def cluster_create(request_data, request_context: ctx.RequestContext):
"""Request handler for cluster create operation.
Required data: org_name, ovdc_name, cluster_name
(data validation handled in broker)
:return: Dict
"""
_raise_error_if_pks_not_enabled()
required = [RequestKey.CLUSTER_NAME]
req_utils.validate_payload(request_data, required)
cluster_name = request_data[RequestKey.CLUSTER_NAME]
request_data['is_org_admin_search'] = True
try:
_get_cluster_and_broker(request_data, request_context, telemetry=False)
raise ClusterAlreadyExistsError(f"Cluster {cluster_name} "
f"already exists.")
except ClusterNotFoundError:
pass
k8s_metadata = \
ovdc_utils.get_ovdc_k8s_provider_metadata(
request_context.sysadmin_client,
org_name=request_data[RequestKey.ORG_NAME],
ovdc_name=request_data[RequestKey.OVDC_NAME],
include_credentials=True,
include_nsxt_info=True)
broker = _get_broker_from_k8s_metadata(k8s_metadata, request_context)
request_data[RequestKey.PKS_PLAN_NAME] = k8s_metadata[PKS_PLANS_KEY][0]
request_data[RequestKey.PKS_EXT_HOST] = \
f"{cluster_name}.{k8s_metadata[PKS_CLUSTER_DOMAIN_KEY]}"
return broker.create_cluster(data=request_data)
def cluster_create(request_data, tenant_auth_token):
"""Request handler for cluster create operation.
Required data: org_name, ovdc_name, cluster_name
Conditional data and default values:
if k8s_provider is 'native':
network_name, num_nodes=2, num_cpu=None, mb_memory=None,
storage_profile_name=None, template_name=default,
template_revision=default, ssh_key=None, enable_nfs=False,
rollback=True
(data validation handled in brokers)
:return: Dict
"""
required = [
RequestKey.CLUSTER_NAME
]
req_utils.validate_payload(request_data, required)
cluster_name = request_data[RequestKey.CLUSTER_NAME]
# TODO HACK 'is_org_admin_search' is used here to prevent users from
# creating clusters with the same name, including clusters in PKS
# True means that the cluster list is filtered by the org name of
# the logged-in user to check that there are no duplicate clusters
request_data['is_org_admin_search'] = True
try:
broker_manager.get_cluster_and_broker(request_data, tenant_auth_token)
raise ClusterAlreadyExistsError(f"Cluster {cluster_name} "
f"already exists.")
except ClusterNotFoundError:
pass
k8s_metadata = \
ovdc_utils.get_ovdc_k8s_provider_metadata(
org_name=request_data[RequestKey.ORG_NAME],
ovdc_name=request_data[RequestKey.OVDC_NAME],
include_credentials=True,
include_nsxt_info=True)
if k8s_metadata.get(K8S_PROVIDER_KEY) == K8sProvider.PKS:
request_data[RequestKey.PKS_PLAN_NAME] = k8s_metadata[PKS_PLANS_KEY][0]
request_data[RequestKey.PKS_EXT_HOST] = \
f"{cluster_name}.{k8s_metadata[PKS_CLUSTER_DOMAIN_KEY]}"
broker = broker_manager.get_broker_from_k8s_metadata(k8s_metadata,
tenant_auth_token)
return broker.create_cluster(request_data)
def _create_cluster(self, **cluster_spec):
cluster_name = cluster_spec['cluster_name']
# 'is_org_admin_search' is used here to prevent cluster creation with
# same cluster-name by users within org.
# If it is true, cluster list is filtered by the org name of the
# logged-in user to check for duplicates.
cluster, _ = self._find_cluster_in_org(cluster_name,
is_org_admin_search=True)
if not cluster:
ctr_prov_ctx = self._get_ctr_prov_ctx_from_ovdc_metadata()
if ctr_prov_ctx.get(K8S_PROVIDER_KEY) == K8sProviders.PKS:
cluster_spec['pks_plan'] = ctr_prov_ctx[PKS_PLANS_KEY][0]
cluster_spec['pks_ext_host'] = f"{cluster_name}." \
f"{ctr_prov_ctx[PKS_CLUSTER_DOMAIN_KEY]}"
broker = self._get_broker_based_on_ctr_prov_ctx(ctr_prov_ctx)
return broker.create_cluster(**cluster_spec)
else:
raise ClusterAlreadyExistsError(
f"Cluster {cluster_name} already exists.")
def create_cluster_thread(self):
network_name = self.req_spec.get(RequestKey.NETWORK_NAME)
try:
clusters = load_from_metadata(self.tenant_client,
name=self.cluster_name)
if len(clusters) != 0:
raise ClusterAlreadyExistsError(f"Cluster {self.cluster_name} "
"already exists.")
org_resource = self.tenant_client.get_org_by_name(
self.req_spec.get(RequestKey.ORG_NAME))
org = Org(self.tenant_client, resource=org_resource)
vdc_resource = org.get_vdc(self.req_spec.get(RequestKey.OVDC_NAME))
vdc = VDC(self.tenant_client, resource=vdc_resource)
template = self._get_template()
self._update_task(
TaskStatus.RUNNING,
message=f"Creating cluster vApp {self.cluster_name}"
f"({self.cluster_id})")
try:
vapp_resource = vdc.create_vapp(
self.cluster_name,
description=f"cluster {self.cluster_name}",
network=network_name,
fence_mode='bridged')
except Exception as e:
raise ClusterOperationError(
"Error while creating vApp:", str(e))
self.tenant_client.get_task_monitor().wait_for_status(
vapp_resource.Tasks.Task[0])
tags = {}
tags['cse.cluster.id'] = self.cluster_id
tags['cse.version'] = pkg_resources.require(
'container-service-extension')[0].version
tags['cse.template'] = template['name']
vapp = VApp(self.tenant_client, href=vapp_resource.get('href'))
for k, v in tags.items():
task = vapp.set_metadata('GENERAL', 'READWRITE', k, v)
self.tenant_client.get_task_monitor().wait_for_status(task)
self._update_task(
TaskStatus.RUNNING,
message=f"Creating master node for {self.cluster_name}"
f"({self.cluster_id})")
vapp.reload()
server_config = get_server_runtime_config()
try:
add_nodes(1, template, NodeType.MASTER, server_config,
self.tenant_client, org, vdc, vapp, self.req_spec)
except Exception as e:
raise MasterNodeCreationError(
"Error while adding master node:", str(e))
self._update_task(
TaskStatus.RUNNING,
message=f"Initializing cluster {self.cluster_name}"
f"({self.cluster_id})")
vapp.reload()
init_cluster(server_config, vapp, template)
master_ip = get_master_ip(server_config, vapp, template)
task = vapp.set_metadata('GENERAL', 'READWRITE', 'cse.master.ip',
master_ip)
self.tenant_client.get_task_monitor().wait_for_status(task)
if self.req_spec.get(RequestKey.NUM_WORKERS) > 0:
self._update_task(
TaskStatus.RUNNING,
message=f"Creating "
f"{self.req_spec.get(RequestKey.NUM_WORKERS)} "
f"node(s) for "
f"{self.cluster_name}({self.cluster_id})")
try:
add_nodes(self.req_spec.get(RequestKey.NUM_WORKERS),
template, NodeType.WORKER, server_config,
self.tenant_client, org, vdc, vapp,
self.req_spec)
except Exception as e:
raise WorkerNodeCreationError(
"Error while creating worker node:", str(e))
self._update_task(
TaskStatus.RUNNING,
message=f"Adding "
f"{self.req_spec.get(RequestKey.NUM_WORKERS)} "
f"node(s) to "
f"{self.cluster_name}({self.cluster_id})")
vapp.reload()
join_cluster(server_config, vapp, template)
if self.req_spec.get(RequestKey.ENABLE_NFS):
self._update_task(
TaskStatus.RUNNING,
message=f"Creating NFS node for {self.cluster_name}"
f"({self.cluster_id})")
try:
add_nodes(1, template, NodeType.NFS,
server_config, self.tenant_client, org, vdc,
vapp, self.req_spec)
except Exception as e:
raise NFSNodeCreationError(
"Error while creating NFS node:", str(e))
self._update_task(
TaskStatus.SUCCESS,
message=f"Created cluster {self.cluster_name}"
f"({self.cluster_id})")
except (MasterNodeCreationError, WorkerNodeCreationError,
NFSNodeCreationError, ClusterJoiningError,
ClusterInitializationError, ClusterOperationError) as e:
LOGGER.error(traceback.format_exc())
error_obj = error_to_json(e)
stack_trace = \
''.join(error_obj[ERROR_MESSAGE_KEY][ERROR_STACKTRACE_KEY])
self._update_task(
TaskStatus.ERROR,
error_message=error_obj[ERROR_MESSAGE_KEY]
[ERROR_DESCRIPTION_KEY],
stack_trace=stack_trace)
raise e
except Exception as e:
LOGGER.error(traceback.format_exc())
error_obj = error_to_json(e)
stack_trace = \
''.join(error_obj[ERROR_MESSAGE_KEY][ERROR_STACKTRACE_KEY])
self._update_task(
TaskStatus.ERROR,
error_message=error_obj[ERROR_MESSAGE_KEY][ERROR_DESCRIPTION_KEY], # noqa: E501
stack_trace=stack_trace)
finally:
self._disconnect_sys_admin()
def create_cluster(self, data):
"""Start the cluster creation operation.
Common broker function that validates data for the 'create cluster'
operation and returns a dictionary with cluster detail and task
information. Calls the asyncronous cluster create function that
actually performs the work. The returned `result['task_href']` can
be polled to get updates on task progress.
Required data: cluster_name, org_name, ovdc_name, network_name
Optional data and default values: num_nodes=2, num_cpu=None,
mb_memory=None, storage_profile_name=None, ssh_key_filepath=None,
template_name=default, template_revision=default, enable_nfs=False,
rollback=True
"""
required = [
RequestKey.CLUSTER_NAME,
RequestKey.ORG_NAME,
RequestKey.OVDC_NAME,
RequestKey.NETWORK_NAME
]
utils.ensure_keys_in_dict(required, data, dict_name='data')
cluster_name = data[RequestKey.CLUSTER_NAME]
# check that cluster name is syntactically valid
if not is_valid_cluster_name(cluster_name):
raise CseServerError(f"Invalid cluster name '{cluster_name}'")
# check that cluster name doesn't already exist
try:
get_cluster(self.tenant_client, cluster_name,
org_name=data[RequestKey.ORG_NAME],
ovdc_name=data[RequestKey.OVDC_NAME])
raise ClusterAlreadyExistsError(f"Cluster {cluster_name} "
f"already exists.")
except ClusterNotFoundError:
pass
# check that requested/default template is valid
template = get_template(
name=data.get(RequestKey.TEMPLATE_NAME),
revision=data.get(RequestKey.TEMPLATE_REVISION))
defaults = {
RequestKey.NUM_WORKERS: 2,
RequestKey.NUM_CPU: None,
RequestKey.MB_MEMORY: None,
RequestKey.STORAGE_PROFILE_NAME: None,
RequestKey.SSH_KEY_FILEPATH: None,
RequestKey.TEMPLATE_NAME: template[LocalTemplateKey.NAME],
RequestKey.TEMPLATE_REVISION: template[LocalTemplateKey.REVISION],
RequestKey.ENABLE_NFS: False,
RequestKey.ROLLBACK: True,
}
validated_data = {**defaults, **data}
# TODO HACK default dictionary combining needs to be fixed
validated_data[RequestKey.TEMPLATE_NAME] = validated_data[RequestKey.TEMPLATE_NAME] or template[LocalTemplateKey.NAME] # noqa: E501
validated_data[RequestKey.TEMPLATE_REVISION] = validated_data[RequestKey.TEMPLATE_REVISION] or template[LocalTemplateKey.REVISION] # noqa: E501
template_name = validated_data[RequestKey.TEMPLATE_NAME]
template_revision = validated_data[RequestKey.TEMPLATE_REVISION]
# check that requested number of worker nodes is at least more than 1
num_workers = validated_data[RequestKey.NUM_WORKERS]
if num_workers < 1:
raise CseServerError(f"Worker node count must be > 0 "
f"(received {num_workers}).")
cluster_id = str(uuid.uuid4())
# must _update_task or else self.task_resource is None
# do not logout of sys admin, or else in pyvcloud's session.request()
# call, session becomes None
self._update_task(
TaskStatus.RUNNING,
message=f"Creating cluster vApp '{cluster_name}' ({cluster_id})"
f" from template '{template_name}' "
f"(revision {template_revision})")
self._create_cluster_async(
org_name=validated_data[RequestKey.ORG_NAME],
ovdc_name=validated_data[RequestKey.OVDC_NAME],
cluster_name=cluster_name,
cluster_id=cluster_id,
template_name=template_name,
template_revision=template_revision,
num_workers=validated_data[RequestKey.NUM_WORKERS],
network_name=validated_data[RequestKey.NETWORK_NAME],
num_cpu=validated_data[RequestKey.NUM_CPU],
mb_memory=validated_data[RequestKey.MB_MEMORY],
storage_profile_name=validated_data[RequestKey.STORAGE_PROFILE_NAME], # noqa: E501
ssh_key_filepath=validated_data[RequestKey.SSH_KEY_FILEPATH],
enable_nfs=validated_data[RequestKey.ENABLE_NFS],
rollback=validated_data[RequestKey.ROLLBACK])
return {
'name': cluster_name,
'cluster_id': cluster_id,
'task_href': self.task_resource.get('href')
}
def create_cluster_thread(self):
network_name = self.body['network']
try:
clusters = load_from_metadata(self.tenant_client,
name=self.cluster_name)
if len(clusters) != 0:
raise ClusterAlreadyExistsError(f"Cluster {self.cluster_name} "
"already exists.")
org_resource = self.tenant_client.get_org()
org = Org(self.tenant_client, resource=org_resource)
vdc_resource = org.get_vdc(self.body['vdc'])
vdc = VDC(self.tenant_client, resource=vdc_resource)
template = self.get_template()
self.update_task(TaskStatus.RUNNING,
message='Creating cluster vApp %s(%s)' %
(self.cluster_name, self.cluster_id))
try:
vapp_resource = vdc.create_vapp(self.cluster_name,
description='cluster %s' %
self.cluster_name,
network=network_name,
fence_mode='bridged')
except Exception as e:
raise ClusterOperationError('Error while creating vApp:',
str(e))
self.tenant_client.get_task_monitor().wait_for_status(
vapp_resource.Tasks.Task[0])
tags = {}
tags['cse.cluster.id'] = self.cluster_id
tags['cse.version'] = pkg_resources.require(
'container-service-extension')[0].version
tags['cse.template'] = template['name']
vapp = VApp(self.tenant_client, href=vapp_resource.get('href'))
for k, v in tags.items():
task = vapp.set_metadata('GENERAL', 'READWRITE', k, v)
self.tenant_client.get_task_monitor().wait_for_status(task)
self.update_task(TaskStatus.RUNNING,
message='Creating master node for %s(%s)' %
(self.cluster_name, self.cluster_id))
vapp.reload()
server_config = get_server_runtime_config()
try:
add_nodes(1, template, TYPE_MASTER, server_config,
self.tenant_client, org, vdc, vapp, self.body)
except Exception as e:
raise MasterNodeCreationError(
"Error while adding master node:", str(e))
self.update_task(TaskStatus.RUNNING,
message='Initializing cluster %s(%s)' %
(self.cluster_name, self.cluster_id))
vapp.reload()
init_cluster(server_config, vapp, template)
master_ip = get_master_ip(server_config, vapp, template)
task = vapp.set_metadata('GENERAL', 'READWRITE', 'cse.master.ip',
master_ip)
self.tenant_client.get_task_monitor().wait_for_status(task)
if self.body['node_count'] > 0:
self.update_task(TaskStatus.RUNNING,
message='Creating %s node(s) for %s(%s)' %
(self.body['node_count'], self.cluster_name,
self.cluster_id))
try:
add_nodes(self.body['node_count'], template, TYPE_NODE,
server_config, self.tenant_client, org, vdc,
vapp, self.body)
except Exception as e:
raise WorkerNodeCreationError(
"Error while creating worker node:", str(e))
self.update_task(TaskStatus.RUNNING,
message='Adding %s node(s) to %s(%s)' %
(self.body['node_count'], self.cluster_name,
self.cluster_id))
vapp.reload()
join_cluster(server_config, vapp, template)
if self.body['enable_nfs']:
self.update_task(TaskStatus.RUNNING,
message='Creating NFS node for %s(%s)' %
(self.cluster_name, self.cluster_id))
try:
add_nodes(1, template, TYPE_NFS, server_config,
self.tenant_client, org, vdc, vapp, self.body)
except Exception as e:
raise NFSNodeCreationError(
"Error while creating NFS node:", str(e))
self.update_task(TaskStatus.SUCCESS,
message='Created cluster %s(%s)' %
(self.cluster_name, self.cluster_id))
except (MasterNodeCreationError, WorkerNodeCreationError,
NFSNodeCreationError, ClusterJoiningError,
ClusterInitializationError, ClusterOperationError) as e:
LOGGER.error(traceback.format_exc())
error_obj = error_to_json(e)
stack_trace = ''.join(error_obj[ERROR_MESSAGE][ERROR_STACKTRACE])
self.update_task(
TaskStatus.ERROR,
error_message=error_obj[ERROR_MESSAGE][ERROR_DESCRIPTION],
stack_trace=stack_trace)
raise e
except Exception as e:
LOGGER.error(traceback.format_exc())
error_obj = error_to_json(e)
stack_trace = ''.join(error_obj[ERROR_MESSAGE][ERROR_STACKTRACE])
self.update_task(
TaskStatus.ERROR,
error_message=error_obj[ERROR_MESSAGE][ERROR_DESCRIPTION],
stack_trace=stack_trace)
finally:
self._disconnect_sys_admin()