magnum/magnum/api/controllers/v1/cluster.py

710 lines
30 KiB
Python
Executable File

# Copyright 2013 UnitedStack Inc.
# All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License"); you may
# not use this file except in compliance with the License. You may obtain
# a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
# WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
# License for the specific language governing permissions and limitations
# under the License.
import uuid
from oslo_log import log as logging
from oslo_utils import strutils
from oslo_utils import timeutils
import pecan
import six
import warnings
import wsme
from wsme import types as wtypes
from magnum.api import attr_validator
from magnum.api.controllers import base
from magnum.api.controllers import link
from magnum.api.controllers.v1 import cluster_actions
from magnum.api.controllers.v1 import collection
from magnum.api.controllers.v1 import nodegroup
from magnum.api.controllers.v1 import types
from magnum.api import expose
from magnum.api import utils as api_utils
from magnum.api import validation
from magnum.common import exception
from magnum.common import name_generator
from magnum.common import policy
import magnum.conf
from magnum.i18n import _
from magnum import objects
from magnum.objects import fields
LOG = logging.getLogger(__name__)
CONF = magnum.conf.CONF
class ClusterID(wtypes.Base):
"""API representation of a cluster ID
This class enforces type checking and value constraints, and converts
between the internal object model and the API representation of a cluster
ID.
"""
uuid = types.uuid
"""Unique UUID for this cluster"""
def __init__(self, uuid):
self.uuid = uuid
class Cluster(base.APIBase):
"""API representation of a cluster.
This class enforces type checking and value constraints, and converts
between the internal object model and the API representation of a Cluster.
"""
_cluster_template_id = None
def _get_cluster_template_id(self):
return self._cluster_template_id
def _set_cluster_template_id(self, value):
if value and self._cluster_template_id != value:
try:
cluster_template = api_utils.get_resource('ClusterTemplate',
value)
self._cluster_template_id = cluster_template.uuid
except exception.ClusterTemplateNotFound as e:
# Change error code because 404 (NotFound) is inappropriate
# response for a POST request to create a Cluster
e.code = 400 # BadRequest
raise
elif value == wtypes.Unset:
self._cluster_template_id = wtypes.Unset
uuid = types.uuid
"""Unique UUID for this cluster"""
name = wtypes.StringType(min_length=1, max_length=242,
pattern='^[a-zA-Z][a-zA-Z0-9_.-]*$')
"""Name of this cluster, max length is limited to 242 because of heat
stack requires max length limit to 255, and Magnum amend a uuid length"""
cluster_template_id = wsme.wsproperty(wtypes.text,
_get_cluster_template_id,
_set_cluster_template_id,
mandatory=True)
"""The cluster_template UUID"""
keypair = wsme.wsattr(wtypes.StringType(min_length=1, max_length=255),
default=None)
"""The name of the nova ssh keypair"""
node_count = wsme.wsattr(wtypes.IntegerType(minimum=0), default=1)
"""The node count for this cluster. Default to 1 if not set"""
master_count = wsme.wsattr(wtypes.IntegerType(minimum=1), default=1)
"""The number of master nodes for this cluster. Default to 1 if not set"""
docker_volume_size = wtypes.IntegerType(minimum=1)
"""The size in GB of the docker volume"""
labels = wtypes.DictType(wtypes.text, types.MultiType(wtypes.text,
six.integer_types,
bool,
float))
"""One or more key/value pairs"""
master_flavor_id = wtypes.StringType(min_length=1, max_length=255)
"""The flavor of the master node for this Cluster"""
flavor_id = wtypes.StringType(min_length=1, max_length=255)
"""The flavor of this Cluster"""
create_timeout = wsme.wsattr(wtypes.IntegerType(minimum=0), default=60)
"""Timeout for creating the cluster in minutes. Default to 60 if not set"""
links = wsme.wsattr([link.Link], readonly=True)
"""A list containing a self link and associated cluster links"""
stack_id = wsme.wsattr(wtypes.text, readonly=True)
"""Stack id of the heat stack"""
status = wtypes.Enum(wtypes.text, *fields.ClusterStatus.ALL)
"""Status of the cluster from the heat stack"""
status_reason = wtypes.text
"""Status reason of the cluster from the heat stack"""
health_status = wtypes.Enum(wtypes.text, *fields.ClusterHealthStatus.ALL)
"""Health status of the cluster from the native COE API"""
health_status_reason = wtypes.DictType(wtypes.text, wtypes.text)
"""Health status reason of the cluster from the native COE API"""
discovery_url = wtypes.text
"""Url used for cluster node discovery"""
api_address = wsme.wsattr(wtypes.text, readonly=True)
"""Api address of cluster master node"""
coe_version = wsme.wsattr(wtypes.text, readonly=True)
"""Version of the COE software currently running in this cluster.
Example: swarm version or kubernetes version."""
container_version = wsme.wsattr(wtypes.text, readonly=True)
"""Version of the container software. Example: docker version."""
project_id = wsme.wsattr(wtypes.text, readonly=True)
"""Project id of the cluster belongs to"""
user_id = wsme.wsattr(wtypes.text, readonly=True)
"""User id of the cluster belongs to"""
node_addresses = wsme.wsattr([wtypes.text], readonly=True)
"""IP addresses of cluster slave nodes"""
master_addresses = wsme.wsattr([wtypes.text], readonly=True)
"""IP addresses of cluster master nodes"""
faults = wsme.wsattr(wtypes.DictType(wtypes.text, wtypes.text))
"""Fault info collected from the heat resources of this cluster"""
fixed_network = wtypes.StringType(min_length=1, max_length=255)
"""The fixed network name to attach to the Cluster"""
fixed_subnet = wtypes.StringType(min_length=1, max_length=255)
"""The fixed subnet name to attach to the Cluster"""
floating_ip_enabled = wsme.wsattr(types.boolean)
"""Indicates whether created clusters should have a floating ip or not."""
merge_labels = wsme.wsattr(types.boolean, default=False)
"""Indicates whether the labels will be merged with the CT labels."""
labels_overridden = wtypes.DictType(
wtypes.text, types.MultiType(
wtypes.text, six.integer_types, bool, float))
"""Contains labels that have a value different than the parent labels."""
labels_added = wtypes.DictType(
wtypes.text, types.MultiType(
wtypes.text, six.integer_types, bool, float))
"""Contains labels that do not exist in the parent."""
labels_skipped = wtypes.DictType(
wtypes.text, types.MultiType(
wtypes.text, six.integer_types, bool, float))
"""Contains labels that exist in the parent but were not inherited."""
master_lb_enabled = wsme.wsattr(types.boolean)
"""Indicates whether created clusters should have a load balancer for master
nodes or not.
"""
def __init__(self, **kwargs):
super(Cluster, self).__init__()
self.fields = []
for field in objects.Cluster.fields:
# Skip fields we do not expose.
if not hasattr(self, field):
continue
self.fields.append(field)
setattr(self, field, kwargs.get(field, wtypes.Unset))
nodegroup_fields = ['node_count', 'master_count',
'node_addresses', 'master_addresses']
for field in nodegroup_fields:
self.fields.append(field)
setattr(self, field, kwargs.get(field, wtypes.Unset))
@staticmethod
def _convert_with_links(cluster, url, expand=True, parent_labels=None):
if not expand:
cluster.unset_fields_except(['uuid', 'name', 'cluster_template_id',
'keypair', 'docker_volume_size',
'labels', 'node_count', 'status',
'master_flavor_id', 'flavor_id',
'create_timeout', 'master_count',
'stack_id', 'health_status'])
else:
overridden, added, skipped = api_utils.get_labels_diff(
parent_labels, cluster.labels)
cluster.labels_overridden = overridden
cluster.labels_added = added
cluster.labels_skipped = skipped
cluster.links = [link.Link.make_link('self', url,
'clusters', cluster.uuid),
link.Link.make_link('bookmark', url,
'clusters', cluster.uuid,
bookmark=True)]
return cluster
@classmethod
def convert_with_links(cls, rpc_cluster, expand=True):
cluster = Cluster(**rpc_cluster.as_dict())
parent_labels = rpc_cluster.cluster_template.labels
return cls._convert_with_links(cluster, pecan.request.host_url, expand,
parent_labels)
@classmethod
def sample(cls, expand=True):
temp_id = '4a96ac4b-2447-43f1-8ca6-9fd6f36d146d'
sample = cls(uuid='27e3153e-d5bf-4b7e-b517-fb518e17f34c',
name='example',
cluster_template_id=temp_id,
keypair=None,
node_count=2,
master_count=1,
docker_volume_size=1,
labels={},
master_flavor_id='m1.small',
flavor_id='m1.small',
create_timeout=15,
stack_id='49dc23f5-ffc9-40c3-9d34-7be7f9e34d63',
status=fields.ClusterStatus.CREATE_COMPLETE,
status_reason="CREATE completed successfully",
health_status=fields.ClusterHealthStatus.HEALTHY,
health_status_reason={"api": "ok",
"node-0.Ready": 'True'},
api_address='172.24.4.3',
node_addresses=['172.24.4.4', '172.24.4.5'],
created_at=timeutils.utcnow(),
updated_at=timeutils.utcnow(),
coe_version=None,
container_version=None,
fixed_network=None,
fixed_subnet=None,
floating_ip_enabled=True,
master_lb_enabled=True)
return cls._convert_with_links(sample, 'http://localhost:9511', expand)
class ClusterPatchType(types.JsonPatchType):
_api_base = Cluster
@staticmethod
def internal_attrs():
internal_attrs = ['/api_address', '/node_addresses',
'/master_addresses', '/stack_id',
'/ca_cert_ref', '/magnum_cert_ref',
'/trust_id', '/trustee_user_name',
'/trustee_password', '/trustee_user_id',
'/etcd_ca_cert_ref', '/front_proxy_ca_cert_ref']
return types.JsonPatchType.internal_attrs() + internal_attrs
class ClusterCollection(collection.Collection):
"""API representation of a collection of clusters."""
clusters = [Cluster]
"""A list containing cluster objects"""
def __init__(self, **kwargs):
self._type = 'clusters'
@staticmethod
def convert_with_links(rpc_clusters, limit, url=None, expand=False,
**kwargs):
collection = ClusterCollection()
collection.clusters = [Cluster.convert_with_links(p, expand)
for p in rpc_clusters]
collection.next = collection.get_next(limit, url=url, **kwargs)
return collection
@classmethod
def sample(cls):
sample = cls()
sample.clusters = [Cluster.sample(expand=False)]
return sample
class ClustersController(base.Controller):
"""REST controller for Clusters."""
def __init__(self):
super(ClustersController, self).__init__()
_custom_actions = {
'detail': ['GET'],
}
_in_tree_cinder_volume_driver_deprecation_note = (
"The in-tree Cinder volume driver is deprecated and will be removed "
"in X cycle in favour of out-of-tree Cinder CSI driver which requires "
"the label cinder_csi_enabled set to True (default behaviour from "
"V cycle) when volume_driver is cinder.")
actions = cluster_actions.ActionsController()
def _generate_name_for_cluster(self, context):
"""Generate a random name like: zeta-22-cluster."""
name_gen = name_generator.NameGenerator()
name = name_gen.generate()
return name + '-cluster'
def _get_clusters_collection(self, marker, limit,
sort_key, sort_dir, expand=False,
resource_url=None):
context = pecan.request.context
if context.is_admin:
if expand:
policy.enforce(context, "cluster:detail_all_projects",
action="cluster:detail_all_projects")
else:
policy.enforce(context, "cluster:get_all_all_projects",
action="cluster:get_all_all_projects")
# TODO(flwang): Instead of asking an extra 'all_project's
# parameter, currently the design is allowing admin user to list
# all clusters from all projects. But the all_tenants is one of
# the condition to do project filter in DB API. And it's also used
# by periodic tasks. So the could be removed in the future and
# a new parameter 'project_id' would be added so that admin user
# can list clusters for a particular project.
context.all_tenants = True
limit = api_utils.validate_limit(limit)
sort_dir = api_utils.validate_sort_dir(sort_dir)
marker_obj = None
if marker:
marker_obj = objects.Cluster.get_by_uuid(pecan.request.context,
marker)
clusters = objects.Cluster.list(pecan.request.context, limit,
marker_obj, sort_key=sort_key,
sort_dir=sort_dir)
return ClusterCollection.convert_with_links(clusters, limit,
url=resource_url,
expand=expand,
sort_key=sort_key,
sort_dir=sort_dir)
nodegroups = nodegroup.NodeGroupController()
@expose.expose(ClusterCollection, types.uuid, int, wtypes.text,
wtypes.text)
def get_all(self, marker=None, limit=None, sort_key='id',
sort_dir='asc'):
"""Retrieve a list of clusters.
:param marker: pagination marker for large data sets.
:param limit: maximum number of resources to return in a single result.
:param sort_key: column to sort results by. Default: id.
:param sort_dir: direction to sort. "asc" or "desc". Default: asc.
"""
context = pecan.request.context
policy.enforce(context, 'cluster:get_all',
action='cluster:get_all')
return self._get_clusters_collection(marker, limit, sort_key,
sort_dir)
@expose.expose(ClusterCollection, types.uuid, int, wtypes.text,
wtypes.text)
def detail(self, marker=None, limit=None, sort_key='id',
sort_dir='asc'):
"""Retrieve a list of clusters with detail.
:param marker: pagination marker for large data sets.
:param limit: maximum number of resources to return in a single result.
:param sort_key: column to sort results by. Default: id.
:param sort_dir: direction to sort. "asc" or "desc". Default: asc.
"""
context = pecan.request.context
policy.enforce(context, 'cluster:detail',
action='cluster:detail')
# NOTE(lucasagomes): /detail should only work against collections
parent = pecan.request.path.split('/')[:-1][-1]
if parent != "clusters":
raise exception.HTTPNotFound
expand = True
resource_url = '/'.join(['clusters', 'detail'])
return self._get_clusters_collection(marker, limit,
sort_key, sort_dir, expand,
resource_url)
def _collect_fault_info(self, context, cluster):
"""Collect fault info from heat resources of given cluster
and store them into cluster.faults.
"""
# Gather fault info from the cluster nodegroups.
return {
ng.name: ng.status_reason for ng in cluster.nodegroups
if ng.status.endswith('FAILED')
}
@expose.expose(Cluster, types.uuid_or_name)
def get_one(self, cluster_ident):
"""Retrieve information about the given Cluster.
:param cluster_ident: UUID or logical name of the Cluster.
"""
context = pecan.request.context
if context.is_admin:
policy.enforce(context, "cluster:get_one_all_projects",
action="cluster:get_one_all_projects")
# TODO(flwang): Instead of asking an extra 'all_project's
# parameter, currently the design is allowing admin user to list
# all clusters from all projects. But the all_tenants is one of
# the condition to do project filter in DB API. And it's also used
# by periodic tasks. So this could be removed in the future and
# a new parameter 'project_id' would be added so that admin user
# can list clusters for a particular project.
context.all_tenants = True
cluster = api_utils.get_resource('Cluster', cluster_ident)
policy.enforce(context, 'cluster:get', cluster.as_dict(),
action='cluster:get')
api_cluster = Cluster.convert_with_links(cluster)
if api_cluster.status in fields.ClusterStatus.STATUS_FAILED:
api_cluster.faults = self._collect_fault_info(context, cluster)
return api_cluster
def _check_cluster_quota_limit(self, context):
try:
# Check if there is any explicit quota limit set in Quotas table
quota = objects.Quota.get_quota_by_project_id_resource(
context,
context.project_id,
'Cluster')
cluster_limit = quota.hard_limit
except exception.QuotaNotFound:
# If explicit quota was not set for the project, use default limit
cluster_limit = CONF.quotas.max_clusters_per_project
if objects.Cluster.get_count_all(context) >= cluster_limit:
msg = _("You have reached the maximum clusters per project, "
"%d. You may delete a cluster to make room for a new "
"one.") % cluster_limit
raise exception.ResourceLimitExceeded(msg=msg)
@base.Controller.api_version("1.1", "1.9")
@expose.expose(ClusterID, body=Cluster, status_code=202)
@validation.enforce_cluster_type_supported()
@validation.enforce_cluster_volume_storage_size()
def post(self, cluster):
if cluster.node_count == 0:
raise exception.ZeroNodeCountNotSupported()
return self._post(cluster)
@base.Controller.api_version("1.10") # noqa
@expose.expose(ClusterID, body=Cluster, status_code=202)
@validation.enforce_cluster_type_supported()
@validation.enforce_cluster_volume_storage_size()
def post(self, cluster): # noqa
return self._post(cluster)
def _post(self, cluster):
"""Create a new cluster.
:param cluster: a cluster within the request body.
"""
context = pecan.request.context
policy.enforce(context, 'cluster:create',
action='cluster:create')
self._check_cluster_quota_limit(context)
temp_id = cluster.cluster_template_id
cluster_template = objects.ClusterTemplate.get_by_uuid(context,
temp_id)
# If keypair not present, use cluster_template value
if cluster.keypair is None:
cluster.keypair = cluster_template.keypair_id
# If labels is not present, use cluster_template value
if cluster.labels == wtypes.Unset or not cluster.labels:
cluster.labels = cluster_template.labels
else:
# If labels are provided check if the user wishes to merge
# them with the values from the cluster template.
if cluster.merge_labels:
labels = cluster_template.labels
labels.update(cluster.labels)
cluster.labels = labels
cinder_csi_enabled = cluster.labels.get('cinder_csi_enabled', True)
if (cluster_template.volume_driver == 'cinder' and
not strutils.bool_from_string(cinder_csi_enabled)):
warnings.warn(self._in_tree_cinder_volume_driver_deprecation_note,
DeprecationWarning)
LOG.warning(self._in_tree_cinder_volume_driver_deprecation_note)
# If floating_ip_enabled is not present, use cluster_template value
if cluster.floating_ip_enabled == wtypes.Unset:
cluster.floating_ip_enabled = cluster_template.floating_ip_enabled
# If master_lb_enabled is not present, use cluster_template value
if cluster.master_lb_enabled == wtypes.Unset:
cluster.master_lb_enabled = cluster_template.master_lb_enabled
attributes = ["docker_volume_size", "master_flavor_id", "flavor_id",
"fixed_network", "fixed_subnet"]
for attr in attributes:
if (getattr(cluster, attr) == wtypes.Unset or
not getattr(cluster, attr)):
setattr(cluster, attr, getattr(cluster_template, attr))
cluster_dict = cluster.as_dict()
attr_validator.validate_os_resources(context,
cluster_template.as_dict(),
cluster_dict)
attr_validator.validate_master_count(cluster_dict,
cluster_template.as_dict())
cluster_dict['project_id'] = context.project_id
cluster_dict['user_id'] = context.user_id
# NOTE(yuywz): We will generate a random human-readable name for
# cluster if the name is not specified by user.
name = cluster_dict.get('name') or \
self._generate_name_for_cluster(context)
cluster_dict['name'] = name
cluster_dict['coe_version'] = None
cluster_dict['container_version'] = None
node_count = cluster_dict.pop('node_count')
master_count = cluster_dict.pop('master_count')
new_cluster = objects.Cluster(context, **cluster_dict)
new_cluster.uuid = uuid.uuid4()
pecan.request.rpcapi.cluster_create_async(new_cluster,
master_count, node_count,
cluster.create_timeout)
return ClusterID(new_cluster.uuid)
@base.Controller.api_version("1.1", "1.2")
@wsme.validate(types.uuid, [ClusterPatchType])
@expose.expose(ClusterID, types.uuid_or_name, body=[ClusterPatchType],
status_code=202)
def patch(self, cluster_ident, patch):
"""Update an existing Cluster.
:param cluster_ident: UUID or logical name of a cluster.
:param patch: a json PATCH document to apply to this cluster.
"""
(cluster, node_count,
health_status,
health_status_reason) = self._patch(cluster_ident, patch)
if node_count == 0:
raise exception.ZeroNodeCountNotSupported()
pecan.request.rpcapi.cluster_update_async(cluster, node_count,
health_status,
health_status_reason)
return ClusterID(cluster.uuid)
@base.Controller.api_version("1.3", "1.9") # noqa
@wsme.validate(types.uuid, bool, [ClusterPatchType])
@expose.expose(ClusterID, types.uuid_or_name, types.boolean,
body=[ClusterPatchType], status_code=202)
def patch(self, cluster_ident, rollback=False, patch=None): # noqa
"""Update an existing Cluster.
:param cluster_ident: UUID or logical name of a cluster.
:param rollback: whether to rollback cluster on update failure.
:param patch: a json PATCH document to apply to this cluster.
"""
(cluster, node_count,
health_status,
health_status_reason) = self._patch(cluster_ident, patch)
if node_count == 0:
raise exception.ZeroNodeCountNotSupported()
pecan.request.rpcapi.cluster_update_async(cluster, node_count,
health_status,
health_status_reason,
rollback)
return ClusterID(cluster.uuid)
@base.Controller.api_version("1.10") # noqa
@wsme.validate(types.uuid, bool, [ClusterPatchType])
@expose.expose(ClusterID, types.uuid_or_name, types.boolean,
body=[ClusterPatchType], status_code=202)
def patch(self, cluster_ident, rollback=False, patch=None): # noqa
"""Update an existing Cluster.
:param cluster_ident: UUID or logical name of a cluster.
:param rollback: whether to rollback cluster on update failure.
:param patch: a json PATCH document to apply to this cluster.
"""
(cluster, node_count,
health_status,
health_status_reason) = self._patch(cluster_ident, patch)
pecan.request.rpcapi.cluster_update_async(cluster, node_count,
health_status,
health_status_reason,
rollback)
return ClusterID(cluster.uuid)
def _patch(self, cluster_ident, patch):
context = pecan.request.context
if context.is_admin:
policy.enforce(context, "cluster:update_all_projects",
action="cluster:update_all_projects")
context.all_tenants = True
cluster = api_utils.get_resource('Cluster', cluster_ident)
policy.enforce(context, 'cluster:update', cluster.as_dict(),
action='cluster:update')
policy.enforce(context, "cluster:update_health_status",
action="cluster:update_health_status")
try:
cluster_dict = cluster.as_dict()
new_cluster = Cluster(**api_utils.apply_jsonpatch(cluster_dict,
patch))
except api_utils.JSONPATCH_EXCEPTIONS as e:
raise exception.PatchError(patch=patch, reason=e)
# NOTE(ttsiouts): magnum.objects.Cluster.node_count will be a
# property so we won't be able to store it in the object. So
# instead of object_what_changed compare the new and the old
# clusters.
delta = set()
for field in new_cluster.fields:
if getattr(cluster, field) != getattr(new_cluster, field):
delta.add(field)
validation.validate_cluster_properties(delta)
# NOTE(brtknr): cluster.node_count is the size of the whole cluster
# which includes non-default nodegroups. However cluster_update expects
# node_count to be the size of the default_ng_worker therefore return
# this value unless the patch object says otherwise.
node_count = cluster.default_ng_worker.node_count
for p in patch:
if p['path'] == '/node_count':
node_count = p.get('value') or new_cluster.node_count
return (cluster, node_count,
new_cluster.health_status, new_cluster.health_status_reason)
@expose.expose(None, types.uuid_or_name, status_code=204)
def delete(self, cluster_ident):
"""Delete a cluster.
:param cluster_ident: UUID of cluster or logical name of the cluster.
"""
context = pecan.request.context
if context.is_admin:
policy.enforce(context, 'cluster:delete_all_projects',
action='cluster:delete_all_projects')
context.all_tenants = True
cluster = api_utils.get_resource('Cluster', cluster_ident)
policy.enforce(context, 'cluster:delete', cluster.as_dict(),
action='cluster:delete')
pecan.request.rpcapi.cluster_delete_async(cluster.uuid)