X-Git-Url: https://osm.etsi.org/gitweb/?p=osm%2FN2VC.git;a=blobdiff_plain;f=n2vc%2Fk8s_helm_conn.py;h=496e320d811a10ed6ecc3c00340072f3a33ba1e5;hp=eac34d4bd721cf969dd5980d8c172e32d9281ff1;hb=65ddf85ef9d11aa4d4f9dc7cb2912ff7069f7c94;hpb=9a1bd31f1e8d0a70b8837a057582cb2e82e66194 diff --git a/n2vc/k8s_helm_conn.py b/n2vc/k8s_helm_conn.py index eac34d4..496e320 100644 --- a/n2vc/k8s_helm_conn.py +++ b/n2vc/k8s_helm_conn.py @@ -20,7 +20,6 @@ # contact with: nfvlabs@tid.es ## -import paramiko import subprocess import os import shutil @@ -30,6 +29,7 @@ import yaml from uuid import uuid4 import random from n2vc.k8s_conn import K8sConnector +from n2vc.exceptions import K8sException class K8sHelmConnector(K8sConnector): @@ -83,6 +83,16 @@ class K8sHelmConnector(K8sConnector): self._helm_command = helm_command self._check_file_exists(filename=helm_command, exception_if_not_exists=True) + # initialize helm client-only + self.debug('Initializing helm client-only...') + command = '{} init --client-only'.format(self._helm_command) + try: + asyncio.ensure_future(self._local_async_exec(command=command, raise_exception_on_error=False)) + # loop = asyncio.get_event_loop() + # loop.run_until_complete(self._local_async_exec(command=command, raise_exception_on_error=False)) + except Exception as e: + self.warning(msg='helm init failed (it was already initialized): {}'.format(e)) + self.info('K8S Helm connector initialized') async def init_env( @@ -91,6 +101,17 @@ class K8sHelmConnector(K8sConnector): namespace: str = 'kube-system', reuse_cluster_uuid=None ) -> (str, bool): + """ + It prepares a given K8s cluster environment to run Charts on both sides: + client (OSM) + server (Tiller) + + :param k8s_creds: credentials to access a given K8s cluster, i.e. a valid '.kube/config' + :param namespace: optional namespace to be used for helm. By default, 'kube-system' will be used + :param reuse_cluster_uuid: existing cluster uuid for reuse + :return: uuid of the K8s cluster and True if connector has installed some software in the cluster + (on error, an exception will be raised) + """ cluster_uuid = reuse_cluster_uuid if not cluster_uuid: @@ -186,7 +207,8 @@ class K8sHelmConnector(K8sConnector): kube_dir, helm_dir, config_filename, cluster_dir = \ self._get_paths(cluster_name=cluster_uuid, create_if_not_exist=True) - command = '{} --kubeconfig={} --home={} repo list --output yaml'.format(self._helm_command, config_filename, helm_dir) + command = '{} --kubeconfig={} --home={} repo list --output yaml'\ + .format(self._helm_command, config_filename, helm_dir) output, rc = await self._local_async_exec(command=command, raise_exception_on_error=True) if output and len(output) > 0: @@ -247,7 +269,7 @@ class K8sHelmConnector(K8sConnector): msg = 'Cluster has releases and not force. Cannot reset K8s environment. Cluster uuid: {}'\ .format(cluster_uuid) self.error(msg) - raise Exception(msg) + raise K8sException(msg) if uninstall_sw: @@ -269,7 +291,6 @@ class K8sHelmConnector(K8sConnector): else: msg = 'Tiller deployment not found in cluster {}'.format(cluster_uuid) self.error(msg) - # raise Exception(msg) self.debug('namespace for tiller: {}'.format(namespace)) @@ -305,14 +326,12 @@ class K8sHelmConnector(K8sConnector): atomic: bool = True, timeout: float = 300, params: dict = None, - db_dict: dict = None + db_dict: dict = None, + kdu_name: str = None ): self.debug('installing {} in cluster {}'.format(kdu_model, cluster_uuid)) - start = time.time() - end = start + timeout - # config filename kube_dir, helm_dir, config_filename, cluster_dir = \ self._get_paths(cluster_name=cluster_uuid, create_if_not_exist=True) @@ -338,7 +357,7 @@ class K8sHelmConnector(K8sConnector): version_str = '--version {}'.format(parts[1]) kdu_model = parts[0] - # generate a name for the releas. Then, check if already exists + # generate a name for the release. Then, check if already exists kdu_instance = None while kdu_instance is None: kdu_instance = K8sHelmConnector._generate_release_name(kdu_model) @@ -351,8 +370,8 @@ class K8sHelmConnector(K8sConnector): if result is not None: # instance already exists: generate a new one kdu_instance = None - except: - kdu_instance = None + except K8sException: + pass # helm repo install command = '{} install {} --output yaml --kubeconfig={} --home={} {} {} --name={} {} {}'\ @@ -365,6 +384,7 @@ class K8sHelmConnector(K8sConnector): exec_task = asyncio.ensure_future( coro_or_future=self._local_async_exec(command=command, raise_exception_on_error=False) ) + # write status in another task status_task = asyncio.ensure_future( coro_or_future=self._store_status( @@ -405,7 +425,7 @@ class K8sHelmConnector(K8sConnector): if rc != 0: msg = 'Error executing command: {}\nOutput: {}'.format(command, output) self.error(msg) - raise Exception(msg) + raise K8sException(msg) self.debug('Returning kdu_instance {}'.format(kdu_instance)) return kdu_instance @@ -450,9 +470,6 @@ class K8sHelmConnector(K8sConnector): self.debug('upgrading {} in cluster {}'.format(kdu_model, cluster_uuid)) - start = time.time() - end = start + timeout - # config filename kube_dir, helm_dir, config_filename, cluster_dir = \ self._get_paths(cluster_name=cluster_uuid, create_if_not_exist=True) @@ -502,7 +519,7 @@ class K8sHelmConnector(K8sConnector): ) # wait for execution task - await asyncio.wait([ exec_task ]) + await asyncio.wait([exec_task]) # cancel status task status_task.cancel() @@ -529,7 +546,7 @@ class K8sHelmConnector(K8sConnector): if rc != 0: msg = 'Error executing command: {}\nOutput: {}'.format(command, output) self.error(msg) - raise Exception(msg) + raise K8sException(msg) # return new revision number instance = await self.get_instance_info(cluster_uuid=cluster_uuid, kdu_instance=kdu_instance) @@ -594,7 +611,7 @@ class K8sHelmConnector(K8sConnector): if rc != 0: msg = 'Error executing command: {}\nOutput: {}'.format(command, output) self.error(msg) - raise Exception(msg) + raise K8sException(msg) # return new revision number instance = await self.get_instance_info(cluster_uuid=cluster_uuid, kdu_instance=kdu_instance) @@ -634,40 +651,113 @@ class K8sHelmConnector(K8sConnector): async def inspect_kdu( self, - kdu_model: str + kdu_model: str, + repo_url: str = None ) -> str: - self.debug('inspect kdu_model {}'.format(kdu_model)) + self.debug('inspect kdu_model {} from (optional) repo: {}'.format(kdu_model, repo_url)) + + return await self._exec_inspect_comand(inspect_command='', kdu_model=kdu_model, repo_url=repo_url) - command = '{} inspect values {}'\ - .format(self._helm_command, kdu_model) + async def values_kdu( + self, + kdu_model: str, + repo_url: str = None + ) -> str: - output, rc = await self._local_async_exec(command=command) + self.debug('inspect kdu_model values {} from (optional) repo: {}'.format(kdu_model, repo_url)) - return output + return await self._exec_inspect_comand(inspect_command='values', kdu_model=kdu_model, repo_url=repo_url) async def help_kdu( self, - kdu_model: str - ): - - self.debug('help kdu_model {}'.format(kdu_model)) - - command = '{} inspect readme {}'\ - .format(self._helm_command, kdu_model) + kdu_model: str, + repo_url: str = None + ) -> str: - output, rc = await self._local_async_exec(command=command, raise_exception_on_error=True) + self.debug('inspect kdu_model {} readme.md from repo: {}'.format(kdu_model, repo_url)) - return output + return await self._exec_inspect_comand(inspect_command='readme', kdu_model=kdu_model, repo_url=repo_url) async def status_kdu( self, cluster_uuid: str, kdu_instance: str - ): + ) -> str: + + # call internal function + return await self._status_kdu( + cluster_uuid=cluster_uuid, + kdu_instance=kdu_instance, + show_error_log=True, + return_text=True + ) + + async def synchronize_repos(self, cluster_uuid: str): + + self.debug("syncronize repos for cluster helm-id: {}",) + try: + update_repos_timeout = 300 # max timeout to sync a single repos, more than this is too much + db_k8scluster = self.db.get_one("k8sclusters", {"_admin.helm-chart.id": cluster_uuid}) + if db_k8scluster: + nbi_repo_list = db_k8scluster.get("_admin").get("helm_chart_repos") or [] + cluster_repo_dict = db_k8scluster.get("_admin").get("helm_charts_added") or {} + # elements that must be deleted + deleted_repo_list = [] + added_repo_dict = {} + self.debug("helm_chart_repos: {}".format(nbi_repo_list)) + self.debug("helm_charts_added: {}".format(cluster_repo_dict)) + + # obtain repos to add: registered by nbi but not added + repos_to_add = [repo for repo in nbi_repo_list if not cluster_repo_dict.get(repo)] + + # obtain repos to delete: added by cluster but not in nbi list + repos_to_delete = [repo for repo in cluster_repo_dict.keys() if repo not in nbi_repo_list] + + # delete repos: must delete first then add because there may be different repos with same name but + # different id and url + self.debug("repos to delete: {}".format(repos_to_delete)) + for repo_id in repos_to_delete: + # try to delete repos + try: + repo_delete_task = asyncio.ensure_future(self.repo_remove(cluster_uuid=cluster_uuid, + name=cluster_repo_dict[repo_id])) + await asyncio.wait_for(repo_delete_task, update_repos_timeout) + except Exception as e: + self.warning("Error deleting repo, id: {}, name: {}, err_msg: {}".format(repo_id, + cluster_repo_dict[repo_id], str(e))) + # always add to the list of to_delete if there is an error because if is not there deleting raises error + deleted_repo_list.append(repo_id) + + # add repos + self.debug("repos to add: {}".format(repos_to_add)) + add_task_list = [] + for repo_id in repos_to_add: + # obtain the repo data from the db + # if there is an error getting the repo in the database we will ignore this repo and continue + # because there is a possible race condition where the repo has been deleted while processing + db_repo = self.db.get_one("k8srepos", {"_id": repo_id}) + self.debug("obtained repo: id, {}, name: {}, url: {}".format(repo_id, db_repo["name"], db_repo["url"])) + try: + repo_add_task = asyncio.ensure_future(self.repo_add(cluster_uuid=cluster_uuid, + name=db_repo["name"], url=db_repo["url"], + repo_type="chart")) + await asyncio.wait_for(repo_add_task, update_repos_timeout) + added_repo_dict[repo_id] = db_repo["name"] + self.debug("added repo: id, {}, name: {}".format(repo_id, db_repo["name"])) + except Exception as e: + # deal with error adding repo, adding a repo that already exists does not raise any error + # will not raise error because a wrong repos added by anyone could prevent instantiating any ns + self.error("Error adding repo id: {}, err_msg: {} ".format(repo_id, repr(e))) - return await self._status_kdu(cluster_uuid=cluster_uuid, kdu_instance=kdu_instance, show_error_log=True) + return deleted_repo_list, added_repo_dict + else: # else db_k8scluster does not exist + raise K8sException("k8cluster with helm-id : {} not found".format(cluster_uuid)) + + except Exception as e: + self.error("Error synchronizing repos: {}".format(str(e))) + raise K8sException("Error synchronizing repos") """ ################################################################################################## @@ -675,11 +765,32 @@ class K8sHelmConnector(K8sConnector): ################################################################################################## """ + async def _exec_inspect_comand( + self, + inspect_command: str, + kdu_model: str, + repo_url: str = None + ): + + repo_str = '' + if repo_url: + repo_str = ' --repo {}'.format(repo_url) + idx = kdu_model.find('/') + if idx >= 0: + idx += 1 + kdu_model = kdu_model[idx:] + + inspect_command = '{} inspect {} {}{}'.format(self._helm_command, inspect_command, kdu_model, repo_str) + output, rc = await self._local_async_exec(command=inspect_command, encode_utf8=True) + + return output + async def _status_kdu( self, cluster_uuid: str, kdu_instance: str, - show_error_log: bool = False + show_error_log: bool = False, + return_text: bool = False ): self.debug('status of kdu_instance {}'.format(kdu_instance)) @@ -697,6 +808,9 @@ class K8sHelmConnector(K8sConnector): show_error_log=show_error_log ) + if return_text: + return str(output) + if rc != 0: return None @@ -734,6 +848,15 @@ class K8sHelmConnector(K8sConnector): def _generate_release_name( chart_name: str ): + # check embeded chart (file or dir) + if chart_name.startswith('/'): + # extract file or directory name + chart_name = chart_name[chart_name.rfind('/')+1:] + # check URL + elif '://' in chart_name: + # extract last portion of URL + chart_name = chart_name[chart_name.rfind('/')+1:] + name = '' for c in chart_name: if c.isalpha() or c.isnumeric(): @@ -752,7 +875,7 @@ class K8sHelmConnector(K8sConnector): def get_random_number(): r = random.randrange(start=1, stop=99999999) s = str(r) - s = s.rjust(width=10, fillchar=' ') + s = s.rjust(10, '0') return s name = name + get_random_number() @@ -772,10 +895,8 @@ class K8sHelmConnector(K8sConnector): await asyncio.sleep(check_every) detailed_status = await self.status_kdu(cluster_uuid=cluster_uuid, kdu_instance=kdu_instance) status = detailed_status.get('info').get('Description') - print('=' * 60) self.debug('STATUS:\n{}'.format(status)) self.debug('DETAILED STATUS:\n{}'.format(detailed_status)) - print('=' * 60) # write status to db result = await self.write_app_status_to_db( db_dict=db_dict, @@ -789,6 +910,7 @@ class K8sHelmConnector(K8sConnector): self.debug('Task cancelled') return except Exception as e: + self.debug('_store_status exception: {}'.format(str(e))) pass finally: if run_once: @@ -800,7 +922,7 @@ class K8sHelmConnector(K8sConnector): kdu_instance: str ) -> bool: - status = await self.status_kdu(cluster_uuid=cluster_uuid, kdu_instance=kdu_instance) + status = await self._status_kdu(cluster_uuid=cluster_uuid, kdu_instance=kdu_instance, return_text=False) # extract info.status.resources-> str # format: @@ -877,7 +999,6 @@ class K8sHelmConnector(K8sConnector): # params for use in -f file # returns values file option and filename (in order to delete it at the end) def _params_to_file_option(self, cluster_uuid: str, params: dict) -> (str, str): - params_str = '' if params and len(params) > 0: kube_dir, helm_dir, config_filename, cluster_dir = \ @@ -894,7 +1015,7 @@ class K8sHelmConnector(K8sConnector): for key in params: value = params.get(key) if '!!yaml' in str(value): - value = yaml.safe_load(value[7:]) + value = yaml.load(value[7:]) params2[key] = value values_file = get_random_number() + '.yaml' @@ -969,7 +1090,7 @@ class K8sHelmConnector(K8sConnector): if not os.path.exists(cluster_dir): msg = 'Base cluster dir {} does not exist'.format(cluster_dir) self.error(msg) - raise Exception(msg) + raise K8sException(msg) # kube dir kube_dir = cluster_dir + '/' + '.kube' @@ -979,7 +1100,7 @@ class K8sHelmConnector(K8sConnector): if not os.path.exists(kube_dir): msg = 'Kube config dir {} does not exist'.format(kube_dir) self.error(msg) - raise Exception(msg) + raise K8sException(msg) # helm home dir helm_dir = cluster_dir + '/' + '.helm' @@ -989,7 +1110,7 @@ class K8sHelmConnector(K8sConnector): if not os.path.exists(helm_dir): msg = 'Helm config dir {} does not exist'.format(helm_dir) self.error(msg) - raise Exception(msg) + raise K8sException(msg) config_filename = kube_dir + '/config' return kube_dir, helm_dir, config_filename, cluster_dir @@ -1022,7 +1143,8 @@ class K8sHelmConnector(K8sConnector): self, command: str, raise_exception_on_error: bool = False, - show_error_log: bool = True + show_error_log: bool = True, + encode_utf8: bool = False ) -> (str, int): command = K8sHelmConnector._remove_multiple_spaces(command) @@ -1046,8 +1168,10 @@ class K8sHelmConnector(K8sConnector): output = '' if stdout: output = stdout.decode('utf-8').strip() + # output = stdout.decode() if stderr: output = stderr.decode('utf-8').strip() + # output = stderr.decode() if return_code != 0 and show_error_log: self.debug('Return code (FAIL): {}\nOutput:\n{}'.format(return_code, output)) @@ -1055,43 +1179,25 @@ class K8sHelmConnector(K8sConnector): self.debug('Return code: {}'.format(return_code)) if raise_exception_on_error and return_code != 0: - raise Exception(output) + raise K8sException(output) + + if encode_utf8: + output = output.encode('utf-8').strip() + output = str(output).replace('\\n', '\n') return output, return_code + except asyncio.CancelledError: + raise + except K8sException: + raise except Exception as e: msg = 'Exception executing command: {} -> {}'.format(command, e) - if show_error_log: - self.error(msg) - return '', -1 - - def _remote_exec( - self, - hostname: str, - username: str, - password: str, - command: str, - timeout: int = 10 - ) -> (str, int): - - command = K8sHelmConnector._remove_multiple_spaces(command) - self.debug('Executing sync remote ssh command: {}'.format(command)) - - ssh = paramiko.SSHClient() - ssh.set_missing_host_key_policy(paramiko.AutoAddPolicy()) - ssh.connect(hostname=hostname, username=username, password=password) - ssh_stdin, ssh_stdout, ssh_stderr = ssh.exec_command(command=command, timeout=timeout) - output = ssh_stdout.read().decode('utf-8') - error = ssh_stderr.read().decode('utf-8') - if error: - self.error('ERROR: {}'.format(error)) - return_code = 1 - else: - return_code = 0 - output = output.replace('\\n', '\n') - self.debug('OUTPUT: {}'.format(output)) - - return output, return_code + self.error(msg) + if raise_exception_on_error: + raise K8sException(e) from e + else: + return '', -1 def _check_file_exists(self, filename: str, exception_if_not_exists: bool = False): self.debug('Checking if file {} exists...'.format(filename)) @@ -1101,5 +1207,6 @@ class K8sHelmConnector(K8sConnector): msg = 'File {} does not exist'.format(filename) if exception_if_not_exists: self.error(msg) - raise Exception(msg) + raise K8sException(msg) +