}], + } + for cp in vld["vnfd-connection-point-ref"]: + cp2target["member_vnf:{}.{}".format(cp["member-vnf-index-ref"], cp["vnfd-connection-point-ref"])] = \ + "nsrs:{}:vld.{}".format(nsr_id, vld_index) + target["ns"]["vld"].append(target_vld) + for vnfr in db_vnfrs.values(): + vnfd = db_vnfds_ref[vnfr["vnfd-ref"]] + target_vnf = deepcopy(vnfr) + for vld in target_vnf.get("vld", ()): + # check if connected to a ns.vld + vnf_cp = next((cp for cp in vnfd.get("connection-point", ()) if + cp.get("internal-vld-ref") == vld["id"]), None) + if vnf_cp: + ns_cp = "member_vnf:{}.{}".format(vnfr["member-vnf-index-ref"], vnf_cp["id"]) + if cp2target.get(ns_cp): + vld["target"] = cp2target[ns_cp] + vld["vim_info"] = [{"vim-network-name": vld.get("vim-network-name"), + "vim_account_id": vnfr["vim-account-id"]}] + + for vdur in target_vnf.get("vdur", ()): + vdur["vim_info"] = [{"vim_account_id": vnfr["vim-account-id"]}] + vdud_index, vdud = next(k for k in enumerate(vnfd["vdu"]) if k[1]["id"] == vdur["vdu-id-ref"]) + # vdur["additionalParams"] = vnfr.get("additionalParamsForVnf") # TODO additional params for VDU + + if ssh_keys: + if deep_get(vdud, ("vdu-configuration", "config-access", "ssh-access", "required")): + vdur["ssh-keys"] = ssh_keys + vdur["ssh-access-required"] = True + elif deep_get(vnfd, ("vnf-configuration", "config-access", "ssh-access", "required")) and \ + any(iface.get("mgmt-vnf") for iface in vdur["interfaces"]): + vdur["ssh-keys"] = ssh_keys + vdur["ssh-access-required"] = True + + # cloud-init + if vdud.get("cloud-init-file"): + vdur["cloud-init"] = "{}:file:{}".format(vnfd["_id"], vdud.get("cloud-init-file")) + elif vdud.get("cloud-init"): + vdur["cloud-init"] = "{}:vdu:{}".format(vnfd["_id"], vdud_index) + + # flavor + ns_flavor = target["flavor"][int(vdur["ns-flavor-id"])] + if not next((vi for vi in ns_flavor["vim_info"] if + vi and vi.get("vim_account_id") == vnfr["vim-account-id"]), None): + ns_flavor["vim_info"].append({"vim_account_id": vnfr["vim-account-id"]}) + # image + ns_image = target["image"][int(vdur["ns-image-id"])] + if not next((vi for vi in ns_image["vim_info"] if + vi and vi.get("vim_account_id") == vnfr["vim-account-id"]), None): + ns_image["vim_info"].append({"vim_account_id": vnfr["vim-account-id"]}) + + vdur["vim_info"] = [{"vim_account_id": vnfr["vim-account-id"]}] + target["vnf"].append(target_vnf) + + desc = await self.RO.deploy(nsr_id, target) + action_id = desc["action_id"] + await self._wait_ng_ro(self, nsr_id, action_id, nslcmop_id, start_deploy, timeout_ns_deploy, stage) + + # Updating NSR + db_nsr_update = { + "_admin.deployed.RO.operational-status": "running", + "detailed-status": " ".join(stage) + } + # db_nsr["_admin.deployed.RO.detailed-status"] = "Deployed at VIM" + self.update_db_2("nsrs", nsr_id, db_nsr_update) + self._write_op_status(nslcmop_id, stage) + self.logger.debug(logging_text + "ns deployed at RO. RO_id={}".format(action_id)) + return + + async def _wait_ng_ro(self, nsr_id, action_id, nslcmop_id, start_time, timeout, stage): + detailed_status_old = None + db_nsr_update = {} + while time() <= start_time + timeout: + desc_status = await self.RO.status(nsr_id, action_id) + if desc_status["status"] == "FAILED": + raise NgRoException(desc_status["details"]) + elif desc_status["status"] == "BUILD": + stage[2] = "VIM: ({})".format(desc_status["details"]) + elif desc_status["status"] == "DONE": + stage[2] = "Deployed at VIM" + break + else: + assert False, "ROclient.check_ns_status returns unknown {}".format(desc_status["status"]) + if stage[2] != detailed_status_old: + detailed_status_old = stage[2] + db_nsr_update["detailed-status"] = " ".join(stage) + self.update_db_2("nsrs", nsr_id, db_nsr_update) + self._write_op_status(nslcmop_id, stage) + await asyncio.sleep(5, loop=self.loop) + else: # timeout_ns_deploy + raise NgRoException("Timeout waiting ns to deploy") + async def _terminate_ng_ro(self, logging_text, nsr_deployed, nsr_id, nslcmop_id, stage): db_nsr_update = {} - RO_descriptor_number = 0 # number of descriptors created at RO - vnf_index_2_RO_id = {} # map between vnfd/nsd id to the id used at RO + failed_detail = [] + action_id = None start_deploy = time() - vdu_flag = False # If any of the VNFDs has VDUs - ns_params = db_nslcmop.get("operationParams") + try: + target = { + "ns": {"vld": []}, + "vnf": [], + "image": [], + "flavor": [], + } + desc = await self.RO.deploy(nsr_id, target) + action_id = desc["action_id"] + db_nsr_update["_admin.deployed.RO.nsr_delete_action_id"] = action_id + db_nsr_update["_admin.deployed.RO.nsr_status"] = "DELETING" + self.logger.debug(logging_text + "ns terminate action at RO. action_id={}".format(action_id)) + + # wait until done + delete_timeout = 20 * 60 # 20 minutes + await self._wait_ng_ro(self, nsr_id, action_id, nslcmop_id, start_deploy, delete_timeout, stage) + + db_nsr_update["_admin.deployed.RO.nsr_delete_action_id"] = None + db_nsr_update["_admin.deployed.RO.nsr_status"] = "DELETED" + # delete all nsr + await self.RO.delete(nsr_id) + except Exception as e: + if isinstance(e, NgRoException) and e.http_code == 404: # not found + db_nsr_update["_admin.deployed.RO.nsr_id"] = None + db_nsr_update["_admin.deployed.RO.nsr_status"] = "DELETED" + db_nsr_update["_admin.deployed.RO.nsr_delete_action_id"] = None + self.logger.debug(logging_text + "RO_action_id={} already deleted".format(action_id)) + elif isinstance(e, NgRoException) and e.http_code == 409: # conflict + failed_detail.append("delete conflict: {}".format(e)) + self.logger.debug(logging_text + "RO_action_id={} delete conflict: {}".format(action_id, e)) + else: + failed_detail.append("delete error: {}".format(e)) + self.logger.error(logging_text + "RO_action_id={} delete error: {}".format(action_id, e)) - # deploy RO - - # get vnfds, instantiate at RO - - for c_vnf in nsd.get("constituent-vnfd", ()): - member_vnf_index = c_vnf["member-vnf-index"] - vnfd = db_vnfds_ref[c_vnf['vnfd-id-ref']] - if vnfd.get("vdu"): - vdu_flag = True - vnfd_ref = vnfd["id"] - step = db_nsr_update["_admin.deployed.RO.detailed-status"] = "Creating vnfd='{}' member_vnf_index='{}' at" \ - " RO".format(vnfd_ref, member_vnf_index) - # self.logger.debug(logging_text + step) - vnfd_id_RO = "{}.{}.{}".format(nsr_id, RO_descriptor_number, member_vnf_index[:23]) - vnf_index_2_RO_id[member_vnf_index] = vnfd_id_RO - RO_descriptor_number += 1 + if failed_detail: + stage[2] = "Error deleting from VIM" + else: + stage[2] = "Deleted from VIM" + db_nsr_update["detailed-status"] = " ".join(stage) + self.update_db_2("nsrs", nsr_id, db_nsr_update) + self._write_op_status(nslcmop_id, stage) - # look position at deployed.RO.vnfd if not present it will be appended at the end - for index, vnf_deployed in enumerate(db_nsr["_admin"]["deployed"]["RO"]["vnfd"]): - if vnf_deployed["member-vnf-index"] == member_vnf_index: - break + if failed_detail: + raise LcmException("; ".join(failed_detail)) + return + + 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 + vnf_index_2_RO_id = {} # map between vnfd/nsd id to the id used at RO + nslcmop_id = db_nslcmop["_id"] + start_deploy = time() + ns_params = db_nslcmop.get("operationParams") + if ns_params and ns_params.get("timeout_ns_deploy"): + timeout_ns_deploy = ns_params["timeout_ns_deploy"] else: - index = len(db_nsr["_admin"]["deployed"]["RO"]["vnfd"]) - db_nsr["_admin"]["deployed"]["RO"]["vnfd"].append(None) - - # look if present - RO_update = {"member-vnf-index": member_vnf_index} - vnfd_list = await self.RO.get_list("vnfd", filter_by={"osm_id": vnfd_id_RO}) - if vnfd_list: - RO_update["id"] = vnfd_list[0]["uuid"] - self.logger.debug(logging_text + "vnfd='{}' member_vnf_index='{}' exists at RO. Using RO_id={}". - format(vnfd_ref, member_vnf_index, vnfd_list[0]["uuid"])) + timeout_ns_deploy = self.timeout.get("ns_deploy", self.timeout_ns_deploy) + + # Check for and optionally request placement optimization. Database will be updated if placement activated + stage[2] = "Waiting for Placement." + 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"] + + if self.ng_ro: + return await self._instantiate_ng_ro(logging_text, nsr_id, nsd, db_nsr, db_nslcmop, db_vnfrs, + db_vnfds_ref, n2vc_key_list, stage, start_deploy, + timeout_ns_deploy) + # deploy RO + # get vnfds, instantiate at RO + for c_vnf in nsd.get("constituent-vnfd", ()): + member_vnf_index = c_vnf["member-vnf-index"] + vnfd = db_vnfds_ref[c_vnf['vnfd-id-ref']] + vnfd_ref = vnfd["id"] + + stage[2] = "Creating vnfd='{}' member_vnf_index='{}' at RO".format(vnfd_ref, member_vnf_index) + db_nsr_update["detailed-status"] = " ".join(stage) + self.update_db_2("nsrs", nsr_id, db_nsr_update) + self._write_op_status(nslcmop_id, stage) + + # self.logger.debug(logging_text + stage[2]) + vnfd_id_RO = "{}.{}.{}".format(nsr_id, RO_descriptor_number, member_vnf_index[:23]) + vnf_index_2_RO_id[member_vnf_index] = vnfd_id_RO + RO_descriptor_number += 1 + + # look position at deployed.RO.vnfd if not present it will be appended at the end + for index, vnf_deployed in enumerate(db_nsr["_admin"]["deployed"]["RO"]["vnfd"]): + if vnf_deployed["member-vnf-index"] == member_vnf_index: + break + else: + index = len(db_nsr["_admin"]["deployed"]["RO"]["vnfd"]) + db_nsr["_admin"]["deployed"]["RO"]["vnfd"].append(None) + + # look if present + RO_update = {"member-vnf-index": member_vnf_index} + vnfd_list = await self.RO.get_list("vnfd", filter_by={"osm_id": vnfd_id_RO}) + if vnfd_list: + RO_update["id"] = vnfd_list[0]["uuid"] + self.logger.debug(logging_text + "vnfd='{}' member_vnf_index='{}' exists at RO. Using RO_id={}". + format(vnfd_ref, member_vnf_index, vnfd_list[0]["uuid"])) + else: + vnfd_RO = self.vnfd2RO(vnfd, vnfd_id_RO, db_vnfrs[c_vnf["member-vnf-index"]]. + get("additionalParamsForVnf"), nsr_id) + desc = await self.RO.create("vnfd", descriptor=vnfd_RO) + RO_update["id"] = desc["uuid"] + self.logger.debug(logging_text + "vnfd='{}' member_vnf_index='{}' created at RO. RO_id={}".format( + vnfd_ref, member_vnf_index, desc["uuid"])) + db_nsr_update["_admin.deployed.RO.vnfd.{}".format(index)] = RO_update + db_nsr["_admin"]["deployed"]["RO"]["vnfd"][index] = RO_update + + # create nsd at RO + nsd_ref = nsd["id"] + + stage[2] = "Creating nsd={} at RO".format(nsd_ref) + db_nsr_update["detailed-status"] = " ".join(stage) + self.update_db_2("nsrs", nsr_id, db_nsr_update) + self._write_op_status(nslcmop_id, stage) + + # self.logger.debug(logging_text + stage[2]) + RO_osm_nsd_id = "{}.{}.{}".format(nsr_id, RO_descriptor_number, nsd_ref[:23]) + RO_descriptor_number += 1 + nsd_list = await self.RO.get_list("nsd", filter_by={"osm_id": RO_osm_nsd_id}) + if nsd_list: + db_nsr_update["_admin.deployed.RO.nsd_id"] = RO_nsd_uuid = nsd_list[0]["uuid"] + self.logger.debug(logging_text + "nsd={} exists at RO. Using RO_id={}".format( + nsd_ref, RO_nsd_uuid)) else: - vnfd_RO = self.vnfd2RO(vnfd, vnfd_id_RO, db_vnfrs[c_vnf["member-vnf-index"]]. - get("additionalParamsForVnf"), nsr_id) - desc = await self.RO.create("vnfd", descriptor=vnfd_RO) - RO_update["id"] = desc["uuid"] - self.logger.debug(logging_text + "vnfd='{}' member_vnf_index='{}' created at RO. RO_id={}".format( - vnfd_ref, member_vnf_index, desc["uuid"])) - db_nsr_update["_admin.deployed.RO.vnfd.{}".format(index)] = RO_update - db_nsr["_admin"]["deployed"]["RO"]["vnfd"][index] = RO_update + nsd_RO = deepcopy(nsd) + nsd_RO["id"] = RO_osm_nsd_id + nsd_RO.pop("_id", None) + nsd_RO.pop("_admin", None) + for c_vnf in nsd_RO.get("constituent-vnfd", ()): + member_vnf_index = c_vnf["member-vnf-index"] + c_vnf["vnfd-id-ref"] = vnf_index_2_RO_id[member_vnf_index] + for c_vld in nsd_RO.get("vld", ()): + for cp in c_vld.get("vnfd-connection-point-ref", ()): + member_vnf_index = cp["member-vnf-index-ref"] + cp["vnfd-id-ref"] = vnf_index_2_RO_id[member_vnf_index] + + desc = await self.RO.create("nsd", descriptor=nsd_RO) + db_nsr_update["_admin.nsState"] = "INSTANTIATED" + 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) - - # create nsd at RO - nsd_ref = nsd["id"] - step = db_nsr_update["_admin.deployed.RO.detailed-status"] = "Creating nsd={} at RO".format(nsd_ref) - # self.logger.debug(logging_text + step) - - RO_osm_nsd_id = "{}.{}.{}".format(nsr_id, RO_descriptor_number, nsd_ref[:23]) - RO_descriptor_number += 1 - nsd_list = await self.RO.get_list("nsd", filter_by={"osm_id": RO_osm_nsd_id}) - if nsd_list: - db_nsr_update["_admin.deployed.RO.nsd_id"] = RO_nsd_uuid = nsd_list[0]["uuid"] - self.logger.debug(logging_text + "nsd={} exists at RO. Using RO_id={}".format( - nsd_ref, RO_nsd_uuid)) - else: - nsd_RO = deepcopy(nsd) - nsd_RO["id"] = RO_osm_nsd_id - nsd_RO.pop("_id", None) - nsd_RO.pop("_admin", None) - for c_vnf in nsd_RO.get("constituent-vnfd", ()): - member_vnf_index = c_vnf["member-vnf-index"] - c_vnf["vnfd-id-ref"] = vnf_index_2_RO_id[member_vnf_index] - for c_vld in nsd_RO.get("vld", ()): - for cp in c_vld.get("vnfd-connection-point-ref", ()): - member_vnf_index = cp["member-vnf-index-ref"] - cp["vnfd-id-ref"] = vnf_index_2_RO_id[member_vnf_index] - desc = await self.RO.create("nsd", descriptor=nsd_RO) - db_nsr_update["_admin.nsState"] = "INSTANTIATED" - 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 + stage[2] = "Creating nsd={} at RO".format(nsd_ref) + db_nsr_update["detailed-status"] = " ".join(stage) + self.update_db_2("nsrs", nsr_id, db_nsr_update) + self._write_op_status(nslcmop_id, stage) - # Crate ns at RO - # if present use it unless in error status - RO_nsr_id = deep_get(db_nsr, ("_admin", "deployed", "RO", "nsr_id")) - if RO_nsr_id: - try: - 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 - RO_nsr_id = db_nsr_update["_admin.deployed.RO.nsr_id"] = None + # if present use it unless in error status + RO_nsr_id = deep_get(db_nsr, ("_admin", "deployed", "RO", "nsr_id")) if 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": - step = db_nsr_update["_admin.deployed.RO.detailed-status"] = "Deleting ns at RO. RO_ns_id={}"\ - .format(RO_nsr_id) - self.logger.debug(logging_text + step) - await self.RO.delete("ns", RO_nsr_id) + try: + stage[2] = "Looking for existing ns at RO" + db_nsr_update["detailed-status"] = " ".join(stage) + self.update_db_2("nsrs", nsr_id, db_nsr_update) + self._write_op_status(nslcmop_id, stage) + # self.logger.debug(logging_text + stage[2] + " 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 RO_nsr_id = db_nsr_update["_admin.deployed.RO.nsr_id"] = None - if not RO_nsr_id: - step = db_nsr_update["_admin.deployed.RO.detailed-status"] = "Checking dependencies" - # self.logger.debug(logging_text + step) + if 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": + stage[2] = "Deleting ns at RO. RO_ns_id={}".format(RO_nsr_id) + self.logger.debug(logging_text + stage[2]) + await self.RO.delete("ns", RO_nsr_id) + RO_nsr_id = db_nsr_update["_admin.deployed.RO.nsr_id"] = None + if not RO_nsr_id: + stage[2] = "Checking dependencies" + db_nsr_update["detailed-status"] = " ".join(stage) + self.update_db_2("nsrs", nsr_id, db_nsr_update) + self._write_op_status(nslcmop_id, stage) + # self.logger.debug(logging_text + stage[2]) + + # check if VIM is creating and wait look if previous tasks in process + task_name, task_dependency = self.lcm_tasks.lookfor_related("vim_account", ns_params["vimAccountId"]) + if task_dependency: + stage[2] = "Waiting for related tasks '{}' to be completed".format(task_name) + self.logger.debug(logging_text + stage[2]) + await asyncio.wait(task_dependency, timeout=3600) + if ns_params.get("vnf"): + for vnf in ns_params["vnf"]: + if "vimAccountId" in vnf: + task_name, task_dependency = self.lcm_tasks.lookfor_related("vim_account", + vnf["vimAccountId"]) + if task_dependency: + stage[2] = "Waiting for related tasks '{}' to be completed.".format(task_name) + self.logger.debug(logging_text + stage[2]) + 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, 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) + self._write_op_status(nslcmop_id, stage) - # check if VIM is creating and wait look if previous tasks in process - task_name, task_dependency = self.lcm_tasks.lookfor_related("vim_account", ns_params["vimAccountId"]) - if task_dependency: - step = "Waiting for related tasks to be completed: {}".format(task_name) - self.logger.debug(logging_text + step) - await asyncio.wait(task_dependency, timeout=3600) - if ns_params.get("vnf"): - for vnf in ns_params["vnf"]: - if "vimAccountId" in vnf: - task_name, task_dependency = self.lcm_tasks.lookfor_related("vim_account", - vnf["vimAccountId"]) - if task_dependency: - step = "Waiting for related tasks to be completed: {}".format(task_name) - self.logger.debug(logging_text + step) - await asyncio.wait(task_dependency, timeout=3600) - - step = db_nsr_update["_admin.deployed.RO.detailed-status"] = "Checking instantiation parameters" - - RO_ns_params = self.ns_params_2_RO(ns_params, nsd, db_vnfds_ref, n2vc_key_list) - - step = db_nsr_update["detailed-status"] = "Deploying ns at VIM" - # step = db_nsr_update["_admin.deployed.RO.detailed-status"] = "Deploying ns at VIM" - desc = await self.RO.create("ns", descriptor=RO_ns_params, name=db_nsr["name"], scenario=RO_nsd_uuid) - RO_nsr_id = db_nsr_update["_admin.deployed.RO.nsr_id"] = desc["uuid"] - db_nsr_update["_admin.nsState"] = "INSTANTIATED" - 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) + desc = await self.RO.create("ns", descriptor=RO_ns_params, name=db_nsr["name"], scenario=RO_nsd_uuid) + RO_nsr_id = db_nsr_update["_admin.deployed.RO.nsr_id"] = desc["uuid"] + db_nsr_update["_admin.nsState"] = "INSTANTIATED" + db_nsr_update["_admin.deployed.RO.nsr_status"] = "BUILD" + self.logger.debug(logging_text + "ns created at RO. RO_id={}".format(desc["uuid"])) - # 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) - - 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) - else: # total_deploy_timeout - raise ROclient.ROClientException("Timeout waiting ns to be ready") + # wait until NS is ready + stage[2] = "Waiting VIM to deploy ns." + db_nsr_update["detailed-status"] = " ".join(stage) + self.update_db_2("nsrs", nsr_id, db_nsr_update) + self._write_op_status(nslcmop_id, stage) + detailed_status_old = None + self.logger.debug(logging_text + stage[2] + " RO_ns_id={}".format(RO_nsr_id)) - step = "Updating NSR" - self.ns_update_nsr(db_nsr_update, db_nsr, desc) + old_desc = None + while time() <= start_deploy + timeout_ns_deploy: + desc = await self.RO.show("ns", RO_nsr_id) - db_nsr_update["_admin.deployed.RO.operational-status"] = "running" - 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) + # 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": + stage[2] = "VIM: ({})".format(ns_status_info) + elif ns_status == "ACTIVE": + stage[2] = "Waiting for management IP address reported by the VIM. Updating VNFRs." + try: + self.ns_update_vnfr(db_vnfrs, desc) + break + except LcmExceptionNoMgmtIP: + pass + else: + assert False, "ROclient.check_ns_status returns unknown {}".format(ns_status) + if stage[2] != detailed_status_old: + detailed_status_old = stage[2] + db_nsr_update["detailed-status"] = " ".join(stage) + self.update_db_2("nsrs", nsr_id, db_nsr_update) + self._write_op_status(nslcmop_id, stage) + await asyncio.sleep(5, loop=self.loop) + else: # timeout_ns_deploy + raise ROclient.ROClientException("Timeout waiting ns to be ready") + + # Updating NSR + self.ns_update_nsr(db_nsr_update, db_nsr, desc) + + db_nsr_update["_admin.deployed.RO.operational-status"] = "running" + # db_nsr["_admin.deployed.RO.detailed-status"] = "Deployed at VIM" + stage[2] = "Deployed at VIM" + db_nsr_update["detailed-status"] = " ".join(stage) + self.update_db_2("nsrs", nsr_id, db_nsr_update) + self._write_op_status(nslcmop_id, stage) + # await self._on_update_n2vc_db("nsrs", {"_id": nsr_id}, "_admin.deployed", db_nsr_update) + # self.logger.debug(logging_text + "Deployed at VIM") + except (ROclient.ROClientException, LcmException, DbException, NgRoException) as e: + stage[2] = "ERROR deploying at VIM" + self.set_vnfr_at_error(db_vnfrs, str(e)) + raise + + async def wait_kdu_up(self, logging_text, nsr_id, vnfr_id, kdu_name): + """ + Wait for kdu to be up, get ip address + :param logging_text: prefix use for logging + :param nsr_id: + :param vnfr_id: + :param kdu_name: + :return: IP address + """ + + # self.logger.debug(logging_text + "Starting wait_kdu_up") + nb_tries = 0 + + while nb_tries < 360: + db_vnfr = self.db.get_one("vnfrs", {"_id": vnfr_id}) + kdur = next((x for x in get_iterable(db_vnfr, "kdur") if x.get("kdu-name") == kdu_name), None) + if not kdur: + raise LcmException("Not found vnfr_id={}, kdu_name={}".format(vnfr_id, kdu_name)) + if kdur.get("status"): + if kdur["status"] in ("READY", "ENABLED"): + return kdur.get("ip-address") + else: + raise LcmException("target KDU={} is in error state".format(kdu_name)) - step = "Deployed at VIM" - self.logger.debug(logging_text + step) + await asyncio.sleep(10, loop=self.loop) + nb_tries += 1 + raise LcmException("Timeout waiting KDU={} instantiated".format(kdu_name)) async def wait_vm_up_insert_key_ro(self, logging_text, nsr_id, vnfr_id, vdu_id, vdu_index, pub_key=None, user=None): """ @@ -890,18 +1337,14 @@ class NsLcm(LcmBase): raise LcmException("Not found _admin.deployed.RO.nsr_id for nsr_id: {}".format(nsr_id)) await asyncio.sleep(10, loop=self.loop) - # wait until NS is deployed at RO - if not ro_nsr_id: - db_nsrs = self.db.get_one("nsrs", {"_id": nsr_id}) - ro_nsr_id = deep_get(db_nsrs, ("_admin", "deployed", "RO", "nsr_id")) - if not ro_nsr_id: - continue # get ip address if not target_vdu_id: db_vnfr = self.db.get_one("vnfrs", {"_id": vnfr_id}) if not vdu_id: # for the VNF case + if db_vnfr.get("status") == "ERROR": + raise LcmException("Cannot inject ssh-key because target VNF is in error state") ip_address = db_vnfr.get("ip-address") if not ip_address: continue @@ -910,12 +1353,13 @@ class NsLcm(LcmBase): vdur = next((x for x in get_iterable(db_vnfr, "vdur") if x.get("vdu-id-ref") == vdu_id and x.get("count-index") == vdu_index), None) + if not vdur and len(db_vnfr.get("vdur", ())) == 1: # If only one, this should be the target vdu + vdur = db_vnfr["vdur"][0] if not vdur: - raise LcmException("Not found vnfr_id={}, vdu_index={}, vdu_index={}".format( - vnfr_id, vdu_id, vdu_index - )) + raise LcmException("Not found vnfr_id={}, vdu_id={}, vdu_index={}".format(vnfr_id, vdu_id, + vdu_index)) - if vdur.get("status") == "ACTIVE": + if vdur.get("pdu-type") or vdur.get("status") == "ACTIVE": ip_address = vdur.get("ip-address") if not ip_address: continue @@ -926,28 +1370,44 @@ class NsLcm(LcmBase): if not target_vdu_id: continue - # self.logger.debug(logging_text + "IP address={}".format(ip_address)) - # inject public key into machine if pub_key and user: + # wait until NS is deployed at RO + if not ro_nsr_id: + db_nsrs = self.db.get_one("nsrs", {"_id": nsr_id}) + ro_nsr_id = deep_get(db_nsrs, ("_admin", "deployed", "RO", "nsr_id")) + if not ro_nsr_id: + continue + # self.logger.debug(logging_text + "Inserting RO key") + if vdur.get("pdu-type"): + self.logger.error(logging_text + "Cannot inject ssh-ky to a PDU") + return ip_address try: ro_vm_id = "{}-{}".format(db_vnfr["member-vnf-index-ref"], target_vdu_id) # TODO add vdu_index - result_dict = await self.RO.create_action( - item="ns", - item_id_name=ro_nsr_id, - descriptor={"add_public_key": pub_key, "vms": [ro_vm_id], "user": user} - ) - # result_dict contains the format {VM-id: {vim_result: 200, description: text}} - if not result_dict or not isinstance(result_dict, dict): - raise LcmException("Unknown response from RO when injecting key") - for result in result_dict.values(): - if result.get("vim_result") == 200: - break - else: - raise ROclient.ROClientException("error injecting key: {}".format( - result.get("description"))) - break + if self.ng_ro: + target = {"action": "inject_ssh_key", "key": pub_key, "user": user, + "vnf": [{"_id": vnfr_id, "vdur": [{"id": vdu_id}]}], + } + await self.RO.deploy(nsr_id, target) + else: + result_dict = await self.RO.create_action( + item="ns", + item_id_name=ro_nsr_id, + descriptor={"add_public_key": pub_key, "vms": [ro_vm_id], "user": user} + ) + # result_dict contains the format {VM-id: {vim_result: 200, description: text}} + if not result_dict or not isinstance(result_dict, dict): + raise LcmException("Unknown response from RO when injecting key") + for result in result_dict.values(): + if result.get("vim_result") == 200: + break + else: + raise ROclient.ROClientException("error injecting key: {}".format( + result.get("description"))) + break + except NgRoException as e: + raise LcmException("Reaching max tries injecting key. Error: {}".format(e)) except ROclient.ROClientException as e: if not nb_tries: self.logger.debug(logging_text + "error injecting key: {}. Retrying until {} seconds". @@ -960,58 +1420,144 @@ class NsLcm(LcmBase): return ip_address - 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 _wait_dependent_n2vc(self, nsr_id, vca_deployed_list, vca_index): + """ + Wait until dependent VCA deployments have been finished. NS wait for VNFs and VDUs. VNFs for VDUs + """ + 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: + 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")): + 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 + + 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, nslcmop_id, stage, vca_type, vca_name, + ee_config_descriptor): nsr_id = db_nsr["_id"] db_update_entry = "_admin.deployed.VCA.{}.".format(vca_index) vca_deployed_list = db_nsr["_admin"]["deployed"]["VCA"] vca_deployed = db_nsr["_admin"]["deployed"]["VCA"][vca_index] + osm_config = {"osm": {"ns_id": db_nsr["_id"]}} db_dict = { 'collection': 'nsrs', 'filter': {'_id': nsr_id}, 'path': db_update_entry } - logging_text += "member_vnf_index={} vdu_id={}, vdu_index={} ".format(db_vnfr["member-vnf-index-ref"], - vdu_id, vdu_index) - step = "" try: + + element_type = 'NS' + element_under_configuration = nsr_id + vnfr_id = None if db_vnfr: vnfr_id = db_vnfr["_id"] + osm_config["osm"]["vnf_id"] = vnfr_id namespace = "{nsi}.{ns}".format( nsi=nsi_id if nsi_id else "", ns=nsr_id) + if vnfr_id: - namespace += "." + vnfr_id + element_type = 'VNF' + element_under_configuration = vnfr_id + namespace += ".{}".format(vnfr_id) if vdu_id: namespace += ".{}-{}".format(vdu_id, vdu_index or 0) + element_type = 'VDU' + element_under_configuration = "{}-{}".format(vdu_id, vdu_index or 0) + osm_config["osm"]["vdu_id"] = vdu_id + elif kdu_name: + namespace += ".{}".format(kdu_name) + element_type = 'KDU' + element_under_configuration = kdu_name + osm_config["osm"]["kdu_name"] = kdu_name # Get artifact path - artifact_path = "/{}/{}/charms/{}".format( + artifact_path = "{}/{}/{}/{}".format( base_folder["folder"], base_folder["pkg-dir"], - config_descriptor["juju"]["charm"] + "charms" if vca_type in ("native_charm", "lxc_proxy_charm", "k8s_proxy_charm") else "helm-charts", + vca_name ) + # get initial_config_primitive_list that applies to this element + initial_config_primitive_list = config_descriptor.get('initial-config-primitive') - is_proxy_charm = deep_get(config_descriptor, ('juju', 'charm')) is not None - if deep_get(config_descriptor, ('juju', 'proxy')) is False: - is_proxy_charm = False + # add config if not present for NS charm + ee_descriptor_id = ee_config_descriptor.get("id") + initial_config_primitive_list = self._get_initial_config_primitive_list(initial_config_primitive_list, + vca_deployed, ee_descriptor_id) # n2vc_redesign STEP 3.1 - # find old ee_id if exists ee_id = vca_deployed.get("ee_id") + vim_account_id = ( + deep_get(db_vnfr, ("vim-account-id",)) or + deep_get(deploy_params, ("OSM", "vim_account_id")) + ) + vca_cloud, vca_cloud_credential = self.get_vca_cloud_and_credentials(vim_account_id) + vca_k8s_cloud, vca_k8s_cloud_credential = self.get_vca_k8s_cloud_and_credentials(vim_account_id) # create or register execution environment in VCA - if is_proxy_charm: + if vca_type in ("lxc_proxy_charm", "k8s_proxy_charm", "helm", "helm-v3"): + + 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: + + ee_id = None + credentials = None + if vca_type == "k8s_proxy_charm": + ee_id = await self.vca_map[vca_type].install_k8s_proxy_charm( + charm_name=artifact_path[artifact_path.rfind("/") + 1:], + namespace=namespace, + artifact_path=artifact_path, + db_dict=db_dict, + cloud_name=vca_k8s_cloud, + credential_name=vca_k8s_cloud_credential, + ) + else: + ee_id, credentials = await self.vca_map[vca_type].create_execution_environment( + namespace=namespace, + reuse_ee_id=ee_id, + db_dict=db_dict, + config=osm_config, + cloud_name=vca_cloud, + credential_name=vca_cloud_credential, + ) + + elif vca_type == "native_charm": step = "Waiting to VM being up and getting IP address" self.logger.debug(logging_text + step) rw_mgmt_ip = await self.wait_vm_up_insert_key_ro(logging_text, nsr_id, vnfr_id, vdu_id, vdu_index, @@ -1021,58 +1567,125 @@ class NsLcm(LcmBase): username = deep_get(config_descriptor, ("config-access", "ssh-access", "default-user")) # TODO remove this when changes on IM regarding config-access:ssh-access:default-user were # merged. Meanwhile let's get username from initial-config-primitive - if not username and config_descriptor.get("initial-config-primitive"): - for config_primitive in config_descriptor["initial-config-primitive"]: + if not username and initial_config_primitive_list: + for config_primitive in initial_config_primitive_list: for param in config_primitive.get("parameter", ()): if param["name"] == "ssh-username": username = param["value"] break if not username: - raise LcmException("Cannot determine the username neither with 'initial-config-promitive' nor with " + raise LcmException("Cannot determine the username neither with 'initial-config-primitive' nor with " "'config-access.ssh-access.default-user'") credentials["username"] = username # n2vc_redesign STEP 3.2 + 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, - db_dict=db_dict) + ee_id = await self.vca_map[vca_type].register_execution_environment( + credentials=credentials, + namespace=namespace, + db_dict=db_dict, + cloud_name=vca_cloud, + credential_name=vca_cloud_credential, + ) # for compatibility with MON/POL modules, the need model and application name at database - # TODO ask to N2VC instead of assuming the format "model_name.application_name" + # TODO ask MON/POL if needed to not assuming anymore the format "model_name.application_name" 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 + "ee_id": ee_id} + if len(ee_id_parts) >= 2: + model_name = ee_id_parts[0] + application_name = ee_id_parts[1] + db_nsr_update[db_update_entry + "model"] = model_name + db_nsr_update[db_update_entry + "application"] = application_name # n2vc_redesign STEP 3.3 - step = "Install configuration Software" + + 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, + other_update=db_nsr_update + ) + # 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) + config = None + if vca_type == "native_charm": + config_primitive = next((p for p in initial_config_primitive_list if p["name"] == "config"), None) + if config_primitive: + config = self._map_primitive_params( + config_primitive, + {}, + deploy_params + ) + num_units = 1 + if vca_type == "lxc_proxy_charm": + if element_type == "NS": + num_units = db_nsr.get("config-units") or 1 + elif element_type == "VNF": + num_units = db_vnfr.get("config-units") or 1 + elif element_type == "VDU": + for v in db_vnfr["vdur"]: + if vdu_id == v["vdu-id-ref"]: + num_units = v.get("config-units") or 1 + break + if vca_type != "k8s_proxy_charm": + await self.vca_map[vca_type].install_configuration_sw( + ee_id=ee_id, + artifact_path=artifact_path, + db_dict=db_dict, + config=config, + num_units=num_units, + ) + + # write in db flag of configuration_sw already installed + self.update_db_2("nsrs", nsr_id, {db_update_entry + "config_sw_installed": True}) + + # add relations for this VCA (wait for other peers related with this VCA) + await self._add_vca_relations(logging_text=logging_text, nsr_id=nsr_id, + vca_index=vca_index, vca_type=vca_type) # if SSH access is required, then get execution environment SSH public - if is_proxy_charm: # if native charm we have waited already to VM be UP + # if native charm we have waited already to VM be UP + if vca_type in ("k8s_proxy_charm", "lxc_proxy_charm", "helm", "helm-v3"): pub_key = None user = None + # self.logger.debug("get ssh key block") if deep_get(config_descriptor, ("config-access", "ssh-access", "required")): + # self.logger.debug("ssh key needed") # Needed to inject a ssh key user = deep_get(config_descriptor, ("config-access", "ssh-access", "default-user")) step = "Install configuration Software, getting public ssh key" - pub_key = await self.n2vc.get_ee_ssh_public__key(ee_id=ee_id, db_dict=db_dict) + pub_key = await self.vca_map[vca_type].get_ee_ssh_public__key(ee_id=ee_id, db_dict=db_dict) - step = "Insert public key into VM" + step = "Insert public key into VM user={} ssh_key={}".format(user, pub_key) else: + # self.logger.debug("no need to get ssh key") step = "Waiting to VM being up and getting IP address" self.logger.debug(logging_text + step) # n2vc_redesign STEP 5.1 # wait for RO (ip-address) Insert pub_key into VM - rw_mgmt_ip = await self.wait_vm_up_insert_key_ro(logging_text, nsr_id, vnfr_id, vdu_id, vdu_index, - user=user, pub_key=pub_key) + if vnfr_id: + if kdu_name: + rw_mgmt_ip = await self.wait_kdu_up(logging_text, nsr_id, vnfr_id, kdu_name) + else: + rw_mgmt_ip = await self.wait_vm_up_insert_key_ro(logging_text, nsr_id, vnfr_id, vdu_id, + vdu_index, user=user, pub_key=pub_key) + else: + rw_mgmt_ip = None # This is for a NS configuration self.logger.debug(logging_text + ' VM_ip_address={}'.format(rw_mgmt_ip)) @@ -1081,46 +1694,225 @@ class NsLcm(LcmBase): # n2vc_redesign STEP 6 Execute initial config primitive step = 'execute initial config primitive' - initial_config_primitive_list = config_descriptor.get('initial-config-primitive') - # sort initial config primitives by 'seq' - try: - initial_config_primitive_list.sort(key=lambda val: int(val['seq'])) - except Exception as e: - self.logger.error(logging_text + step + ": " + str(e)) + # 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[0] = 'Stage 3/5: running Day-1 primitives for VDU.' + elif my_vca.get("member-vnf-index"): + # VNF + stage[0] = 'Stage 4/5: running Day-1 primitives for VNF.' + else: + # NS + stage[0] = 'Stage 5/5: running Day-1 primitives for NS.' - # add config if not present for NS charm - initial_config_primitive_list = self._get_initial_config_primitive_list(initial_config_primitive_list, - vca_deployed) + 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 + ) + check_if_terminated_needed = True 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"]: - deploy_params["ns_config_info"] = self._get_ns_config_info(vca_deployed_list) + deploy_params["ns_config_info"] = json.dumps(self._get_ns_config_info(nsr_id)) # TODO check if already done primitive_params_ = self._map_primitive_params(initial_config_primitive, {}, deploy_params) step = "execute primitive '{}' params '{}'".format(initial_config_primitive["name"], primitive_params_) self.logger.debug(logging_text + step) - await self.n2vc.exec_primitive( + await self.vca_map[vca_type].exec_primitive( ee_id=ee_id, primitive_name=initial_config_primitive["name"], params_dict=primitive_params_, db_dict=db_dict ) + # Once some primitive has been exec, check and write at db if it needs to exec terminated primitives + if check_if_terminated_needed: + if config_descriptor.get('terminate-config-primitive'): + self.update_db_2("nsrs", nsr_id, {db_update_entry + "needed_terminate": True}) + check_if_terminated_needed = False + # TODO register in database that primitive is done + # STEP 7 Configure metrics + if vca_type == "helm" or vca_type == "helm-v3": + prometheus_jobs = await self.add_prometheus_metrics( + 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, + ) + if prometheus_jobs: + self.update_db_2("nsrs", nsr_id, {db_update_entry + "prometheus_jobs": prometheus_jobs}) + step = "instantiated at VCA" self.logger.debug(logging_text + step) + 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 - raise Exception("{} {}".format(step, e)) from e - # TODO raise N2VC exception with 'step' extra information + # self.update_db_2("nsrs", nsr_id, {db_update_entry + "instantiation": "FAILED"}) + if not isinstance(e, (DbException, N2VCException, LcmException, asyncio.CancelledError)): + self.logger.error("Exception while {} : {}".format(step, e), exc_info=True) + self._write_configuration_status( + nsr_id=nsr_id, + vca_index=vca_index, + status='BROKEN' + ) + raise LcmException("{} {}".format(step, e)) from e - async def instantiate(self, nsr_id, nslcmop_id): + 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, other_update: dict = None): """ - - :param nsr_id: ns instance to deploy + Update db_nsr fields. + :param nsr_id: + :param ns_state: + :param current_operation: + :param current_operation_id: + :param error_description: + :param error_detail: + :param other_update: Other required changes at database if provided, will be cleared + :return: + """ + try: + db_dict = other_update or {} + db_dict["_admin.nslcmop"] = current_operation_id # for backward compatibility + db_dict["_admin.current-operation"] = current_operation_id + db_dict["_admin.operation-type"] = current_operation if current_operation != "IDLE" else None + db_dict["currentOperation"] = current_operation + db_dict["currentOperationID"] = current_operation_id + db_dict["errorDescription"] = error_description + db_dict["errorDetail"] = error_detail + + if ns_state: + db_dict["nsState"] = ns_state + self.update_db_2("nsrs", nsr_id, db_dict) + except DbException as e: + self.logger.warn('Error writing NS status, ns={}: {}'.format(nsr_id, e)) + + def _write_op_status(self, op_id: str, stage: list = None, error_message: str = None, queuePosition: int = 0, + operation_state: str = None, other_update: dict = None): + try: + db_dict = other_update or {} + db_dict['queuePosition'] = queuePosition + if isinstance(stage, list): + db_dict['stage'] = stage[0] + db_dict['detailed-status'] = " ".join(stage) + elif stage is not None: + db_dict['stage'] = str(stage) + + if error_message is not None: + db_dict['errorMessage'] = error_message + if operation_state is not None: + db_dict['operationState'] = operation_state + db_dict["statusEnteredTime"] = time() + self.update_db_2("nslcmops", op_id, db_dict) + except DbException as e: + self.logger.warn('Error writing OPERATION status for op_id: {} -> {}'.format(op_id, e)) + + def _write_all_config_status(self, db_nsr: dict, status: str): + try: + 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 + 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, + 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 = other_update or {} + if status: + 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 DbException as e: + 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): + """ + 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 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 * 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": nslcmop_id}) + pla_result = deep_get(db_nslcmop, ('_admin', 'pla')) + + if not pla_result: + 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']}) + # Modifies db_vnfrs + vnfr["vim-account-id"] = pla_vnf['vimAccountId'] + return modified + + def update_nsrs_with_pla_result(self, params): + try: + nslcmop_id = deep_get(params, ('placement', 'nslcmopId')) + self.update_db_2("nslcmops", nslcmop_id, {"_admin.pla": params.get('placement')}) + except Exception as e: + self.logger.warn('Update failed for nslcmop_id={}:{}'.format(nslcmop_id, e)) + + async def instantiate(self, nsr_id, nslcmop_id): + """ + + :param nsr_id: ns instance to deploy :param nslcmop_id: operation to run :return: """ @@ -1128,7 +1920,7 @@ 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) @@ -1143,40 +1935,63 @@ class NsLcm(LcmBase): db_nslcmop = None # update operation on nsrs - db_nsr_update = {"_admin.nslcmop": nslcmop_id, - "_admin.current-operation": nslcmop_id, - "_admin.operation-type": "instantiate"} - self.update_db_2("nsrs", nsr_id, db_nsr_update) - + db_nsr_update = {} # update operation on nslcmops db_nslcmop_update = {} nslcmop_operation_state = None db_vnfrs = {} # vnf's info indexed by member-index # n2vc_info = {} - task_instantiation_list = [] + tasks_dict_info = {} # from task to info text exc = None + error_list = [] + stage = ['Stage 1/5: preparation of the environment.', "Waiting for previous operations to terminate.", ""] + # ^ stage, step, VIM progress 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) + stage[1] = "Sync filesystem from database." + self.fs.sync() # TODO, make use of partial sync, only for the needed packages + # STEP 0: Reading database (nslcmops, nsrs, nsds, vnfrs, vnfds) + stage[1] = "Reading from database." + # nsState="BUILDING", currentOperation="INSTANTIATING", currentOperationID=nslcmop_id + db_nsr_update["detailed-status"] = "creating" + db_nsr_update["operational-status"] = "init" + self._write_ns_status( + nsr_id=nsr_id, + ns_state="BUILDING", + current_operation="INSTANTIATING", + current_operation_id=nslcmop_id, + other_update=db_nsr_update + ) + self._write_op_status( + op_id=nslcmop_id, + stage=stage, + queuePosition=0 + ) # read from db: operation - step = "Getting nslcmop={} from db".format(nslcmop_id) + stage[1] = "Getting nslcmop={} from db.".format(nslcmop_id) db_nslcmop = self.db.get_one("nslcmops", {"_id": nslcmop_id}) + ns_params = db_nslcmop.get("operationParams") + if ns_params and ns_params.get("timeout_ns_deploy"): + timeout_ns_deploy = ns_params["timeout_ns_deploy"] + else: + timeout_ns_deploy = self.timeout.get("ns_deploy", self.timeout_ns_deploy) # read from db: ns - step = "Getting nsr={} from db".format(nsr_id) + 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 - step = "Getting vnfrs from db" - self.logger.debug(logging_text + step) + stage[1] = "Getting vnfrs from db." + self.logger.debug(logging_text + stage[1]) db_vnfrs_list = self.db.get_list("vnfrs", {"nsr-id-ref": nsr_id}) # read from db: vnfd's for every vnf @@ -1189,11 +2004,12 @@ class NsLcm(LcmBase): db_vnfrs[vnfr["member-vnf-index-ref"]] = vnfr # vnf's dict indexed by member-index: '1', '2', etc vnfd_id = vnfr["vnfd-id"] # vnfd uuid for this vnf vnfd_ref = vnfr["vnfd-ref"] # vnfd name for this vnf + # if we haven't this vnfd, read it from db if vnfd_id not in db_vnfds: - # read from cb - step = "Getting vnfd={} id='{}' from db".format(vnfd_id, vnfd_ref) - self.logger.debug(logging_text + step) + # read from db + stage[1] = "Getting vnfd={} id='{}' from db.".format(vnfd_id, vnfd_ref) + self.logger.debug(logging_text + stage[1]) vnfd = self.db.get_one("vnfds", {"_id": vnfd_id}) # store vnfd @@ -1207,7 +2023,9 @@ 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): @@ -1216,9 +2034,6 @@ class NsLcm(LcmBase): db_nsr_update["_admin.deployed.VCA"] = vca_deployed_list populate_dict(db_nsr, ("_admin", "deployed", "VCA"), vca_deployed_list) - db_nsr_update["detailed-status"] = "creating" - db_nsr_update["operational-status"] = "init" - if not isinstance(deep_get(db_nsr, ("_admin", "deployed", "RO", "vnfd")), list): populate_dict(db_nsr, ("_admin", "deployed", "RO", "vnfd"), []) db_nsr_update["_admin.deployed.RO.vnfd"] = [] @@ -1226,18 +2041,28 @@ 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) - self.logger.debug(logging_text + "Before deploy_kdus") + self.db.set_list("vnfrs", {"nsr-id-ref": nsr_id}, {"_admin.nsState": "INSTANTIATED"}) + + # n2vc_redesign STEP 2 Deploy Network Scenario + stage[0] = 'Stage 2/5: deployment of KDUs, VMs and execution environments.' + self._write_op_status( + op_id=nslcmop_id, + stage=stage + ) + + stage[1] = "Deploying KDUs." + # self.logger.debug(logging_text + "Before deploy_kdus") # Call to deploy_kdus in case exists the "vdu:kdu" param - task_kdu = asyncio.ensure_future( - self.deploy_kdus( - logging_text=logging_text, - nsr_id=nsr_id, - db_nsr=db_nsr, - db_vnfrs=db_vnfrs, - ) + await self.deploy_kdus( + logging_text=logging_text, + nsr_id=nsr_id, + nslcmop_id=nslcmop_id, + db_vnfrs=db_vnfrs, + db_vnfds=db_vnfds, + task_instantiation_info=tasks_dict_info, ) - self.lcm_tasks.register("ns", nsr_id, nslcmop_id, "instantiate_KDUs", task_kdu) - task_instantiation_list.append(task_kdu) + + stage[1] = "Getting VCA public key." # n2vc_redesign STEP 1 Get VCA public ssh-key # feature 1429. Add n2vc public key to needed VMs n2vc_key = self.n2vc.get_public_key() @@ -1245,7 +2070,7 @@ 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 + stage[1] = "Deploying NS at VIM." task_ro = asyncio.ensure_future( self.instantiate_RO( logging_text=logging_text, @@ -1255,15 +2080,16 @@ class NsLcm(LcmBase): db_nslcmop=db_nslcmop, db_vnfrs=db_vnfrs, db_vnfds_ref=db_vnfds_ref, - n2vc_key_list=n2vc_key_list + n2vc_key_list=n2vc_key_list, + stage=stage ) ) self.lcm_tasks.register("ns", nsr_id, nslcmop_id, "instantiate_RO", task_ro) - task_instantiation_list.append(task_ro) + tasks_dict_info[task_ro] = "Deploying at VIM" # n2vc_redesign STEP 3 to 6 Deploy N2VC - step = "Looking for needed vnfd to configure with proxy charm" - self.logger.debug(logging_text + step) + stage[1] = "Deploying Execution Environments." + self.logger.debug(logging_text + stage[1]) nsi_id = None # TODO put nsi_id when this nsr belongs to a NSI # get_iterable() returns a value from a dict or empty tuple if key does not exist @@ -1279,14 +2105,14 @@ class NsLcm(LcmBase): kdu_name = None # Get additional parameters - deploy_params = {} + deploy_params = {"OSM": self._get_osm_params(db_vnfr)} if db_vnfr.get("additionalParamsForVnf"): - deploy_params = self._format_additional_params(db_vnfr["additionalParamsForVnf"].copy()) + deploy_params.update(self._format_additional_params(db_vnfr["additionalParamsForVnf"].copy())) descriptor_config = vnfd.get("vnf-configuration") - if descriptor_config and descriptor_config.get("juju"): + if descriptor_config: self._deploy_n2vc( - logging_text=logging_text, + logging_text=logging_text + "member_vnf_index={} ".format(member_vnf_index), db_nsr=db_nsr, db_vnfr=db_vnfr, nslcmop_id=nslcmop_id, @@ -1301,7 +2127,8 @@ class NsLcm(LcmBase): deploy_params=deploy_params, descriptor_config=descriptor_config, base_folder=base_folder, - task_instantiation_list=task_instantiation_list + task_instantiation_info=tasks_dict_info, + stage=stage ) # Deploy charms for each VDU that supports one. @@ -1313,21 +2140,15 @@ class NsLcm(LcmBase): deploy_params_vdu = self._format_additional_params(vdur["additionalParams"]) else: deploy_params_vdu = deploy_params - if descriptor_config and descriptor_config.get("juju"): - # look for vdu index in the db_vnfr["vdu"] section - # for vdur_index, vdur in enumerate(db_vnfr["vdur"]): - # if vdur["vdu-id-ref"] == vdu_id: - # break - # else: - # raise LcmException("Mismatch vdu_id={} not found in the vnfr['vdur'] list for " - # "member_vnf_index={}".format(vdu_id, member_vnf_index)) - # vdu_name = vdur.get("name") + deploy_params_vdu["OSM"] = self._get_osm_params(db_vnfr, vdu_id, vdu_count_index=0) + if descriptor_config: vdu_name = None kdu_name = None for vdu_index in range(int(vdud.get("count", 1))): # TODO vnfr_params["rw_mgmt_ip"] = vdur["ip-address"] self._deploy_n2vc( - logging_text=logging_text, + logging_text=logging_text + "member_vnf_index={}, vdu_id={}, vdu_index={} ".format( + member_vnf_index, vdu_id, vdu_index), db_nsr=db_nsr, db_vnfr=db_vnfr, nslcmop_id=nslcmop_id, @@ -1342,24 +2163,20 @@ class NsLcm(LcmBase): deploy_params=deploy_params_vdu, descriptor_config=descriptor_config, base_folder=base_folder, - task_instantiation_list=task_instantiation_list + task_instantiation_info=tasks_dict_info, + stage=stage ) for kdud in get_iterable(vnfd, 'kdu'): kdu_name = kdud["name"] descriptor_config = kdud.get('kdu-configuration') - if descriptor_config and descriptor_config.get("juju"): + if descriptor_config: vdu_id = None vdu_index = 0 vdu_name = None - # look for vdu index in the db_vnfr["vdu"] section - # for vdur_index, vdur in enumerate(db_vnfr["vdur"]): - # if vdur["vdu-id-ref"] == vdu_id: - # break - # else: - # raise LcmException("Mismatch vdu_id={} not found in the vnfr['vdur'] list for " - # "member_vnf_index={}".format(vdu_id, member_vnf_index)) - # vdu_name = vdur.get("name") - # vdu_name = None + kdur = next(x for x in db_vnfr["kdur"] if x["kdu-name"] == kdu_name) + deploy_params_kdu = {"OSM": self._get_osm_params(db_vnfr)} + if kdur.get("additionalParams"): + deploy_params_kdu = self._format_additional_params(kdur["additionalParams"]) self._deploy_n2vc( logging_text=logging_text, @@ -1374,10 +2191,11 @@ class NsLcm(LcmBase): member_vnf_index=member_vnf_index, vdu_index=vdu_index, vdu_name=vdu_name, - deploy_params=deploy_params, + deploy_params=deploy_params_kdu, descriptor_config=descriptor_config, base_folder=base_folder, - task_instantiation_list=task_instantiation_list + task_instantiation_info=tasks_dict_info, + stage=stage ) # Check if this NS has a charm configuration @@ -1392,9 +2210,9 @@ class NsLcm(LcmBase): vdu_name = None # Get additional parameters - deploy_params = {} + deploy_params = {"OSM": self._get_osm_params(db_vnfr)} if db_nsr.get("additionalParamsForNs"): - deploy_params = self._format_additional_params(db_nsr["additionalParamsForNs"].copy()) + deploy_params.update(self._format_additional_params(db_nsr["additionalParamsForNs"].copy())) base_folder = nsd["_admin"]["storage"] self._deploy_n2vc( logging_text=logging_text, @@ -1412,74 +2230,88 @@ class NsLcm(LcmBase): deploy_params=deploy_params, descriptor_config=descriptor_config, base_folder=base_folder, - task_instantiation_list=task_instantiation_list + task_instantiation_info=tasks_dict_info, + stage=stage ) - # Wait until all tasks of "task_instantiation_list" have been finished - - # while time() <= start_deploy + self.total_deploy_timeout: - error_text = None - timeout = 3600 # time() - start_deploy - task_instantiation_set = set(task_instantiation_list) # build a set with tasks - done = None - pending = None - if len(task_instantiation_set) > 0: - done, pending = await asyncio.wait(task_instantiation_set, timeout=timeout) - if pending: - error_text = "timeout" - for task in done: - if task.cancelled(): - if not error_text: - error_text = "cancelled" - elif task.done(): - exc = task.exception() - if exc: - error_text = str(exc) - - if error_text: - db_nsr_update["config-status"] = "failed" - error_text = "fail configuring " + error_text - db_nsr_update["detailed-status"] = error_text - db_nslcmop_update["operationState"] = nslcmop_operation_state = "FAILED_TEMP" - db_nslcmop_update["detailed-status"] = error_text - db_nslcmop_update["statusEnteredTime"] = time() - else: - # all is done - db_nslcmop_update["operationState"] = nslcmop_operation_state = "COMPLETED" - db_nslcmop_update["statusEnteredTime"] = time() - db_nslcmop_update["detailed-status"] = "done" - db_nsr_update["config-status"] = "configured" - db_nsr_update["detailed-status"] = "done" + # rest of staff will be done at finally - except (ROclient.ROClientException, DbException, LcmException) as e: - self.logger.error(logging_text + "Exit Exception while '{}': {}".format(step, e)) + except (ROclient.ROClientException, DbException, LcmException, N2VCException) as e: + self.logger.error(logging_text + "Exit Exception while '{}': {}".format(stage[1], e)) exc = e except asyncio.CancelledError: - self.logger.error(logging_text + "Cancelled Exception while '{}'".format(step)) + self.logger.error(logging_text + "Cancelled Exception while '{}'".format(stage[1])) exc = "Operation was cancelled" except Exception as e: exc = traceback.format_exc() - self.logger.critical(logging_text + "Exit Exception {} while '{}': {}".format(type(e).__name__, step, e), - exc_info=True) + self.logger.critical(logging_text + "Exit Exception while '{}': {}".format(stage[1], e), exc_info=True) finally: if exc: - if db_nsr: - db_nsr_update["detailed-status"] = "ERROR {}: {}".format(step, exc) - db_nsr_update["operational-status"] = "failed" - 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() + error_list.append(str(exc)) try: - if db_nsr: - db_nsr_update["_admin.nslcmop"] = None - db_nsr_update["_admin.current-operation"] = None - db_nsr_update["_admin.operation-type"] = None - self.update_db_2("nsrs", nsr_id, db_nsr_update) - if db_nslcmop_update: - self.update_db_2("nslcmops", nslcmop_id, db_nslcmop_update) - except DbException as e: - self.logger.error(logging_text + "Cannot update database: {}".format(e)) + # wait for pending tasks + if tasks_dict_info: + stage[1] = "Waiting for instantiate pending tasks." + self.logger.debug(logging_text + stage[1]) + error_list += await self._wait_for_tasks(logging_text, tasks_dict_info, timeout_ns_deploy, + stage, nslcmop_id, nsr_id=nsr_id) + stage[1] = stage[2] = "" + except asyncio.CancelledError: + error_list.append("Cancelled") + # TODO cancel all tasks + except Exception as exc: + error_list.append(str(exc)) + + # update operation-status + db_nsr_update["operational-status"] = "running" + # let's begin with VCA 'configured' status (later we can change it) + db_nsr_update["config-status"] = "configured" + 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): + # A N2VC task is pending + db_nsr_update["config-status"] = "failed" + else: + # RO or KDU task is pending + db_nsr_update["operational-status"] = "failed" + + # update status at database + if error_list: + error_detail = ". ".join(error_list) + self.logger.error(logging_text + error_detail) + error_description_nslcmop = '{} Detail: {}'.format(stage[0], error_detail) + error_description_nsr = 'Operation: INSTANTIATING.{}, {}'.format(nslcmop_id, stage[0]) + + db_nsr_update["detailed-status"] = error_description_nsr + " Detail: " + error_detail + db_nslcmop_update["detailed-status"] = error_detail + nslcmop_operation_state = "FAILED" + ns_state = "BROKEN" + else: + error_detail = None + error_description_nsr = error_description_nslcmop = None + ns_state = "READY" + db_nsr_update["detailed-status"] = "Done" + db_nslcmop_update["detailed-status"] = "Done" + nslcmop_operation_state = "COMPLETED" + + if db_nsr: + self._write_ns_status( + nsr_id=nsr_id, + ns_state=ns_state, + current_operation="IDLE", + current_operation_id=None, + error_description=error_description_nsr, + error_detail=error_detail, + 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", "instantiated", {"nsr_id": nsr_id, "nslcmop_id": nslcmop_id, @@ -1491,178 +2323,525 @@ class NsLcm(LcmBase): self.logger.debug(logging_text + "Exit") self.lcm_tasks.remove("ns", nsr_id, nslcmop_id, "ns_instantiate") - async def deploy_kdus(self, logging_text, nsr_id, db_nsr, db_vnfrs): + async def _add_vca_relations(self, logging_text, nsr_id, vca_index: int, + timeout: int = 3600, vca_type: str = None) -> bool: + + # steps: + # 1. find all relations for this VCA + # 2. wait for other peers related + # 3. add relations + + try: + vca_type = vca_type or "lxc_proxy_charm" + + # STEP 1: find all relations for this VCA + + # 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(nsd, ('ns-configuration', 'relation')) + if db_ns_relations: + for r in db_ns_relations: + # check if this VCA is in the relation + if my_vca.get('member-vnf-index') in\ + (r.get('entities')[0].get('id'), r.get('entities')[1].get('id')): + ns_relations.append(r) + + # read all vnf-configuration relations + vnf_relations = list() + db_vnfd_list = db_nsr.get('vnfd-id') + if db_vnfd_list: + for vnfd in db_vnfd_list: + db_vnfd = self.db.get_one("vnfds", {"_id": vnfd}) + db_vnf_relations = deep_get(db_vnfd, ('vnf-configuration', 'relation')) + if db_vnf_relations: + for r in db_vnf_relations: + # check if this VCA is in the relation + if my_vca.get('vdu_id') in (r.get('entities')[0].get('id'), r.get('entities')[1].get('id')): + vnf_relations.append(r) + + # if no relations, terminate + if not ns_relations and not vnf_relations: + self.logger.debug(logging_text + ' No relations') + return True + + self.logger.debug(logging_text + ' adding relations\n {}\n {}'.format(ns_relations, vnf_relations)) + + # add all relations + start = time() + while True: + # check timeout + now = time() + if now - start >= timeout: + self.logger.error(logging_text + ' : timeout adding relations') + return False + + # reload nsr from database (we need to update record: _admin.deloyed.VCA) + db_nsr = self.db.get_one("nsrs", {"_id": nsr_id}) + + # for each defined NS relation, find the VCA's related + for r in ns_relations.copy(): + from_vca_ee_id = None + to_vca_ee_id = None + from_vca_endpoint = None + to_vca_endpoint = None + vca_list = deep_get(db_nsr, ('_admin', 'deployed', 'VCA')) + for vca in vca_list: + if vca.get('member-vnf-index') == 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('member-vnf-index') == 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: + # add relation + await self.vca_map[vca_type].add_relation( + ee_id_1=from_vca_ee_id, + ee_id_2=to_vca_ee_id, + endpoint_1=from_vca_endpoint, + endpoint_2=to_vca_endpoint) + # remove entry from relations list + ns_relations.remove(r) + else: + # check failed peers + try: + vca_status_list = db_nsr.get('configurationStatus') + if vca_status_list: + for i in range(len(vca_list)): + vca = vca_list[i] + vca_status = vca_status_list[i] + if vca.get('member-vnf-index') == r.get('entities')[0].get('id'): + if vca_status.get('status') == 'BROKEN': + # peer broken: remove relation from list + ns_relations.remove(r) + if vca.get('member-vnf-index') == r.get('entities')[1].get('id'): + if vca_status.get('status') == 'BROKEN': + # peer broken: remove relation from list + ns_relations.remove(r) + except Exception: + # ignore + pass + + # for each defined VNF relation, find the VCA's related + for r in vnf_relations.copy(): + from_vca_ee_id = None + to_vca_ee_id = None + from_vca_endpoint = None + to_vca_endpoint = None + vca_list = deep_get(db_nsr, ('_admin', 'deployed', 'VCA')) + for vca in vca_list: + 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(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: + # add relation + await self.vca_map[vca_type].add_relation( + ee_id_1=from_vca_ee_id, + ee_id_2=to_vca_ee_id, + endpoint_1=from_vca_endpoint, + endpoint_2=to_vca_endpoint) + # remove entry from relations list + vnf_relations.remove(r) + else: + # check failed peers + try: + vca_status_list = db_nsr.get('configurationStatus') + if vca_status_list: + for i in range(len(vca_list)): + vca = vca_list[i] + vca_status = vca_status_list[i] + if vca.get('vdu_id') == r.get('entities')[0].get('id'): + if vca_status.get('status') == 'BROKEN': + # peer broken: remove relation from list + vnf_relations.remove(r) + if vca.get('vdu_id') == r.get('entities')[1].get('id'): + if vca_status.get('status') == 'BROKEN': + # peer broken: remove relation from list + vnf_relations.remove(r) + except Exception: + # ignore + pass + + # wait for next try + await asyncio.sleep(5.0) + + if not ns_relations and not vnf_relations: + self.logger.debug('Relations added') + break + + return True + + except Exception as e: + self.logger.warn(logging_text + ' ERROR adding relations: {}'.format(e)) + return False + + async def _install_kdu(self, nsr_id: str, nsr_db_path: str, vnfr_data: dict, kdu_index: int, kdud: dict, + vnfd: dict, k8s_instance_info: dict, k8params: dict = None, timeout: int = 600): + + try: + k8sclustertype = k8s_instance_info["k8scluster-type"] + # Instantiate kdu + db_dict_install = {"collection": "nsrs", + "filter": {"_id": nsr_id}, + "path": nsr_db_path} + + kdu_instance = await self.k8scluster_map[k8sclustertype].install( + cluster_uuid=k8s_instance_info["k8scluster-uuid"], + kdu_model=k8s_instance_info["kdu-model"], + atomic=True, + params=k8params, + db_dict=db_dict_install, + timeout=timeout, + kdu_name=k8s_instance_info["kdu-name"], + namespace=k8s_instance_info["namespace"]) + self.update_db_2("nsrs", nsr_id, {nsr_db_path + ".kdu-instance": kdu_instance}) + + # Obtain services to obtain management service ip + services = await self.k8scluster_map[k8sclustertype].get_services( + cluster_uuid=k8s_instance_info["k8scluster-uuid"], + kdu_instance=kdu_instance, + namespace=k8s_instance_info["namespace"]) + + # Obtain management service info (if exists) + vnfr_update_dict = {} + if services: + vnfr_update_dict["kdur.{}.services".format(kdu_index)] = services + mgmt_services = [service for service in kdud.get("service", []) if service.get("mgmt-service")] + for mgmt_service in mgmt_services: + for service in services: + if service["name"].startswith(mgmt_service["name"]): + # Mgmt service found, Obtain service ip + ip = service.get("external_ip", service.get("cluster_ip")) + if isinstance(ip, list) and len(ip) == 1: + ip = ip[0] + + vnfr_update_dict["kdur.{}.ip-address".format(kdu_index)] = ip + + # Check if must update also mgmt ip at the vnf + service_external_cp = mgmt_service.get("external-connection-point-ref") + if service_external_cp: + if deep_get(vnfd, ("mgmt-interface", "cp")) == service_external_cp: + vnfr_update_dict["ip-address"] = ip + + break + else: + self.logger.warn("Mgmt service name: {} not found".format(mgmt_service["name"])) + + vnfr_update_dict["kdur.{}.status".format(kdu_index)] = "READY" + self.update_db_2("vnfrs", vnfr_data.get("_id"), vnfr_update_dict) + + kdu_config = kdud.get("kdu-configuration") + if kdu_config and kdu_config.get("initial-config-primitive") and kdu_config.get("juju") is None: + initial_config_primitive_list = kdu_config.get("initial-config-primitive") + initial_config_primitive_list.sort(key=lambda val: int(val["seq"])) + + for initial_config_primitive in initial_config_primitive_list: + primitive_params_ = self._map_primitive_params(initial_config_primitive, {}, {}) + + await asyncio.wait_for( + self.k8scluster_map[k8sclustertype].exec_primitive( + cluster_uuid=k8s_instance_info["k8scluster-uuid"], + kdu_instance=kdu_instance, + primitive_name=initial_config_primitive["name"], + params=primitive_params_, db_dict={}), + timeout=timeout) + + except Exception as e: + # Prepare update db with error and raise exception + try: + self.update_db_2("nsrs", nsr_id, {nsr_db_path + ".detailed-status": str(e)}) + self.update_db_2("vnfrs", vnfr_data.get("_id"), {"kdur.{}.status".format(kdu_index): "ERROR"}) + except Exception: + # ignore to keep original exception + pass + # reraise original error + raise + + return kdu_instance + + 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 - k8scluster_id_2_uuic = {"helm-chart": {}, "juju-bundle": {}} + k8scluster_id_2_uuic = {"helm-chart-v3": {}, "helm-chart": {}, "juju-bundle": {}} - def _get_cluster_id(cluster_id, cluster_type): + async def _get_cluster_id(cluster_id, cluster_type): nonlocal k8scluster_id_2_uuic if cluster_id in k8scluster_id_2_uuic[cluster_type]: return k8scluster_id_2_uuic[cluster_type][cluster_id] + # check if K8scluster is creating and wait look if previous tasks in process + task_name, task_dependency = self.lcm_tasks.lookfor_related("k8scluster", cluster_id) + if task_dependency: + text = "Waiting for related tasks '{}' on k8scluster {} to be completed".format(task_name, cluster_id) + self.logger.debug(logging_text + text) + await asyncio.wait(task_dependency, timeout=3600) + db_k8scluster = self.db.get_one("k8sclusters", {"_id": cluster_id}, fail_on_empty=False) if not db_k8scluster: raise LcmException("K8s cluster {} cannot be found".format(cluster_id)) + k8s_id = deep_get(db_k8scluster, ("_admin", cluster_type, "id")) if not k8s_id: - raise LcmException("K8s cluster '{}' has not been initilized for '{}'".format(cluster_id, cluster_type)) + if cluster_type == "helm-chart-v3": + try: + # backward compatibility for existing clusters that have not been initialized for helm v3 + k8s_credentials = yaml.safe_dump(db_k8scluster.get("credentials")) + k8s_id, uninstall_sw = await self.k8sclusterhelm3.init_env(k8s_credentials, + reuse_cluster_uuid=cluster_id) + db_k8scluster_update = {} + db_k8scluster_update["_admin.helm-chart-v3.error_msg"] = None + db_k8scluster_update["_admin.helm-chart-v3.id"] = k8s_id + db_k8scluster_update["_admin.helm-chart-v3.created"] = uninstall_sw + db_k8scluster_update["_admin.helm-chart-v3.operationalState"] = "ENABLED" + self.update_db_2("k8sclusters", cluster_id, db_k8scluster_update) + except Exception as e: + self.logger.error(logging_text + "error initializing helm-v3 cluster: {}".format(str(e))) + raise LcmException("K8s cluster '{}' has not been initialized for '{}'".format(cluster_id, + cluster_type)) + else: + raise LcmException("K8s cluster '{}' has not been initialized for '{}'". + format(cluster_id, cluster_type)) k8scluster_id_2_uuic[cluster_type][cluster_id] = k8s_id return k8s_id logging_text += "Deploy kdus: " + step = "" try: db_nsr_update = {"_admin.deployed.K8s": []} self.update_db_2("nsrs", nsr_id, db_nsr_update) - # Look for all vnfds - pending_tasks = {} index = 0 + updated_cluster_list = [] + updated_v3_cluster_list = [] + for vnfr_data in db_vnfrs.values(): - for kdur in get_iterable(vnfr_data, "kdur"): + for kdu_index, kdur in enumerate(get_iterable(vnfr_data, "kdur")): + # Step 0: Prepare and set parameters desc_params = self._format_additional_params(kdur.get("additionalParams")) - kdumodel = None - k8sclustertype = None - error_text = None - cluster_uuid = None + vnfd_id = vnfr_data.get('vnfd-id') + kdud = next(kdud for kdud in db_vnfds[vnfd_id]["kdu"] if kdud["name"] == kdur["kdu-name"]) + namespace = kdur.get("k8s-namespace") if kdur.get("helm-chart"): kdumodel = kdur["helm-chart"] - k8sclustertype = "chart" - k8sclustertype_full = "helm-chart" + # 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" - k8sclustertype_full = "juju-bundle" + k8sclustertype = "juju-bundle" else: - error_text = "kdu type is neither helm-chart not juju-bundle. Maybe an old NBI version is" \ - " running" + raise LcmException("kdu type for kdu='{}.{}' is neither helm-chart nor " + "juju-bundle. Maybe an old NBI version is running". + format(vnfr_data["member-vnf-index-ref"], kdur["kdu-name"])) + # check if kdumodel is a file and exists try: - if not error_text: - cluster_uuid = _get_cluster_id(kdur["k8s-cluster"]["id"], k8sclustertype_full) - except LcmException as e: - error_text = str(e) - step = "Instantiate KDU {} in k8s cluster {}".format(kdur["kdu-name"], cluster_uuid) - - k8s_instace_info = {"kdu-instance": None, "k8scluster-uuid": cluster_uuid, - "k8scluster-type": k8sclustertype, - "kdu-name": kdur["kdu-name"], "kdu-model": kdumodel} - if error_text: - k8s_instace_info["detailed-status"] = error_text - db_nsr_update["_admin.deployed.K8s.{}".format(index)] = k8s_instace_info + 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 + + k8s_cluster_id = kdur["k8s-cluster"]["id"] + step = "Synchronize repos for k8s cluster '{}'".format(k8s_cluster_id) + cluster_uuid = await _get_cluster_id(k8s_cluster_id, k8sclustertype) + + # Synchronize repos + if (k8sclustertype == "helm-chart" and cluster_uuid not in updated_cluster_list)\ + or (k8sclustertype == "helm-chart-v3" and cluster_uuid not in updated_v3_cluster_list): + del_repo_list, added_repo_dict = await asyncio.ensure_future( + self.k8scluster_map[k8sclustertype].synchronize_repos(cluster_uuid=cluster_uuid)) + if del_repo_list or added_repo_dict: + if k8sclustertype == "helm-chart": + unset = {'_admin.helm_charts_added.' + item: None for item in del_repo_list} + updated = {'_admin.helm_charts_added.' + + item: name for item, name in added_repo_dict.items()} + updated_cluster_list.append(cluster_uuid) + elif k8sclustertype == "helm-chart-v3": + unset = {'_admin.helm_charts_v3_added.' + item: None for item in del_repo_list} + updated = {'_admin.helm_charts_v3_added.' + + item: name for item, name in added_repo_dict.items()} + updated_v3_cluster_list.append(cluster_uuid) + self.logger.debug(logging_text + "repos synchronized on k8s cluster " + "'{}' to_delete: {}, to_add: {}". + format(k8s_cluster_id, del_repo_list, added_repo_dict)) + self.db.set_one("k8sclusters", {"_id": k8s_cluster_id}, updated, unset=unset) + + # Instantiate kdu + step = "Instantiating KDU {}.{} in k8s cluster {}".format(vnfr_data["member-vnf-index-ref"], + kdur["kdu-name"], k8s_cluster_id) + k8s_instance_info = {"kdu-instance": None, + "k8scluster-uuid": cluster_uuid, + "k8scluster-type": k8sclustertype, + "member-vnf-index": vnfr_data["member-vnf-index-ref"], + "kdu-name": kdur["kdu-name"], + "kdu-model": kdumodel, + "namespace": namespace} + db_path = "_admin.deployed.K8s.{}".format(index) + db_nsr_update[db_path] = k8s_instance_info self.update_db_2("nsrs", nsr_id, db_nsr_update) - if error_text: - continue - db_dict = {"collection": "nsrs", "filter": {"_id": nsr_id}, "path": "_admin.deployed.K8s." - "{}".format(index)} - if k8sclustertype == "chart": - task = asyncio.ensure_future( - self.k8sclusterhelm.install(cluster_uuid=cluster_uuid, kdu_model=kdumodel, atomic=True, - params=desc_params, db_dict=db_dict, timeout=3600) - ) - else: - task = self.k8sclusterjuju.install(cluster_uuid=cluster_uuid, kdu_model=kdumodel, - atomic=True, params=desc_params, - db_dict=db_dict, timeout=600) + task = asyncio.ensure_future( + self._install_kdu(nsr_id, db_path, vnfr_data, kdu_index, kdud, db_vnfds[vnfd_id], + k8s_instance_info, k8params=desc_params, timeout=600)) + self.lcm_tasks.register("ns", nsr_id, nslcmop_id, "instantiate_KDU-{}".format(index), task) + task_instantiation_info[task] = "Deploying KDU {}".format(kdur["kdu-name"]) - pending_tasks[task] = "_admin.deployed.K8s.{}.".format(index) index += 1 - if not pending_tasks: - return - self.logger.debug(logging_text + 'Waiting for terminate pending tasks...') - pending_list = list(pending_tasks.keys()) - while pending_list: - done_list, pending_list = await asyncio.wait(pending_list, timeout=30*60, - return_when=asyncio.FIRST_COMPLETED) - if not done_list: # timeout - for task in pending_list: - db_nsr_update[pending_tasks(task) + "detailed-status"] = "Timeout" - break - for task in done_list: - exc = task.exception() - if exc: - db_nsr_update[pending_tasks[task] + "detailed-status"] = "{}".format(exc) - else: - db_nsr_update[pending_tasks[task] + "kdu-instance"] = task.result() + except (LcmException, asyncio.CancelledError): + raise 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)) + msg = "Exception {} while {}: {}".format(type(e).__name__, step, e) + if isinstance(e, (N2VCException, DbException)): + self.logger.error(logging_text + msg) + else: + self.logger.critical(logging_text + msg, exc_info=True) + raise LcmException(msg) finally: - # TODO Write in data base if db_nsr_update: self.update_db_2("nsrs", nsr_id, db_nsr_update) def _deploy_n2vc(self, logging_text, db_nsr, db_vnfr, nslcmop_id, nsr_id, nsi_id, vnfd_id, vdu_id, kdu_name, member_vnf_index, vdu_index, vdu_name, deploy_params, descriptor_config, - base_folder, task_instantiation_list): + base_folder, task_instantiation_info, stage): # launch instantiate_N2VC in a asyncio task and register task object # Look where information of this charm is at database ._admin.deployed.VCA # if not found, create one entry and update database - # fill db_nsr._admin.deployed.VCA. - vca_index = -1 - for vca_index, vca_deployed in enumerate(db_nsr["_admin"]["deployed"]["VCA"]): - if not vca_deployed: + + self.logger.debug(logging_text + "_deploy_n2vc vnfd_id={}, vdu_id={}".format(vnfd_id, vdu_id)) + if descriptor_config.get("juju"): # There is one execution envioronment of type juju + ee_list = [descriptor_config] + elif descriptor_config.get("execution-environment-list"): + ee_list = descriptor_config.get("execution-environment-list") + else: # other types as script are not supported + ee_list = [] + + for ee_item in ee_list: + self.logger.debug(logging_text + "_deploy_n2vc ee_item juju={}, helm={}".format(ee_item.get('juju'), + ee_item.get("helm-chart"))) + ee_descriptor_id = ee_item.get("id") + if ee_item.get("juju"): + vca_name = ee_item['juju'].get('charm') + vca_type = "lxc_proxy_charm" if ee_item['juju'].get('charm') is not None else "native_charm" + if ee_item['juju'].get('cloud') == "k8s": + vca_type = "k8s_proxy_charm" + elif ee_item['juju'].get('proxy') is False: + 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" + else: + self.logger.debug(logging_text + "skipping non juju neither charm configuration") continue - if vca_deployed.get("member-vnf-index") == member_vnf_index and \ - vca_deployed.get("vdu_id") == vdu_id and \ - vca_deployed.get("kdu_name") == kdu_name and \ - vca_deployed.get("vdu_count_index", 0) == vdu_index: - break - else: - # not found, create one. - vca_deployed = { - "member-vnf-index": member_vnf_index, - "vdu_id": vdu_id, - "kdu_name": kdu_name, - "vdu_count_index": vdu_index, - "operational-status": "init", # TODO revise - "detailed-status": "", # TODO revise - "step": "initial-deploy", # TODO revise - "vnfd_id": vnfd_id, - "vdu_name": vdu_name, - } - vca_index += 1 - self.update_db_2("nsrs", nsr_id, {"_admin.deployed.VCA.{}".format(vca_index): vca_deployed}) - db_nsr["_admin"]["deployed"]["VCA"].append(vca_deployed) - # Launch task - task_n2vc = asyncio.ensure_future( - self.instantiate_N2VC( - logging_text=logging_text, - vca_index=vca_index, - nsi_id=nsi_id, - db_nsr=db_nsr, - db_vnfr=db_vnfr, - vdu_id=vdu_id, - kdu_name=kdu_name, - vdu_index=vdu_index, - deploy_params=deploy_params, - config_descriptor=descriptor_config, - base_folder=base_folder, + vca_index = -1 + for vca_index, vca_deployed in enumerate(db_nsr["_admin"]["deployed"]["VCA"]): + if not vca_deployed: + continue + if vca_deployed.get("member-vnf-index") == member_vnf_index and \ + vca_deployed.get("vdu_id") == vdu_id and \ + vca_deployed.get("kdu_name") == kdu_name and \ + vca_deployed.get("vdu_count_index", 0) == vdu_index and \ + vca_deployed.get("ee_descriptor_id") == ee_descriptor_id: + break + else: + # not found, create one. + target = "ns" if not member_vnf_index else "vnf/{}".format(member_vnf_index) + if vdu_id: + target += "/vdu/{}/{}".format(vdu_id, vdu_index or 0) + elif kdu_name: + target += "/kdu/{}".format(kdu_name) + vca_deployed = { + "target_element": target, + # ^ target_element will replace member-vnf-index, kdu_name, vdu_id ... in a single string + "member-vnf-index": member_vnf_index, + "vdu_id": vdu_id, + "kdu_name": kdu_name, + "vdu_count_index": vdu_index, + "operational-status": "init", # TODO revise + "detailed-status": "", # TODO revise + "step": "initial-deploy", # TODO revise + "vnfd_id": vnfd_id, + "vdu_name": vdu_name, + "type": vca_type, + "ee_descriptor_id": ee_descriptor_id + } + vca_index += 1 + + # 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 + task_n2vc = asyncio.ensure_future( + self.instantiate_N2VC( + logging_text=logging_text, + vca_index=vca_index, + nsi_id=nsi_id, + db_nsr=db_nsr, + db_vnfr=db_vnfr, + vdu_id=vdu_id, + kdu_name=kdu_name, + vdu_index=vdu_index, + deploy_params=deploy_params, + config_descriptor=descriptor_config, + base_folder=base_folder, + nslcmop_id=nslcmop_id, + stage=stage, + vca_type=vca_type, + vca_name=vca_name, + ee_config_descriptor=ee_item + ) ) - ) - self.lcm_tasks.register("ns", nsr_id, nslcmop_id, "instantiate_N2VC-{}".format(vca_index), task_n2vc) - task_instantiation_list.append(task_n2vc) - - # Check if this VNFD has a configured terminate action - def _has_terminate_config_primitive(self, vnfd): - vnf_config = vnfd.get("vnf-configuration") - if vnf_config and vnf_config.get("terminate-config-primitive"): - return True - else: - return False + self.lcm_tasks.register("ns", nsr_id, nslcmop_id, "instantiate_N2VC-{}".format(vca_index), task_n2vc) + task_instantiation_info[task_n2vc] = self.task_name_deploy_vca + " {}.{}".format( + member_vnf_index or "", vdu_id or "") @staticmethod - def _get_terminate_config_primitive_seq_list(vnfd): - """ Get a numerically sorted list of the sequences for this VNFD's terminate action """ - # No need to check for existing primitive twice, already done before - vnf_config = vnfd.get("vnf-configuration") - seq_list = vnf_config.get("terminate-config-primitive") - # Get all 'seq' tags in seq_list, order sequences numerically, ascending. - seq_list_sorted = sorted(seq_list, key=lambda x: int(x['seq'])) - return seq_list_sorted + def _get_terminate_config_primitive(primitive_list, vca_deployed): + """ Get a sorted terminate config primitive list. In case ee_descriptor_id is present at vca_deployed, + it get only those primitives for this execution envirom""" + + primitive_list = primitive_list or [] + # filter primitives by ee_descriptor_id + ee_descriptor_id = vca_deployed.get("ee_descriptor_id") + primitive_list = [p for p in primitive_list if p.get("execution-environment-ref") == ee_descriptor_id] + + if primitive_list: + primitive_list.sort(key=lambda val: int(val['seq'])) + + return primitive_list @staticmethod def _create_nslcmop(nsr_id, operation, params): @@ -1718,16 +2897,16 @@ 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( @@ -1740,7 +2919,7 @@ 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): @@ -1768,7 +2947,7 @@ class NsLcm(LcmBase): def _add_suboperation(self, db_nslcmop, vnf_index, vdu_id, vdu_count_index, vdu_name, primitive, mapped_primitive_params, operationState=None, detailed_status=None, operationType=None, RO_nsr_id=None, RO_scaling_info=None): - if not (db_nslcmop): + if not db_nslcmop: return self.SUBOPERATION_STATUS_NOT_FOUND # Get the "_admin.operations" list, if it exists db_nslcmop_admin = db_nslcmop.get('_admin', {}) @@ -1808,11 +2987,11 @@ 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, @@ -1827,7 +3006,7 @@ 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 @@ -1835,7 +3014,7 @@ 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: @@ -1861,83 +3040,240 @@ 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) - - # Helper methods for terminate() - - async def _terminate_action(self, db_nslcmop, nslcmop_id, nsr_id): - """ Create a primitive with params from VNFD - Called from terminate() before deleting instance - Calls action() to execute the primitive """ - logging_text = "Task ns={} _terminate_action={} ".format(nsr_id, nslcmop_id) - db_vnfrs_list = self.db.get_list("vnfrs", {"nsr-id-ref": nsr_id}) - db_vnfds = {} - # Loop over VNFRs - for vnfr in db_vnfrs_list: - vnfd_id = vnfr["vnfd-id"] - vnf_index = vnfr["member-vnf-index-ref"] - if vnfd_id not in db_vnfds: - step = "Getting vnfd={} id='{}' from db".format(vnfd_id, vnfd_id) - vnfd = self.db.get_one("vnfds", {"_id": vnfd_id}) - db_vnfds[vnfd_id] = vnfd - vnfd = db_vnfds[vnfd_id] - if not self._has_terminate_config_primitive(vnfd): - continue - # Get the primitive's sorted sequence list - seq_list = self._get_terminate_config_primitive_seq_list(vnfd) - for seq in seq_list: - # For each sequence in list, get primitive and call _ns_execute_primitive() - step = "Calling terminate action for vnf_member_index={} primitive={}".format( - vnf_index, seq.get("name")) - self.logger.debug(logging_text + step) - # Create the primitive for each sequence, i.e. "primitive": "touch" - primitive = seq.get('name') - mapped_primitive_params = self._get_terminate_primitive_params(seq, vnf_index) - # The following 3 parameters are currently set to None for 'terminate': - # vdu_id, vdu_count_index, vdu_name - vdu_id = db_nslcmop["operationParams"].get("vdu_id") - vdu_count_index = db_nslcmop["operationParams"].get("vdu_count_index") - vdu_name = db_nslcmop["operationParams"].get("vdu_name") - # Add sub-operation - self._add_suboperation(db_nslcmop, - nslcmop_id, - vnf_index, - vdu_id, - vdu_count_index, - vdu_name, - primitive, - mapped_primitive_params) - # Sub-operations: Call _ns_execute_primitive() instead of action() - # db_nsr = self.db.get_one("nsrs", {"_id": nsr_id}) - # nsr_deployed = db_nsr["_admin"]["deployed"] - - # nslcmop_operation_state, nslcmop_operation_state_detail = await self.action( - # nsr_id, nslcmop_terminate_action_id) - # Launch Exception if action() returns other than ['COMPLETED', 'PARTIALLY_COMPLETED'] - # result_ok = ['COMPLETED', 'PARTIALLY_COMPLETED'] - # if result not in result_ok: - # raise LcmException( - # "terminate_primitive_action for vnf_member_index={}", - # " primitive={} fails with error {}".format( - # vnf_index, seq.get("name"), result_detail)) - - # TODO: find ee_id - ee_id = None + return self._retry_or_skip_suboperation(db_nslcmop, op_index) + + # Function to return execution_environment id + + def _get_ee_id(self, vnf_index, vdu_id, vca_deployed_list): + # TODO vdu_index_count + for vca in vca_deployed_list: + if vca["member-vnf-index"] == vnf_index and vca["vdu_id"] == vdu_id: + return vca["ee_id"] + + async def destroy_N2VC(self, logging_text, db_nslcmop, vca_deployed, config_descriptor, + vca_index, destroy_ee=True, exec_primitives=True): + """ + Execute the terminate primitives and destroy the execution environment (if destroy_ee=False + :param logging_text: + :param db_nslcmop: + :param vca_deployed: Dictionary of deployment info at db_nsr._admin.depoloyed.VCA. + :param config_descriptor: Configuration descriptor of the NSD, VNFD, VNFD.vdu or VNFD.kdu + :param vca_index: index in the database _admin.deployed.VCA + :param destroy_ee: False to do not destroy, because it will be destroyed all of then at once + :param exec_primitives: False to do not execute terminate primitives, because the config is not completed or has + not executed properly + :return: None or exception + """ + + self.logger.debug( + logging_text + " vca_index: {}, vca_deployed: {}, config_descriptor: {}, destroy_ee: {}".format( + vca_index, vca_deployed, config_descriptor, destroy_ee + ) + ) + + vca_type = vca_deployed.get("type", "lxc_proxy_charm") + + # execute terminate_primitives + if exec_primitives: + terminate_primitives = self._get_terminate_config_primitive( + config_descriptor.get("terminate-config-primitive"), vca_deployed) + vdu_id = vca_deployed.get("vdu_id") + vdu_count_index = vca_deployed.get("vdu_count_index") + vdu_name = vca_deployed.get("vdu_name") + vnf_index = vca_deployed.get("member-vnf-index") + if terminate_primitives and vca_deployed.get("needed_terminate"): + for seq in terminate_primitives: + # For each sequence in list, get primitive and call _ns_execute_primitive() + step = "Calling terminate action for vnf_member_index={} primitive={}".format( + vnf_index, seq.get("name")) + self.logger.debug(logging_text + step) + # Create the primitive for each sequence, i.e. "primitive": "touch" + primitive = seq.get('name') + mapped_primitive_params = self._get_terminate_primitive_params(seq, vnf_index) + + # Add sub-operation + self._add_suboperation(db_nslcmop, + vnf_index, + vdu_id, + vdu_count_index, + vdu_name, + primitive, + mapped_primitive_params) + # Sub-operations: Call _ns_execute_primitive() instead of action() + try: + result, result_detail = await self._ns_execute_primitive(vca_deployed["ee_id"], primitive, + mapped_primitive_params, + vca_type=vca_type) + except LcmException: + # this happens when VCA is not deployed. In this case it is not needed to terminate + continue + result_ok = ['COMPLETED', 'PARTIALLY_COMPLETED'] + if result not in result_ok: + raise LcmException("terminate_primitive {} for vnf_member_index={} fails with " + "error {}".format(seq.get("name"), vnf_index, result_detail)) + # set that this VCA do not need terminated + db_update_entry = "_admin.deployed.VCA.{}.needed_terminate".format(vca_index) + self.update_db_2("nsrs", db_nslcmop["nsInstanceId"], {db_update_entry: False}) + + if vca_deployed.get("prometheus_jobs") and self.prometheus: + await self.prometheus.update(remove_jobs=vca_deployed["prometheus_jobs"]) + + if destroy_ee: + await self.vca_map[vca_type].delete_execution_environment(vca_deployed["ee_id"]) + + 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): + """ + Terminates a deployment from RO + :param logging_text: + :param nsr_deployed: db_nsr._admin.deployed + :param nsr_id: + :param nslcmop_id: + :param stage: list of string with the content to write on db_nslcmop.detailed-status. + this method will update only the index 2, but it will write on database the concatenated content of the list + :return: + """ + db_nsr_update = {} + failed_detail = [] + ro_nsr_id = ro_delete_action = None + if nsr_deployed and nsr_deployed.get("RO"): + ro_nsr_id = nsr_deployed["RO"].get("nsr_id") + ro_delete_action = nsr_deployed["RO"].get("nsr_delete_action_id") + try: + if ro_nsr_id: + stage[2] = "Deleting ns from VIM." + db_nsr_update["detailed-status"] = " ".join(stage) + self._write_op_status(nslcmop_id, stage) + self.logger.debug(logging_text + stage[2]) + self.update_db_2("nsrs", nsr_id, db_nsr_update) + self._write_op_status(nslcmop_id, stage) + desc = await self.RO.delete("ns", ro_nsr_id) + ro_delete_action = desc["action_id"] + db_nsr_update["_admin.deployed.RO.nsr_delete_action_id"] = ro_delete_action + db_nsr_update["_admin.deployed.RO.nsr_id"] = None + db_nsr_update["_admin.deployed.RO.nsr_status"] = "DELETED" + if ro_delete_action: + # wait until NS is deleted from VIM + stage[2] = "Waiting ns deleted from VIM." + detailed_status_old = None + self.logger.debug(logging_text + stage[2] + " RO_id={} ro_delete_action={}".format(ro_nsr_id, + ro_delete_action)) + self.update_db_2("nsrs", nsr_id, db_nsr_update) + self._write_op_status(nslcmop_id, stage) + + delete_timeout = 20 * 60 # 20 minutes + while delete_timeout > 0: + desc = await self.RO.show( + "ns", + 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) + elif ns_status == "BUILD": + stage[2] = "Deleting from VIM {}".format(ns_status_info) + elif ns_status == "ACTIVE": + db_nsr_update["_admin.deployed.RO.nsr_delete_action_id"] = None + db_nsr_update["_admin.deployed.RO.nsr_status"] = "DELETED" + break + else: + assert False, "ROclient.check_action_status returns unknown {}".format(ns_status) + if stage[2] != detailed_status_old: + detailed_status_old = stage[2] + db_nsr_update["detailed-status"] = " ".join(stage) + self._write_op_status(nslcmop_id, stage) + self.update_db_2("nsrs", nsr_id, db_nsr_update) + await asyncio.sleep(5, loop=self.loop) + delete_timeout -= 5 + else: # delete_timeout <= 0: + raise ROclient.ROClientException("Timeout waiting ns deleted from VIM") + + except Exception as e: + self.update_db_2("nsrs", nsr_id, db_nsr_update) + if isinstance(e, ROclient.ROClientException) and e.http_code == 404: # not found + db_nsr_update["_admin.deployed.RO.nsr_id"] = None + db_nsr_update["_admin.deployed.RO.nsr_status"] = "DELETED" + db_nsr_update["_admin.deployed.RO.nsr_delete_action_id"] = None + self.logger.debug(logging_text + "RO_ns_id={} already deleted".format(ro_nsr_id)) + elif isinstance(e, ROclient.ROClientException) and e.http_code == 409: # conflict + failed_detail.append("delete conflict: {}".format(e)) + self.logger.debug(logging_text + "RO_ns_id={} delete conflict: {}".format(ro_nsr_id, e)) + else: + failed_detail.append("delete error: {}".format(e)) + self.logger.error(logging_text + "RO_ns_id={} delete error: {}".format(ro_nsr_id, e)) + + # Delete nsd + if not failed_detail and deep_get(nsr_deployed, ("RO", "nsd_id")): + ro_nsd_id = nsr_deployed["RO"]["nsd_id"] + try: + stage[2] = "Deleting nsd from RO." + db_nsr_update["detailed-status"] = " ".join(stage) + self.update_db_2("nsrs", nsr_id, db_nsr_update) + self._write_op_status(nslcmop_id, stage) + await self.RO.delete("nsd", ro_nsd_id) + self.logger.debug(logging_text + "ro_nsd_id={} deleted".format(ro_nsd_id)) + db_nsr_update["_admin.deployed.RO.nsd_id"] = None + except Exception as e: + if isinstance(e, ROclient.ROClientException) and e.http_code == 404: # not found + db_nsr_update["_admin.deployed.RO.nsd_id"] = None + self.logger.debug(logging_text + "ro_nsd_id={} already deleted".format(ro_nsd_id)) + elif isinstance(e, ROclient.ROClientException) and e.http_code == 409: # conflict + failed_detail.append("ro_nsd_id={} delete conflict: {}".format(ro_nsd_id, e)) + self.logger.debug(logging_text + failed_detail[-1]) + else: + failed_detail.append("ro_nsd_id={} delete error: {}".format(ro_nsd_id, e)) + self.logger.error(logging_text + failed_detail[-1]) + + if not failed_detail and deep_get(nsr_deployed, ("RO", "vnfd")): + for index, vnf_deployed in enumerate(nsr_deployed["RO"]["vnfd"]): + if not vnf_deployed or not vnf_deployed["id"]: + continue try: - await self.n2vc.exec_primitive( - ee_id=ee_id, - primitive_name=primitive, - params_dict=mapped_primitive_params - ) + ro_vnfd_id = vnf_deployed["id"] + stage[2] = "Deleting member_vnf_index={} ro_vnfd_id={} from RO.".format( + vnf_deployed["member-vnf-index"], ro_vnfd_id) + db_nsr_update["detailed-status"] = " ".join(stage) + self.update_db_2("nsrs", nsr_id, db_nsr_update) + self._write_op_status(nslcmop_id, stage) + await self.RO.delete("vnfd", ro_vnfd_id) + self.logger.debug(logging_text + "ro_vnfd_id={} deleted".format(ro_vnfd_id)) + db_nsr_update["_admin.deployed.RO.vnfd.{}.id".format(index)] = None except Exception as e: - self.logger.error('Error executing primitive {}: {}'.format(primitive, e)) - raise LcmException( - "terminate_primitive_action for vnf_member_index={}, primitive={} fails with error {}" - .format(vnf_index, seq.get("name"), e), - ) + if isinstance(e, ROclient.ROClientException) and e.http_code == 404: # not found + db_nsr_update["_admin.deployed.RO.vnfd.{}.id".format(index)] = None + self.logger.debug(logging_text + "ro_vnfd_id={} already deleted ".format(ro_vnfd_id)) + elif isinstance(e, ROclient.ROClientException) and e.http_code == 409: # conflict + failed_detail.append("ro_vnfd_id={} delete conflict: {}".format(ro_vnfd_id, e)) + self.logger.debug(logging_text + failed_detail[-1]) + else: + failed_detail.append("ro_vnfd_id={} delete error: {}".format(ro_vnfd_id, e)) + self.logger.error(logging_text + failed_detail[-1]) - async def terminate(self, nsr_id, nslcmop_id): + if failed_detail: + stage[2] = "Error deleting from VIM" + else: + stage[2] = "Deleted from VIM" + db_nsr_update["detailed-status"] = " ".join(stage) + self.update_db_2("nsrs", nsr_id, db_nsr_update) + self._write_op_status(nslcmop_id, stage) + + if failed_detail: + raise LcmException("; ".join(failed_detail)) + async def terminate(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) if not task_is_locked_by_me: @@ -1945,237 +3281,226 @@ class NsLcm(LcmBase): logging_text = "Task ns={} terminate={} ".format(nsr_id, nslcmop_id) self.logger.debug(logging_text + "Enter") + timeout_ns_terminate = self.timeout_ns_terminate db_nsr = None db_nslcmop = None + operation_params = None exc = None - failed_detail = [] # annotates all failed error messages - db_nsr_update = {"_admin.nslcmop": nslcmop_id, - "_admin.current-operation": nslcmop_id, - "_admin.operation-type": "terminate"} - self.update_db_2("nsrs", nsr_id, db_nsr_update) + error_list = [] # annotates all failed error messages db_nslcmop_update = {} - nslcmop_operation_state = None autoremove = False # autoremove after terminated - pending_tasks = [] + tasks_dict_info = {} + db_nsr_update = {} + stage = ["Stage 1/3: Preparing task.", "Waiting for previous operations to terminate.", ""] + # ^ contains [stage, step, VIM-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) - step = "Getting nslcmop={} from db".format(nslcmop_id) + stage[1] = "Getting nslcmop={} from db.".format(nslcmop_id) db_nslcmop = self.db.get_one("nslcmops", {"_id": nslcmop_id}) - step = "Getting nsr={} from db".format(nsr_id) + operation_params = db_nslcmop.get("operationParams") or {} + if operation_params.get("timeout_ns_terminate"): + timeout_ns_terminate = operation_params["timeout_ns_terminate"] + stage[1] = "Getting nsr={} from db.".format(nsr_id) db_nsr = self.db.get_one("nsrs", {"_id": nsr_id}) - # nsd = db_nsr["nsd"] - nsr_deployed = deepcopy(db_nsr["_admin"].get("deployed")) - if db_nsr["_admin"]["nsState"] == "NOT_INSTANTIATED": - return - # #TODO check if VIM is creating and wait - # RO_vim_id = db_vim["_admin"]["deployed"]["RO"] - # Call internal terminate action - await self._terminate_action(db_nslcmop, nslcmop_id, nsr_id) - - pending_tasks = [] db_nsr_update["operational-status"] = "terminating" db_nsr_update["config-status"] = "terminating" + self._write_ns_status( + nsr_id=nsr_id, + ns_state="TERMINATING", + current_operation="TERMINATING", + current_operation_id=nslcmop_id, + other_update=db_nsr_update + ) + self._write_op_status( + op_id=nslcmop_id, + queuePosition=0, + stage=stage + ) + nsr_deployed = deepcopy(db_nsr["_admin"].get("deployed")) or {} + if db_nsr["_admin"]["nsState"] == "NOT_INSTANTIATED": + return - # remove NS - try: - step = "delete execution environment" - self.logger.debug(logging_text + step) - - 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) - self.logger.error(msg) - failed_detail.append(msg) - - try: - # Delete from k8scluster - step = "delete kdus" - self.logger.debug(logging_text + step) - # print(nsr_deployed) - if nsr_deployed: - for kdu in nsr_deployed.get("K8s", ()): - kdu_instance = kdu.get("kdu-instance") - if not kdu_instance: - continue - if kdu.get("k8scluster-type") == "chart": - task_delete_kdu_instance = asyncio.ensure_future( - self.k8sclusterhelm.uninstall(cluster_uuid=kdu.get("k8scluster-uuid"), - kdu_instance=kdu_instance)) - elif kdu.get("k8scluster-type") == "juju": - task_delete_kdu_instance = asyncio.ensure_future( - self.k8sclusterjuju.uninstall(cluster_uuid=kdu.get("k8scluster-uuid"), - kdu_instance=kdu_instance)) - else: - self.error(logging_text + "Unknown k8s deployment type {}". - format(kdu.get("k8scluster-type"))) - continue - pending_tasks.append(task_delete_kdu_instance) - except LcmException as e: - msg = "Failed while deleting KDUs from NS: {}".format(e) - self.logger.error(msg) - failed_detail.append(msg) - - # remove from RO - RO_fail = False - - # Delete ns - RO_nsr_id = RO_delete_action = None - if nsr_deployed and nsr_deployed.get("RO"): - RO_nsr_id = nsr_deployed["RO"].get("nsr_id") - RO_delete_action = nsr_deployed["RO"].get("nsr_delete_action_id") - try: - if RO_nsr_id: - step = db_nsr_update["detailed-status"] = db_nslcmop_update["detailed-status"] = \ - "Deleting ns from VIM" - self.update_db_2("nslcmops", nslcmop_id, db_nslcmop_update) - self.update_db_2("nsrs", nsr_id, db_nsr_update) - self.logger.debug(logging_text + step) - desc = await self.RO.delete("ns", RO_nsr_id) - RO_delete_action = desc["action_id"] - db_nsr_update["_admin.deployed.RO.nsr_delete_action_id"] = RO_delete_action - db_nsr_update["_admin.deployed.RO.nsr_id"] = None - db_nsr_update["_admin.deployed.RO.nsr_status"] = "DELETED" - if RO_delete_action: - # wait until NS is deleted from VIM - step = detailed_status = "Waiting ns deleted from VIM. RO_id={} RO_delete_action={}".\ - format(RO_nsr_id, RO_delete_action) - detailed_status_old = None - self.logger.debug(logging_text + step) - - delete_timeout = 20 * 60 # 20 minutes - while delete_timeout > 0: - desc = await self.RO.show( - "ns", - item_id_name=RO_nsr_id, - extra_item="action", - extra_item_id=RO_delete_action) - ns_status, ns_status_info = self.RO.check_action_status(desc) - if ns_status == "ERROR": - raise ROclient.ROClientException(ns_status_info) - elif ns_status == "BUILD": - detailed_status = step + "; {}".format(ns_status_info) - elif ns_status == "ACTIVE": - db_nsr_update["_admin.deployed.RO.nsr_delete_action_id"] = None - db_nsr_update["_admin.deployed.RO.nsr_status"] = "DELETED" - break - else: - assert False, "ROclient.check_action_status returns unknown {}".format(ns_status) - if detailed_status != detailed_status_old: - detailed_status_old = db_nslcmop_update["detailed-status"] = \ - db_nsr_update["detailed-status"] = detailed_status - self.update_db_2("nslcmops", nslcmop_id, db_nslcmop_update) - self.update_db_2("nsrs", nsr_id, db_nsr_update) - await asyncio.sleep(5, loop=self.loop) - delete_timeout -= 5 - else: # delete_timeout <= 0: - raise ROclient.ROClientException("Timeout waiting ns deleted from VIM") - - except ROclient.ROClientException as e: - if e.http_code == 404: # not found - db_nsr_update["_admin.deployed.RO.nsr_id"] = None - db_nsr_update["_admin.deployed.RO.nsr_status"] = "DELETED" - db_nsr_update["_admin.deployed.RO.nsr_delete_action_id"] = None - self.logger.debug(logging_text + "RO_ns_id={} already deleted".format(RO_nsr_id)) - elif e.http_code == 409: # conflict - failed_detail.append("RO_ns_id={} delete conflict: {}".format(RO_nsr_id, e)) - self.logger.debug(logging_text + failed_detail[-1]) - RO_fail = True + stage[1] = "Getting vnf descriptors from db." + db_vnfrs_list = self.db.get_list("vnfrs", {"nsr-id-ref": nsr_id}) + db_vnfds_from_id = {} + db_vnfds_from_member_index = {} + # Loop over VNFRs + for vnfr in db_vnfrs_list: + vnfd_id = vnfr["vnfd-id"] + if vnfd_id not in db_vnfds_from_id: + vnfd = self.db.get_one("vnfds", {"_id": vnfd_id}) + db_vnfds_from_id[vnfd_id] = vnfd + db_vnfds_from_member_index[vnfr["member-vnf-index-ref"]] = db_vnfds_from_id[vnfd_id] + + # Destroy individual execution environments when there are terminating primitives. + # Rest of EE will be deleted at once + # TODO - check before calling _destroy_N2VC + # if not operation_params.get("skip_terminate_primitives"):# + # or not vca.get("needed_terminate"): + stage[0] = "Stage 2/3 execute terminating primitives." + self.logger.debug(logging_text + stage[0]) + stage[1] = "Looking execution environment that needs terminate." + self.logger.debug(logging_text + stage[1]) + # self.logger.debug("nsr_deployed: {}".format(nsr_deployed)) + for vca_index, vca in enumerate(get_iterable(nsr_deployed, "VCA")): + config_descriptor = None + if not vca or not vca.get("ee_id"): + continue + if not vca.get("member-vnf-index"): + # ns + config_descriptor = db_nsr.get("ns-configuration") + elif vca.get("vdu_id"): + db_vnfd = db_vnfds_from_member_index[vca["member-vnf-index"]] + vdud = next((vdu for vdu in db_vnfd.get("vdu", ()) if vdu["id"] == vca.get("vdu_id")), None) + if vdud: + config_descriptor = vdud.get("vdu-configuration") + elif vca.get("kdu_name"): + db_vnfd = db_vnfds_from_member_index[vca["member-vnf-index"]] + kdud = next((kdu for kdu in db_vnfd.get("kdu", ()) if kdu["name"] == vca.get("kdu_name")), None) + if kdud: + config_descriptor = kdud.get("kdu-configuration") else: - failed_detail.append("RO_ns_id={} delete error: {}".format(RO_nsr_id, e)) - self.logger.error(logging_text + failed_detail[-1]) - RO_fail = True - - # Delete nsd - if not RO_fail and nsr_deployed and nsr_deployed.get("RO") and nsr_deployed["RO"].get("nsd_id"): - RO_nsd_id = nsr_deployed["RO"]["nsd_id"] - try: - step = db_nsr_update["detailed-status"] = db_nslcmop_update["detailed-status"] =\ - "Deleting nsd from RO" - await self.RO.delete("nsd", RO_nsd_id) - self.logger.debug(logging_text + "RO_nsd_id={} deleted".format(RO_nsd_id)) - db_nsr_update["_admin.deployed.RO.nsd_id"] = None - except ROclient.ROClientException as e: - if e.http_code == 404: # not found - db_nsr_update["_admin.deployed.RO.nsd_id"] = None - self.logger.debug(logging_text + "RO_nsd_id={} already deleted".format(RO_nsd_id)) - elif e.http_code == 409: # conflict - failed_detail.append("RO_nsd_id={} delete conflict: {}".format(RO_nsd_id, e)) - self.logger.debug(logging_text + failed_detail[-1]) - RO_fail = True - else: - failed_detail.append("RO_nsd_id={} delete error: {}".format(RO_nsd_id, e)) - self.logger.error(logging_text + failed_detail[-1]) - RO_fail = True - - if not RO_fail and nsr_deployed and nsr_deployed.get("RO") and nsr_deployed["RO"].get("vnfd"): - for index, vnf_deployed in enumerate(nsr_deployed["RO"]["vnfd"]): - if not vnf_deployed or not vnf_deployed["id"]: - continue - try: - RO_vnfd_id = vnf_deployed["id"] - step = db_nsr_update["detailed-status"] = db_nslcmop_update["detailed-status"] =\ - "Deleting member_vnf_index={} RO_vnfd_id={} from RO".format( - vnf_deployed["member-vnf-index"], RO_vnfd_id) - await self.RO.delete("vnfd", RO_vnfd_id) - self.logger.debug(logging_text + "RO_vnfd_id={} deleted".format(RO_vnfd_id)) - db_nsr_update["_admin.deployed.RO.vnfd.{}.id".format(index)] = None - except ROclient.ROClientException as e: - if e.http_code == 404: # not found - db_nsr_update["_admin.deployed.RO.vnfd.{}.id".format(index)] = None - self.logger.debug(logging_text + "RO_vnfd_id={} already deleted ".format(RO_vnfd_id)) - elif e.http_code == 409: # conflict - failed_detail.append("RO_vnfd_id={} delete conflict: {}".format(RO_vnfd_id, e)) - self.logger.debug(logging_text + failed_detail[-1]) - else: - failed_detail.append("RO_vnfd_id={} delete error: {}".format(RO_vnfd_id, e)) - self.logger.error(logging_text + failed_detail[-1]) + config_descriptor = db_vnfds_from_member_index[vca["member-vnf-index"]].get("vnf-configuration") + vca_type = vca.get("type") + exec_terminate_primitives = (not operation_params.get("skip_terminate_primitives") 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 + # 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( + self.destroy_N2VC(logging_text, db_nslcmop, vca, config_descriptor, vca_index, + destroy_ee, exec_terminate_primitives)) + tasks_dict_info[task] = "Terminating VCA {}".format(vca.get("ee_id")) + + # wait for pending tasks of terminate primitives + if tasks_dict_info: + self.logger.debug(logging_text + 'Waiting for tasks {}'.format(list(tasks_dict_info.keys()))) + error_list = await self._wait_for_tasks(logging_text, tasks_dict_info, + min(self.timeout_charm_delete, timeout_ns_terminate), + stage, nslcmop_id) + tasks_dict_info.clear() + if error_list: + return # raise LcmException("; ".join(error_list)) + + # remove All execution environments at once + stage[0] = "Stage 3/3 delete all." + + 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." + self.logger.debug(logging_text + stage[1]) + # print(nsr_deployed) + for kdu in get_iterable(nsr_deployed, "K8s"): + if not kdu or not kdu.get("kdu-instance"): + continue + kdu_instance = kdu.get("kdu-instance") + if kdu.get("k8scluster-type") in self.k8scluster_map: + task_delete_kdu_instance = asyncio.ensure_future( + self.k8scluster_map[kdu["k8scluster-type"]].uninstall( + cluster_uuid=kdu.get("k8scluster-uuid"), + kdu_instance=kdu_instance)) + else: + self.logger.error(logging_text + "Unknown k8s deployment type {}". + format(kdu.get("k8scluster-type"))) + continue + tasks_dict_info[task_delete_kdu_instance] = "Terminating KDU '{}'".format(kdu.get("kdu-name")) - if failed_detail: - self.logger.error(logging_text + " ;".join(failed_detail)) - db_nsr_update["operational-status"] = "failed" - db_nsr_update["detailed-status"] = "Deletion errors " + "; ".join(failed_detail) - db_nslcmop_update["detailed-status"] = "; ".join(failed_detail) - db_nslcmop_update["operationState"] = nslcmop_operation_state = "FAILED" - db_nslcmop_update["statusEnteredTime"] = time() + # remove from RO + stage[1] = "Deleting ns from VIM." + if self.ng_ro: + task_delete_ro = asyncio.ensure_future( + self._terminate_ng_ro(logging_text, nsr_deployed, nsr_id, nslcmop_id, stage)) else: - db_nsr_update["operational-status"] = "terminated" - db_nsr_update["detailed-status"] = "Done" - db_nsr_update["_admin.nsState"] = "NOT_INSTANTIATED" - db_nslcmop_update["detailed-status"] = "Done" - db_nslcmop_update["operationState"] = nslcmop_operation_state = "COMPLETED" - db_nslcmop_update["statusEnteredTime"] = time() - if db_nslcmop["operationParams"].get("autoremove"): - autoremove = True + task_delete_ro = asyncio.ensure_future( + self._terminate_RO(logging_text, nsr_deployed, nsr_id, nslcmop_id, stage)) + tasks_dict_info[task_delete_ro] = "Removing deployment from VIM" - except (ROclient.ROClientException, DbException, LcmException) as e: + # rest of staff will be done at finally + + except (ROclient.ROClientException, DbException, LcmException, N2VCException) 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)) + self.logger.error(logging_text + "Cancelled Exception while '{}'".format(stage[1])) exc = "Operation was cancelled" except Exception as e: exc = traceback.format_exc() - self.logger.critical(logging_text + "Exit Exception {}".format(e), exc_info=True) + self.logger.critical(logging_text + "Exit Exception while '{}': {}".format(stage[1], e), exc_info=True) finally: - if exc and db_nslcmop: - db_nslcmop_update["detailed-status"] = "FAILED {}: {}".format(step, exc) - db_nslcmop_update["operationState"] = nslcmop_operation_state = "FAILED" - db_nslcmop_update["statusEnteredTime"] = time() + if exc: + error_list.append(str(exc)) try: - if db_nslcmop and db_nslcmop_update: - self.update_db_2("nslcmops", nslcmop_id, db_nslcmop_update) - if db_nsr: - db_nsr_update["_admin.nslcmop"] = None - db_nsr_update["_admin.current-operation"] = None - db_nsr_update["_admin.operation-type"] = None - self.update_db_2("nsrs", nsr_id, db_nsr_update) - except DbException as e: - self.logger.error(logging_text + "Cannot update database: {}".format(e)) + # wait for pending tasks + if tasks_dict_info: + stage[1] = "Waiting for terminate pending tasks." + self.logger.debug(logging_text + stage[1]) + error_list += await self._wait_for_tasks(logging_text, tasks_dict_info, timeout_ns_terminate, + stage, nslcmop_id) + stage[1] = stage[2] = "" + except asyncio.CancelledError: + error_list.append("Cancelled") + # TODO cancell all tasks + except Exception as exc: + error_list.append(str(exc)) + # update status at database + if error_list: + error_detail = "; ".join(error_list) + # self.logger.error(logging_text + error_detail) + error_description_nslcmop = '{} Detail: {}'.format(stage[0], error_detail) + error_description_nsr = 'Operation: TERMINATING.{}, {}.'.format(nslcmop_id, stage[0]) + + db_nsr_update["operational-status"] = "failed" + db_nsr_update["detailed-status"] = error_description_nsr + " Detail: " + error_detail + db_nslcmop_update["detailed-status"] = error_detail + nslcmop_operation_state = "FAILED" + ns_state = "BROKEN" + else: + error_detail = None + error_description_nsr = error_description_nslcmop = None + ns_state = "NOT_INSTANTIATED" + db_nsr_update["operational-status"] = "terminated" + db_nsr_update["detailed-status"] = "Done" + db_nsr_update["_admin.nsState"] = "NOT_INSTANTIATED" + db_nslcmop_update["detailed-status"] = "Done" + nslcmop_operation_state = "COMPLETED" + + if db_nsr: + self._write_ns_status( + nsr_id=nsr_id, + ns_state=ns_state, + current_operation="IDLE", + current_operation_id=None, + error_description=error_description_nsr, + error_detail=error_detail, + 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 ns_state == "NOT_INSTANTIATED": + try: + self.db.set_list("vnfrs", {"nsr-id-ref": nsr_id}, {"_admin.nsState": "NOT_INSTANTIATED"}) + except DbException as e: + self.logger.warn(logging_text + 'Error writing VNFR status for nsr-id-ref: {} -> {}'. + format(nsr_id, e)) + if operation_params: + autoremove = operation_params.get("autoremove", False) if nslcmop_operation_state: try: await self.msg.aiowrite("ns", "terminated", {"nsr_id": nsr_id, "nslcmop_id": nslcmop_id, @@ -2185,20 +3510,58 @@ class NsLcm(LcmBase): except Exception as e: self.logger.error(logging_text + "kafka_write notification Exception {}".format(e)) - # wait for pending tasks - done = None - pending = None - if pending_tasks: - self.logger.debug(logging_text + 'Waiting for terminate pending tasks...') - done, pending = await asyncio.wait(pending_tasks, timeout=3600) - if not pending: - self.logger.debug(logging_text + 'All tasks finished...') - else: - self.logger.info(logging_text + 'There are pending tasks: {}'.format(pending)) - self.logger.debug(logging_text + "Exit") self.lcm_tasks.remove("ns", nsr_id, nslcmop_id, "ns_terminate") + async def _wait_for_tasks(self, logging_text, created_tasks_info, timeout, stage, nslcmop_id, nsr_id=None): + time_start = time() + error_detail_list = [] + error_list = [] + pending_tasks = list(created_tasks_info.keys()) + num_tasks = len(pending_tasks) + num_done = 0 + stage[1] = "{}/{}.".format(num_done, num_tasks) + self._write_op_status(nslcmop_id, stage) + while pending_tasks: + new_error = None + _timeout = timeout + time_start - time() + done, pending_tasks = await asyncio.wait(pending_tasks, timeout=_timeout, + return_when=asyncio.FIRST_COMPLETED) + num_done += len(done) + if not done: # Timeout + for task in pending_tasks: + new_error = created_tasks_info[task] + ": Timeout" + error_detail_list.append(new_error) + error_list.append(new_error) + break + for task in done: + if task.cancelled(): + exc = "Cancelled" + else: + exc = task.exception() + if exc: + if isinstance(exc, asyncio.TimeoutError): + exc = "Timeout" + 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, + K8sException)): + self.logger.error(logging_text + new_error) + else: + exc_traceback = "".join(traceback.format_exception(None, exc, exc.__traceback__)) + self.logger.error(logging_text + created_tasks_info[task] + exc_traceback) + else: + self.logger.debug(logging_text + created_tasks_info[task] + ": Done") + stage[1] = "{}/{}.".format(num_done, num_tasks) + if new_error: + stage[1] += " Errors: " + ". ".join(error_detail_list) + "." + if nsr_id: # update also nsr + self.update_db_2("nsrs", nsr_id, {"errorDescription": "Error at: " + ", ".join(error_list), + "errorDetail": ". ".join(error_detail_list)}) + self._write_op_status(nslcmop_id, stage) + return error_detail_list + @staticmethod def _map_primitive_params(primitive_desc, params, instantiation_params): """ @@ -2235,6 +3598,14 @@ class NsLcm(LcmBase): width=256) elif isinstance(calculated_params[param_name], str) and calculated_params[param_name].startswith("!!yaml "): calculated_params[param_name] = calculated_params[param_name][7:] + if parameter.get("data-type") == "INTEGER": + try: + calculated_params[param_name] = int(calculated_params[param_name]) + except ValueError: # error converting string to int + raise LcmException( + "Parameter {} of primitive {} must be integer".format(param_name, primitive_desc["name"])) + elif parameter.get("data-type") == "BOOLEAN": + calculated_params[param_name] = not ((str(calculated_params[param_name])).lower() == 'false') # add always ns_config_info if primitive name is config if primitive_desc["name"] == "config": @@ -2242,58 +3613,77 @@ class NsLcm(LcmBase): calculated_params["ns_config_info"] = instantiation_params["ns_config_info"] return calculated_params - async def _ns_execute_primitive(self, db_deployed, member_vnf_index, vdu_id, vdu_name, vdu_count_index, - primitive, primitive_params, retries=0, retries_interval=30) -> (str, str): - - # find vca_deployed record for this action + def _look_for_deployed_vca(self, deployed_vca, member_vnf_index, vdu_id, vdu_count_index, kdu_name=None, + ee_descriptor_id=None): + # find vca_deployed record for this action. Raise LcmException if not found or there is not any id. + for vca in deployed_vca: + if not vca: + continue + if member_vnf_index != vca["member-vnf-index"] or vdu_id != vca["vdu_id"]: + continue + if vdu_count_index is not None and vdu_count_index != vca["vdu_count_index"]: + continue + if kdu_name and kdu_name != vca["kdu_name"]: + continue + if ee_descriptor_id and ee_descriptor_id != vca["ee_descriptor_id"]: + continue + break + else: + # vca_deployed not found + raise LcmException("charm for member_vnf_index={} vdu_id={}.{} kdu_name={} execution-environment-list.id={}" + " is not deployed".format(member_vnf_index, vdu_id, vdu_count_index, kdu_name, + ee_descriptor_id)) + + # get ee_id + ee_id = vca.get("ee_id") + vca_type = vca.get("type", "lxc_proxy_charm") # default value for backward compatibility - proxy charm + if not ee_id: + raise LcmException("charm for member_vnf_index={} vdu_id={} kdu_name={} vdu_count_index={} has not " + "execution environment" + .format(member_vnf_index, vdu_id, kdu_name, vdu_count_index)) + return ee_id, vca_type + + async def _ns_execute_primitive(self, ee_id, primitive, primitive_params, retries=0, + retries_interval=30, timeout=None, + vca_type=None, db_dict=None) -> (str, str): try: - for vca_deployed in db_deployed["VCA"]: - if not vca_deployed: - continue - if member_vnf_index != vca_deployed["member-vnf-index"] or vdu_id != vca_deployed["vdu_id"]: - continue - if vdu_name and vdu_name != vca_deployed["vdu_name"]: - continue - if vdu_count_index and vdu_count_index != vca_deployed["vdu_count_index"]: - continue - break - else: - # vca_deployed not found - raise LcmException("charm for member_vnf_index={} vdu_id={} vdu_name={} vdu_count_index={} is not " - "deployed".format(member_vnf_index, vdu_id, vdu_name, vdu_count_index)) - - # get ee_id - ee_id = vca_deployed.get("ee_id") - if not ee_id: - raise LcmException("charm for member_vnf_index={} vdu_id={} vdu_name={} vdu_count_index={} has not " - "execution environment" - .format(member_vnf_index, vdu_id, vdu_name, vdu_count_index)) - if primitive == "config": primitive_params = {"params": primitive_params} + vca_type = vca_type or "lxc_proxy_charm" + while retries >= 0: try: - output = await self.n2vc.exec_primitive( - ee_id=ee_id, - primitive_name=primitive, - params_dict=primitive_params - ) + output = await asyncio.wait_for( + self.vca_map[vca_type].exec_primitive( + ee_id=ee_id, + primitive_name=primitive, + params_dict=primitive_params, + progress_timeout=self.timeout_progress_primitive, + total_timeout=self.timeout_primitive, + db_dict=db_dict), + timeout=timeout or self.timeout_primitive) # execution was OK break - except Exception as e: + except asyncio.CancelledError: + raise + except Exception as e: # asyncio.TimeoutError + if isinstance(e, asyncio.TimeoutError): + e = "Timeout" retries -= 1 if retries >= 0: self.logger.debug('Error executing action {} on {} -> {}'.format(primitive, ee_id, e)) # wait and retry await asyncio.sleep(retries_interval, loop=self.loop) else: - return 'Cannot execute action {} on {}: {}'.format(primitive, ee_id, e), 'FAIL' + return 'FAILED', str(e) - return output, 'OK' + return 'COMPLETED', output + except (LcmException, asyncio.CancelledError): + raise except Exception as e: - return 'Error executing action {}: {}'.format(primitive, e), 'FAIL' + return 'FAIL', 'Error executing action {}: {}'.format(primitive, e) async def action(self, nsr_id, nslcmop_id): @@ -2307,19 +3697,23 @@ class NsLcm(LcmBase): # get all needed from database db_nsr = None db_nslcmop = None - db_nsr_update = {"_admin.nslcmop": nslcmop_id, - "_admin.current-operation": nslcmop_id, - "_admin.operation-type": "action"} - self.update_db_2("nsrs", nsr_id, db_nsr_update) + db_nsr_update = {} db_nslcmop_update = {} nslcmop_operation_state = None - nslcmop_operation_state_detail = None + error_description_nslcmop = None exc = None 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="RUNNING ACTION", + current_operation_id=nslcmop_id + ) + step = "Getting information from database" db_nslcmop = self.db.get_one("nslcmops", {"_id": nslcmop_id}) db_nsr = self.db.get_one("nsrs", {"_id": nsr_id}) @@ -2329,7 +3723,9 @@ class NsLcm(LcmBase): vdu_id = db_nslcmop["operationParams"].get("vdu_id") kdu_name = db_nslcmop["operationParams"].get("kdu_name") vdu_count_index = db_nslcmop["operationParams"].get("vdu_count_index") - vdu_name = db_nslcmop["operationParams"].get("vdu_name") + primitive = db_nslcmop["operationParams"]["primitive"] + primitive_params = db_nslcmop["operationParams"]["primitive_params"] + timeout_ns_action = db_nslcmop["operationParams"].get("timeout_ns_action", self.timeout_primitive) if vnf_index: step = "Getting vnfr from database" @@ -2337,11 +3733,8 @@ class NsLcm(LcmBase): step = "Getting vnfd from database" db_vnfd = self.db.get_one("vnfds", {"_id": db_vnfr["vnfd-id"]}) else: - if db_nsr.get("nsd"): - db_nsd = db_nsr.get("nsd") # TODO this will be removed - else: - step = "Getting nsd from database" - db_nsd = self.db.get_one("nsds", {"_id": db_nsr["nsd-id"]}) + step = "Getting nsd from database" + db_nsd = self.db.get_one("nsds", {"_id": db_nsr["nsd-id"]}) # for backward compatibility if nsr_deployed and isinstance(nsr_deployed.get("VCA"), dict): @@ -2349,172 +3742,184 @@ class NsLcm(LcmBase): db_nsr_update["_admin.deployed.VCA"] = nsr_deployed["VCA"] self.update_db_2("nsrs", nsr_id, db_nsr_update) - primitive = db_nslcmop["operationParams"]["primitive"] - primitive_params = db_nslcmop["operationParams"]["primitive_params"] - # look for primitive - config_primitive_desc = None + config_primitive_desc = descriptor_configuration = None if vdu_id: for vdu in get_iterable(db_vnfd, "vdu"): if vdu_id == vdu["id"]: - for config_primitive in vdu.get("vdu-configuration", {}).get("config-primitive", ()): - if config_primitive["name"] == primitive: - config_primitive_desc = config_primitive - break + descriptor_configuration = vdu.get("vdu-configuration") + break elif kdu_name: - self.logger.debug(logging_text + "Checking actions in KDUs") - kdur = next((x for x in db_vnfr["kdur"] if x["kdu_name"] == kdu_name), None) - desc_params = self._format_additional_params(kdur.get("additionalParams")) or {} - if primitive_params: - desc_params.update(primitive_params) - # TODO Check if we will need something at vnf level - index = 0 - for kdu in get_iterable(nsr_deployed, "K8s"): - if kdu_name == kdu["kdu-name"]: - db_dict = {"collection": "nsrs", "filter": {"_id": nsr_id}, - "path": "_admin.deployed.K8s.{}".format(index)} - if primitive == "upgrade": - if desc_params.get("kdu_model"): - kdu_model = desc_params.get("kdu_model") - del desc_params["kdu_model"] - else: - kdu_model = kdu.get("kdu-model") - parts = kdu_model.split(sep=":") - if len(parts) == 2: - kdu_model = parts[0] - - if kdu.get("k8scluster-type") == "chart": - output = await self.k8sclusterhelm.upgrade(cluster_uuid=kdu.get("k8scluster-uuid"), - kdu_instance=kdu.get("kdu-instance"), - atomic=True, kdu_model=kdu_model, - params=desc_params, db_dict=db_dict, - timeout=300) - elif kdu.get("k8scluster-type") == "juju": - output = await self.k8sclusterjuju.upgrade(cluster_uuid=kdu.get("k8scluster-uuid"), - kdu_instance=kdu.get("kdu-instance"), - atomic=True, kdu_model=kdu_model, - params=desc_params, db_dict=db_dict, - timeout=300) - - else: - msg = "k8scluster-type not defined" - raise LcmException(msg) - - self.logger.debug(logging_text + " Upgrade of kdu {} done".format(output)) - break - elif primitive == "rollback": - if kdu.get("k8scluster-type") == "chart": - output = await self.k8sclusterhelm.rollback(cluster_uuid=kdu.get("k8scluster-uuid"), - kdu_instance=kdu.get("kdu-instance"), - db_dict=db_dict) - elif kdu.get("k8scluster-type") == "juju": - output = await self.k8sclusterjuju.rollback(cluster_uuid=kdu.get("k8scluster-uuid"), - kdu_instance=kdu.get("kdu-instance"), - db_dict=db_dict) - else: - msg = "k8scluster-type not defined" - raise LcmException(msg) - break - elif primitive == "status": - if kdu.get("k8scluster-type") == "chart": - output = await self.k8sclusterhelm.status_kdu(cluster_uuid=kdu.get("k8scluster-uuid"), - kdu_instance=kdu.get("kdu-instance")) - elif kdu.get("k8scluster-type") == "juju": - output = await self.k8sclusterjuju.status_kdu(cluster_uuid=kdu.get("k8scluster-uuid"), - kdu_instance=kdu.get("kdu-instance")) - else: - msg = "k8scluster-type not defined" - raise LcmException(msg) - break - index += 1 - - else: - raise LcmException("KDU '{}' not found".format(kdu_name)) - if output: - db_nslcmop_update["detailed-status"] = output - db_nslcmop_update["operationState"] = 'COMPLETED' - db_nslcmop_update["statusEnteredTime"] = time() - else: - db_nslcmop_update["detailed-status"] = '' - db_nslcmop_update["operationState"] = 'FAILED' - db_nslcmop_update["statusEnteredTime"] = time() - return - elif vnf_index: - for config_primitive in db_vnfd.get("vnf-configuration", {}).get("config-primitive", ()): - if config_primitive["name"] == primitive: - config_primitive_desc = config_primitive + for kdu in get_iterable(db_vnfd, "kdu"): + if kdu_name == kdu["name"]: + descriptor_configuration = kdu.get("kdu-configuration") break + elif vnf_index: + descriptor_configuration = db_vnfd.get("vnf-configuration") else: - for config_primitive in db_nsd.get("ns-configuration", {}).get("config-primitive", ()): + descriptor_configuration = db_nsd.get("ns-configuration") + + if descriptor_configuration and descriptor_configuration.get("config-primitive"): + for config_primitive in descriptor_configuration["config-primitive"]: if config_primitive["name"] == primitive: config_primitive_desc = config_primitive break if not config_primitive_desc: - raise LcmException("Primitive {} not found at [ns|vnf|vdu]-configuration:config-primitive ". - format(primitive)) + if not (kdu_name and primitive in ("upgrade", "rollback", "status")): + raise LcmException("Primitive {} not found at [ns|vnf|vdu]-configuration:config-primitive ". + format(primitive)) + primitive_name = primitive + ee_descriptor_id = None + else: + primitive_name = config_primitive_desc.get("execution-environment-primitive", primitive) + ee_descriptor_id = config_primitive_desc.get("execution-environment-ref") - desc_params = {} if vnf_index: - if db_vnfr.get("additionalParamsForVnf"): - desc_params = self._format_additional_params(db_vnfr["additionalParamsForVnf"]) if vdu_id: vdur = next((x for x in db_vnfr["vdur"] if x["vdu-id-ref"] == vdu_id), None) - if vdur.get("additionalParams"): - desc_params = self._format_additional_params(vdur["additionalParams"]) + desc_params = self._format_additional_params(vdur.get("additionalParams")) + elif kdu_name: + kdur = next((x for x in db_vnfr["kdur"] if x["kdu-name"] == kdu_name), None) + desc_params = self._format_additional_params(kdur.get("additionalParams")) + else: + desc_params = self._format_additional_params(db_vnfr.get("additionalParamsForVnf")) else: - if db_nsr.get("additionalParamsForNs"): - desc_params.update(self._format_additional_params(db_nsr["additionalParamsForNs"])) + 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 - output, detail = await self._ns_execute_primitive( - db_deployed=nsr_deployed, - member_vnf_index=vnf_index, - vdu_id=vdu_id, - vdu_name=vdu_name, - vdu_count_index=vdu_count_index, - primitive=primitive, - primitive_params=self._map_primitive_params(config_primitive_desc, primitive_params, desc_params)) - - detailed_status = output - if detail == 'OK': - result = 'COMPLETED' + if kdu_name and (primitive_name in ("upgrade", "rollback", "status") or kdu_action): + # kdur and desc_params already set from before + if primitive_params: + desc_params.update(primitive_params) + # TODO Check if we will need something at vnf level + for index, kdu in enumerate(get_iterable(nsr_deployed, "K8s")): + if kdu_name == kdu["kdu-name"] and kdu["member-vnf-index"] == vnf_index: + break + else: + raise LcmException("KDU '{}' for vnf '{}' not deployed".format(kdu_name, vnf_index)) + + if kdu.get("k8scluster-type") not in self.k8scluster_map: + msg = "unknown k8scluster-type '{}'".format(kdu.get("k8scluster-type")) + raise LcmException(msg) + + db_dict = {"collection": "nsrs", + "filter": {"_id": nsr_id}, + "path": "_admin.deployed.K8s.{}".format(index)} + self.logger.debug(logging_text + "Exec k8s {} on {}.{}".format(primitive_name, vnf_index, kdu_name)) + step = "Executing kdu {}".format(primitive_name) + if primitive_name == "upgrade": + if desc_params.get("kdu_model"): + kdu_model = desc_params.get("kdu_model") + del desc_params["kdu_model"] + else: + kdu_model = kdu.get("kdu-model") + parts = kdu_model.split(sep=":") + if len(parts) == 2: + kdu_model = parts[0] + + detailed_status = await asyncio.wait_for( + self.k8scluster_map[kdu["k8scluster-type"]].upgrade( + cluster_uuid=kdu.get("k8scluster-uuid"), + kdu_instance=kdu.get("kdu-instance"), + atomic=True, kdu_model=kdu_model, + params=desc_params, db_dict=db_dict, + timeout=timeout_ns_action), + timeout=timeout_ns_action + 10) + self.logger.debug(logging_text + " Upgrade of kdu {} done".format(detailed_status)) + elif primitive_name == "rollback": + detailed_status = await asyncio.wait_for( + self.k8scluster_map[kdu["k8scluster-type"]].rollback( + cluster_uuid=kdu.get("k8scluster-uuid"), + kdu_instance=kdu.get("kdu-instance"), + db_dict=db_dict), + timeout=timeout_ns_action) + elif primitive_name == "status": + detailed_status = await asyncio.wait_for( + self.k8scluster_map[kdu["k8scluster-type"]].status_kdu( + 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_name, + 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: - result = 'FAILED' - - db_nslcmop_update["detailed-status"] = nslcmop_operation_state_detail = detailed_status - db_nslcmop_update["operationState"] = nslcmop_operation_state = result - db_nslcmop_update["statusEnteredTime"] = time() - self.logger.debug(logging_text + " task Done with result {} {}".format(result, detailed_status)) + ee_id, vca_type = self._look_for_deployed_vca(nsr_deployed["VCA"], + member_vnf_index=vnf_index, + vdu_id=vdu_id, + vdu_count_index=vdu_count_index, + ee_descriptor_id=ee_descriptor_id) + db_nslcmop_notif = {"collection": "nslcmops", + "filter": {"_id": nslcmop_id}, + "path": "admin.VCA"} + nslcmop_operation_state, detailed_status = await self._ns_execute_primitive( + ee_id, + primitive=primitive_name, + primitive_params=self._map_primitive_params(config_primitive_desc, primitive_params, desc_params), + timeout=timeout_ns_action, + vca_type=vca_type, + db_dict=db_nslcmop_notif) + + db_nslcmop_update["detailed-status"] = detailed_status + error_description_nslcmop = detailed_status if nslcmop_operation_state == "FAILED" else "" + self.logger.debug(logging_text + " task Done with result {} {}".format(nslcmop_operation_state, + detailed_status)) return # database update is called inside finally - except (DbException, LcmException) as e: + 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 and db_nslcmop: - db_nslcmop_update["detailed-status"] = nslcmop_operation_state_detail = \ + if exc: + db_nslcmop_update["detailed-status"] = detailed_status = error_description_nslcmop = \ "FAILED {}: {}".format(step, exc) - db_nslcmop_update["operationState"] = nslcmop_operation_state = "FAILED" - db_nslcmop_update["statusEnteredTime"] = time() - try: - if db_nslcmop_update: - self.update_db_2("nslcmops", nslcmop_id, db_nslcmop_update) - if db_nsr: - db_nsr_update["_admin.nslcmop"] = None - db_nsr_update["_admin.operation-type"] = None - db_nsr_update["_admin.nslcmop"] = None - db_nsr_update["_admin.current-operation"] = None - self.update_db_2("nsrs", nsr_id, db_nsr_update) - except DbException as e: - self.logger.error(logging_text + "Cannot update database: {}".format(e)) - self.logger.debug(logging_text + "Exit") + nslcmop_operation_state = "FAILED" + if db_nsr: + self._write_ns_status( + nsr_id=nsr_id, + ns_state=db_nsr["nsState"], # TODO check if degraded. For the moment use previous status + current_operation="IDLE", + current_operation_id=None, + # error_description=error_description_nsr, + # error_detail=error_detail, + 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", "actioned", {"nsr_id": nsr_id, "nslcmop_id": nslcmop_id, @@ -2524,7 +3929,7 @@ class NsLcm(LcmBase): 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_action") - return nslcmop_operation_state, nslcmop_operation_state_detail + return nslcmop_operation_state, detailed_status async def scale(self, nsr_id, nslcmop_id): @@ -2540,21 +3945,24 @@ class NsLcm(LcmBase): db_nslcmop = None db_nslcmop_update = {} nslcmop_operation_state = None - db_nsr_update = {"_admin.nslcmop": nslcmop_id, - "_admin.current-operation": nslcmop_id, - "_admin.operation-type": "scale"} - self.update_db_2("nsrs", nsr_id, db_nsr_update) + db_nsr_update = {} exc = None # in case of error, indicates what part of scale was failed to put nsr at error status scale_process = None old_operational_status = "" old_config_status = "" - vnfr_scaled = False 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="SCALING", + 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}) @@ -2637,8 +4045,25 @@ class NsLcm(LcmBase): vdu_scaling_info["scaling_direction"] = "OUT" vdu_scaling_info["vdu-create"] = {} for vdu_scale_info in scaling_descriptor["vdu"]: + vdud = next(vdu for vdu in db_vnfd.get("vdu") if vdu["id"] == vdu_scale_info["vdu-id-ref"]) + vdu_index = len([x for x in db_vnfr.get("vdur", ()) + if x.get("vdu-id-ref") == vdu_scale_info["vdu-id-ref"] and + x.get("member-vnf-index-ref") == vnf_index]) + cloud_init_text = self._get_cloud_init(vdud, db_vnfd) + if cloud_init_text: + additional_params = self._get_vdu_additional_params(db_vnfr, vdud["id"]) or {} + cloud_init_list = [] + for x in range(vdu_scale_info.get("count", 1)): + if cloud_init_text: + # TODO Information of its own ip is not available because db_vnfr is not updated. + additional_params["OSM"] = self._get_osm_params(db_vnfr, vdu_scale_info["vdu-id-ref"], + vdu_index + x) + cloud_init_list.append(self._parse_cloud_init(cloud_init_text, additional_params, + db_vnfd["id"], vdud["id"])) RO_scaling_info.append({"osm_vdu_id": vdu_scale_info["vdu-id-ref"], "member-vnf-index": vnf_index, "type": "create", "count": vdu_scale_info.get("count", 1)}) + if cloud_init_list: + RO_scaling_info[-1]["cloud_init"] = cloud_init_list vdu_scaling_info["vdu-create"][vdu_scale_info["vdu-id-ref"]] = vdu_scale_info.get("count", 1) elif scaling_type == "SCALE_IN": @@ -2695,7 +4120,7 @@ class NsLcm(LcmBase): raise LcmException( "Invalid vnfd descriptor at scaling-group-descriptor[name='{}']:scaling-config-action" "[vnf-config-primitive-name-ref='{}'] does not match any vnf-configuration:config-" - "primitive".format(scaling_group, config_primitive)) + "primitive".format(scaling_group, vnf_config_primitive)) vnfr_params = {"VDU_SCALE_INFO": vdu_scaling_info} if db_vnfr.get("additionalParamsForVnf"): @@ -2705,10 +4130,10 @@ 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' @@ -2716,22 +4141,30 @@ 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 + ee_descriptor_id = config_primitive.get("execution-environment-ref") + primitive_name = config_primitive.get("execution-environment-primitive", + vnf_config_primitive) + ee_id, vca_type = self._look_for_deployed_vca(nsr_deployed["VCA"], + member_vnf_index=vnf_index, + vdu_id=None, + vdu_count_index=None, + ee_descriptor_id=ee_descriptor_id) result, result_detail = await self._ns_execute_primitive( - nsr_deployed, vnf_index, None, None, None, vnf_config_primitive, primitive_params) + ee_id, primitive_name, primitive_params, vca_type) self.logger.debug(logging_text + "vnf_config_primitive={} Done with result {} {}".format( vnf_config_primitive, result, result_detail)) # Update operationState = COMPLETED | FAILED @@ -2749,26 +4182,26 @@ 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}) @@ -2779,7 +4212,7 @@ class NsLcm(LcmBase): db_nslcmop_update["_admin.deploy.RO"] = RO_nslcmop_id RO_task_done = False - step = detailed_status = "Waiting RO_task_id={} to complete the scale action.".format(RO_nslcmop_id) + step = detailed_status = "Waiting for VIM to scale. RO_task_id={}.".format(RO_nslcmop_id) detailed_status_old = None self.logger.debug(logging_text + step) @@ -2788,6 +4221,10 @@ 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) @@ -2795,11 +4232,16 @@ class NsLcm(LcmBase): detailed_status = step + "; {}".format(ns_status_info) elif ns_status == "ACTIVE": RO_task_done = True + self.scale_vnfr(db_vnfr, vdu_create=vdu_create, vdu_delete=vdu_delete) step = detailed_status = "Waiting ns ready at RO. RO_id={}".format(RO_nsr_id) self.logger.debug(logging_text + step) else: assert False, "ROclient.check_action_status returns unknown {}".format(ns_status) else: + desc = await self.RO.show("ns", RO_nsr_id) + ns_status, ns_status_info = self.RO.check_ns_status(desc) + # deploymentStatus + self._on_update_ro_db(nsrs_id=nsr_id, ro_descriptor=desc) if ns_status == "ERROR": raise ROclient.ROClientException(ns_status_info) @@ -2808,11 +4250,7 @@ class NsLcm(LcmBase): elif ns_status == "ACTIVE": step = detailed_status = \ "Waiting for management IP address reported by the VIM. Updating VNFRs" - if not vnfr_scaled: - self.scale_vnfr(db_vnfr, vdu_create=vdu_create, vdu_delete=vdu_delete) - vnfr_scaled = True try: - desc = await self.RO.show("ns", RO_nsr_id) # nsr_deployed["nsr_ip"] = RO.get_ns_vnf_info(desc) self.ns_update_vnfr({db_vnfr["member-vnf-index-ref"]: db_vnfr}, desc) break @@ -2878,18 +4316,18 @@ class NsLcm(LcmBase): if config_primitive["name"] == vnf_config_primitive: break else: - raise LcmException("Invalid vnfd descriptor at scaling-group-descriptor[name='{}']:" - "scaling-config-action[vnf-config-primitive-name-ref='{}'] does not " - "match any vnf-configuration:config-primitive".format(scaling_group, - config_primitive)) + raise LcmException( + "Invalid vnfd descriptor at scaling-group-descriptor[name='{}']:scaling-config-" + "action[vnf-config-primitive-name-ref='{}'] does not match any vnf-configuration:" + "config-primitive".format(scaling_group, vnf_config_primitive)) scale_process = "VCA" 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): + if op_index == self.SUBOPERATION_STATUS_SKIP: # Skip sub-operation result = 'COMPLETED' result_detail = 'Done' @@ -2897,22 +4335,30 @@ 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 + ee_descriptor_id = config_primitive.get("execution-environment-ref") + primitive_name = config_primitive.get("execution-environment-primitive", + vnf_config_primitive) + ee_id, vca_type = self._look_for_deployed_vca(nsr_deployed["VCA"], + member_vnf_index=vnf_index, + vdu_id=None, + vdu_count_index=None, + ee_descriptor_id=ee_descriptor_id) result, result_detail = await self._ns_execute_primitive( - nsr_deployed, vnf_index, None, None, None, vnf_config_primitive, primitive_params) + ee_id, primitive_name, primitive_params, vca_type) self.logger.debug(logging_text + "vnf_config_primitive={} Done with result {} {}".format( vnf_config_primitive, result, result_detail)) # Update operationState = COMPLETED | FAILED @@ -2925,9 +4371,6 @@ 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 @@ -2943,16 +4386,19 @@ 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) - 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 db_nsr_update["detailed-status"] = "" - db_nsr_update["_admin.nslcmop"] = None if scale_process: if "VCA" in scale_process: db_nsr_update["config-status"] = "failed" @@ -2960,16 +4406,27 @@ 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: - db_nsr_update["_admin.current-operation"] = None - db_nsr_update["_admin.operation-type"] = None - db_nsr_update["_admin.nslcmop"] = None - self.update_db_2("nsrs", nsr_id, db_nsr_update) - except DbException as e: - self.logger.error(logging_text + "Cannot update database: {}".format(e)) + 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 + ) + if nslcmop_operation_state: try: await self.msg.aiowrite("ns", "scaled", {"nsr_id": nsr_id, "nslcmop_id": nslcmop_id, @@ -2982,3 +4439,68 @@ class NsLcm(LcmBase): 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_scale") + + async def add_prometheus_metrics(self, ee_id, artifact_path, ee_config_descriptor, vnfr_id, nsr_id, target_ip): + if not self.prometheus: + return + # look if exist a file called 'prometheus*.j2' and + artifact_content = self.fs.dir_ls(artifact_path) + job_file = next((f for f in artifact_content if f.startswith("prometheus") and f.endswith(".j2")), None) + if not job_file: + return + with self.fs.file_open((artifact_path, job_file), "r") as f: + job_data = f.read() + + # TODO get_service + _, _, service = ee_id.partition(".") # remove prefix "namespace." + host_name = "{}-{}".format(service, ee_config_descriptor["metric-service"]) + host_port = "80" + vnfr_id = vnfr_id.replace("-", "") + variables = { + "JOB_NAME": vnfr_id, + "TARGET_IP": target_ip, + "EXPORTER_POD_IP": host_name, + "EXPORTER_POD_PORT": host_port, + } + job_list = self.prometheus.parse_job(job_data, variables) + # ensure job_name is using the vnfr_id. Adding the metadata nsr_id + for job in job_list: + if not isinstance(job.get("job_name"), str) or vnfr_id not in job["job_name"]: + job["job_name"] = vnfr_id + "_" + str(randint(1, 10000)) + job["nsr_id"] = nsr_id + job_dict = {jl["job_name"]: jl for jl in job_list} + if await self.prometheus.update(job_dict): + return list(job_dict.keys()) + + def get_vca_cloud_and_credentials(self, vim_account_id: str) -> (str, str): + """ + Get VCA Cloud and VCA Cloud Credentials for the VIM account + + :param: vim_account_id: VIM Account ID + + :return: (cloud_name, cloud_credential) + """ + config = self.get_vim_account_config(vim_account_id) + return config.get("vca_cloud"), config.get("vca_cloud_credential") + + def get_vca_k8s_cloud_and_credentials(self, vim_account_id: str) -> (str, str): + """ + Get VCA K8s Cloud and VCA K8s Cloud Credentials for the VIM account + + :param: vim_account_id: VIM Account ID + + :return: (cloud_name, cloud_credential) + """ + config = self.get_vim_account_config(vim_account_id) + return config.get("vca_k8s_cloud"), config.get("vca_k8s_cloud_credential") + + def get_vim_account_config(self, vim_account_id: str) -> dict: + """ + Get VIM Account config from the OSM Database + + :param: vim_account_id: VIM Account ID + + :return: Dictionary with the config of the vim account + """ + vim_account = self.db.get_one(table="vim_accounts", q_filter={"_id": vim_account_id}, fail_on_empty=False) + return vim_account.get("config", {}) if vim_account else {}