def run(self):
super(FakeDeletionThread, self).run()
receiver = NailgunReceiver
kwargs = {
'task_uuid': self.task_uuid,
'nodes': self.data['args']['nodes'],
'status': 'ready'
}
nodes_to_restore = self.data['args'].get('nodes_to_restore', [])
resp_method = getattr(receiver, self.respond_to)
resp_method(**kwargs)
for node_data in nodes_to_restore:
node = Node(**node_data)
# Offline node just deleted from db
# and could not recreated with status
# discover
if not node.online:
continue
node.status = 'discover'
db().add(node)
db().commit()
node.attributes = NodeAttributes(node_id=node.id)
node.attributes.volumes = node.volume_manager.gen_volumes_info()
NetworkManager.update_interfaces_info(node)
db().commit()
ram = round(node.meta.get('ram') or 0, 1)
cores = node.meta.get('cores') or 'unknown'
notifier.notify("discover",
"New node with %s CPU core(s) "
"and %s GB memory is discovered" %
(cores, ram), node_id=node.id)
def test_session_update(self):
node = Node()
node.mac = u"ASDFGHJKLMNOPR"
node.timestamp = datetime.now()
self.db.add(node)
self.db.commit()
node2 = self.db2.query(Node).filter(Node.id == node.id).first()
node2.mac = u"12345678"
self.db2.add(node2)
self.db2.commit()
self.db.query(Node).filter(Node.id == node.id).first()
self.assertEqual(node.mac, u"12345678")
def test_session_update(self):
node = Node()
node.mac = "aa:bb:cc:dd:ff:11"
node.timestamp = datetime.now()
self.db.add(node)
self.db.commit()
node2 = self.db2.query(Node).filter(Node.id == node.id).first()
node2.mac = "aa:bb:cc:dd:ff:11"
self.db2.add(node2)
self.db2.commit()
self.db.query(Node).filter(Node.id == node.id).first()
self.assertEqual(node.mac, "aa:bb:cc:dd:ff:11")
def test_session_update(self):
node = Node()
node.mac = u"ASDFGHJKLMNOPR"
node.timestamp = datetime.now()
self.db.add(node)
self.db.commit()
node2 = self.db2.query(Node).filter(
Node.id == node.id
).first()
node2.mac = u"12345678"
self.db2.add(node2)
self.db2.commit()
self.db.query(Node).filter(
Node.id == node.id
).first()
self.assertEqual(node.mac, u"12345678")
def test_session_update(self):
node = Node()
node.mac = "aa:bb:cc:dd:ff:11"
node.timestamp = datetime.now()
self.db.add(node)
self.db.commit()
node2 = self.db2.query(Node).filter(
Node.id == node.id
).first()
node2.mac = "aa:bb:cc:dd:ff:11"
self.db2.add(node2)
self.db2.commit()
self.db.query(Node).filter(
Node.id == node.id
).first()
self.assertEqual(node.mac, "aa:bb:cc:dd:ff:11")
def execute(self, nodes, mclient_remove=True):
cluster_id = None
if hasattr(self, 'cluster'):
cluster_id = self.cluster.id
objects.TaskCollection.lock_cluster_tasks(cluster_id)
logger.info("Trying to execute node deletion task with nodes %s",
', '.join(str(node.id) for node in nodes))
self.verify_nodes_with_cluster(nodes)
objects.NodeCollection.lock_nodes(nodes)
if cluster_id is None:
# DeletionTask operates on cluster's nodes.
# Nodes that are not in cluster are simply deleted.
Node.delete_by_ids([n.id for n in nodes])
db().flush()
task = Task(name=consts.TASK_NAMES.node_deletion,
progress=100,
status=consts.TASK_STATUSES.ready)
db().add(task)
db().flush()
return task
task = Task(name=consts.TASK_NAMES.node_deletion,
cluster=self.cluster)
db().add(task)
for node in nodes:
objects.Node.update(node,
{'status': consts.NODE_STATUSES.removing})
db().flush()
self._call_silently(
task,
tasks.DeletionTask,
nodes=tasks.DeletionTask.prepare_nodes_for_task(
nodes, mclient_remove=mclient_remove))
return task
def execute(self, nodes, mclient_remove=True):
cluster_id = None
if hasattr(self, 'cluster'):
cluster_id = self.cluster.id
objects.TaskCollection.lock_cluster_tasks(cluster_id)
logger.info("Trying to execute node deletion task with nodes %s",
', '.join(str(node.id) for node in nodes))
self.verify_nodes_with_cluster(nodes)
objects.NodeCollection.lock_nodes(nodes)
if cluster_id is None:
# DeletionTask operates on cluster's nodes.
# Nodes that are not in cluster are simply deleted.
Node.delete_by_ids([n.id for n in nodes])
db().flush()
task = Task(name=consts.TASK_NAMES.node_deletion,
progress=100,
status=consts.TASK_STATUSES.ready)
db().add(task)
db().flush()
return task
task = Task(name=consts.TASK_NAMES.node_deletion, cluster=self.cluster)
db().add(task)
for node in nodes:
objects.Node.update(node,
{'status': consts.NODE_STATUSES.removing})
db().flush()
self._call_silently(task,
tasks.DeletionTask,
nodes=tasks.DeletionTask.prepare_nodes_for_task(
nodes, mclient_remove=mclient_remove))
return task
def remove_nodes_resp(cls, **kwargs):
logger.info(
"RPC method remove_nodes_resp received: %s" %
jsonutils.dumps(kwargs)
)
task_uuid = kwargs.get('task_uuid')
nodes = kwargs.get('nodes') or []
error_nodes = kwargs.get('error_nodes') or []
inaccessible_nodes = kwargs.get('inaccessible_nodes') or []
error_msg = kwargs.get('error')
status = kwargs.get('status')
progress = kwargs.get('progress')
if status in [consts.TASK_STATUSES.ready, consts.TASK_STATUSES.error]:
progress = 100
# locking tasks on cluster
task = objects.Task.get_by_uuid(task_uuid, fail_if_not_found=True)
objects.TaskCollection.lock_cluster_tasks(task.cluster_id)
task = objects.Task.get_by_uuid(
task_uuid,
fail_if_not_found=True,
lock_for_update=True
)
# locking cluster
if task.cluster_id is not None:
objects.Cluster.get_by_uid(
task.cluster_id,
fail_if_not_found=True,
lock_for_update=True
)
# locking nodes
all_nodes = itertools.chain(nodes, error_nodes, inaccessible_nodes)
all_nodes_ids = [
node['id'] if 'id' in node else node['uid']
for node in all_nodes
]
locked_nodes = objects.NodeCollection.filter_by_list(
None,
'id',
all_nodes_ids,
order_by='id'
)
objects.NodeCollection.lock_for_update(locked_nodes).all()
def get_node_id(n):
return n.get('id', int(n.get('uid')))
Node.delete_by_ids([get_node_id(n) for n in nodes])
if(len(inaccessible_nodes) > 0):
inaccessible_node_ids = [
get_node_id(n) for n in inaccessible_nodes]
logger.warn(u'Nodes %s not answered by RPC, removing from db',
inaccessible_nodes)
Node.delete_by_ids(inaccessible_node_ids)
for node in error_nodes:
node_db = objects.Node.get_by_uid(node['uid'])
if not node_db:
logger.error(
u"Failed to delete node '%s' marked as error from Astute:"
" node doesn't exist", str(node)
)
else:
node_db.pending_deletion = False
node_db.status = 'error'
db().add(node_db)
node['name'] = node_db.name
db().flush()
success_msg = u"No nodes were removed"
err_msg = u"No errors occurred"
if nodes:
success_msg = u"Successfully removed {0} node(s)".format(
len(nodes)
)
notifier.notify("done", success_msg)
if error_nodes:
err_msg = u"Failed to remove {0} node(s): {1}".format(
len(error_nodes),
', '.join(
[n.get('name') or "ID: {0}".format(n['uid'])
for n in error_nodes])
)
notifier.notify("error", err_msg)
if not error_msg:
error_msg = ". ".join([success_msg, err_msg])
data = {
'status': status,
'progress': progress,
'message': error_msg,
}
objects.Task.update(task, data)
cls._update_action_log_entry(status, task.name, task_uuid, nodes)
def POST(self):
""":returns: JSONized Node object.
:http: * 201 (cluster successfully created)
* 400 (invalid node data specified)
* 403 (node has incorrect status)
* 409 (node with such parameters already exists)
"""
data = self.checked_data()
if data.get("status", "") != "discover":
error = web.forbidden()
error.data = "Only bootstrap nodes are allowed to be registered."
msg = u"Node with mac '{0}' was not created, " \
u"because request status is '{1}'."\
.format(data[u'mac'], data.get(u'status'))
logger.warning(msg)
raise error
node = Node(
name="Untitled (%s)" % data['mac'][-5:],
timestamp=datetime.now()
)
if "cluster_id" in data:
# FIXME(vk): this part is needed only for tests. Normally,
# nodes are created only by agent and POST requests don't contain
# cluster_id, but our integration and unit tests widely use it.
# We need to assign cluster first
cluster_id = data.pop("cluster_id")
if cluster_id:
node.cluster = db.query(Cluster).get(cluster_id)
for key, value in data.iteritems():
if key == "id":
continue
elif key == "meta":
node.create_meta(value)
else:
setattr(node, key, value)
db().add(node)
db().commit()
node.attributes = NodeAttributes()
try:
node.attributes.volumes = node.volume_manager.gen_volumes_info()
if node.cluster:
node.cluster.add_pending_changes(
"disks",
node_id=node.id
)
except Exception as exc:
msg = (
u"Failed to generate volumes "
"info for node '{0}': '{1}'"
).format(
node.name or data.get("mac") or data.get("id"),
str(exc) or "see logs for details"
)
logger.warning(traceback.format_exc())
notifier.notify("error", msg, node_id=node.id)
db().add(node)
db().commit()
network_manager = NetworkManager
# Add interfaces for node from 'meta'.
if node.meta and node.meta.get('interfaces'):
network_manager.update_interfaces_info(node)
if node.cluster_id:
network_manager = node.cluster.network_manager
network_manager.assign_networks_by_default(node)
network_manager.allow_network_assignment_to_all_interfaces(node)
try:
# we use multiplier of 1024 because there are no problems here
# with unfair size calculation
ram = str(round(float(
node.meta['memory']['total']) / 1073741824, 1)) + " GB RAM"
except Exception as exc:
logger.warning(traceback.format_exc())
ram = "unknown RAM"
try:
# we use multiplier of 1000 because disk vendors specify HDD size
# in terms of decimal capacity. Sources:
# http://knowledge.seagate.com/articles/en_US/FAQ/172191en
# http://physics.nist.gov/cuu/Units/binary.html
hd_size = round(float(
sum([d["size"] for d in node.meta["disks"]]) / 1000000000), 1)
# if HDD > 100 GB we show it's size in TB
if hd_size > 100:
hd_size = str(hd_size / 1000) + " TB HDD"
else:
hd_size = str(hd_size) + " GB HDD"
except Exception as exc:
logger.warning(traceback.format_exc())
hd_size = "unknown HDD"
cores = str(node.meta.get('cpu', {}).get('total', "unknown"))
notifier.notify(
"discover",
"New node is discovered: %s CPUs / %s / %s " %
(cores, ram, hd_size),
node_id=node.id
)
raise web.webapi.created(json.dumps(
NodeHandler.render(node),
indent=4
))
def create_node(self,
api=False,
exclude=None,
expect_http=201,
expect_message=None,
**kwargs):
#TODO(alekseyk) Simplify 'interfaces' and 'mac' manipulation logic
metadata = kwargs.get('meta')
default_metadata = self.default_metadata()
if metadata:
default_metadata.update(metadata)
meta_ifaces = 'interfaces' in metadata
mac = kwargs.get('mac', self._generate_random_mac())
if default_metadata['interfaces']:
default_metadata['interfaces'][0]['mac'] = mac
if not metadata or not meta_ifaces:
for iface in default_metadata['interfaces'][1:]:
if 'mac' in iface:
iface['mac'] = self._generate_random_mac()
node_data = {
'mac': mac,
'status': 'discover',
'meta': default_metadata
}
if kwargs:
meta = kwargs.pop('meta', None)
node_data.update(kwargs)
if meta:
kwargs['meta'] = meta
if exclude and isinstance(exclude, list):
for ex in exclude:
try:
del node_data[ex]
except KeyError as err:
logger.warning(err)
if api:
resp = self.app.post(reverse('NodeCollectionHandler'),
json.dumps(node_data),
headers=self.default_headers,
expect_errors=True)
self.tester.assertEquals(resp.status, expect_http)
if expect_message:
self.tester.assertEquals(resp.body, expect_message)
if str(expect_http)[0] != "2":
return None
self.tester.assertEquals(resp.status, expect_http)
node = json.loads(resp.body)
node_db = self.db.query(Node).get(node['id'])
if 'interfaces' not in node_data['meta'] \
or not node_data['meta']['interfaces']:
self._set_interfaces_if_not_set_in_meta(
node_db.id, kwargs.get('meta', None))
self.nodes.append(node_db)
else:
node = Node()
node.timestamp = datetime.now()
if 'cluster_id' in node_data:
cluster_id = node_data.pop('cluster_id')
for cluster in self.clusters:
if cluster.id == cluster_id:
node.cluster = cluster
break
else:
node.cluster_id = cluster_id
for key, value in node_data.iteritems():
setattr(node, key, value)
node.attributes = self.create_attributes()
node.attributes.volumes = node.volume_manager.gen_volumes_info()
self.db.add(node)
self.db.commit()
if node.meta and node.meta.get('interfaces'):
self._create_interfaces_from_meta(node)
self.nodes.append(node)
return node
def create_node(
self, api=False,
exclude=None, expect_http=201,
expect_message=None,
**kwargs):
metadata = kwargs.get('meta')
default_metadata = self.default_metadata()
if metadata:
default_metadata.update(metadata)
mac = self._generate_random_mac()
if default_metadata['interfaces']:
default_metadata['interfaces'][0]['mac'] = kwargs.get('mac', mac)
node_data = {
'mac': mac,
'status': 'discover',
'meta': default_metadata
}
if kwargs:
meta = kwargs.pop('meta', None)
node_data.update(kwargs)
if meta:
kwargs['meta'] = meta
if exclude and isinstance(exclude, list):
for ex in exclude:
try:
del node_data[ex]
except KeyError as err:
logger.warning(err)
if api:
resp = self.app.post(
reverse('NodeCollectionHandler'),
json.dumps(node_data),
headers=self.default_headers,
expect_errors=True
)
self.tester.assertEquals(resp.status, expect_http)
if expect_message:
self.tester.assertEquals(resp.body, expect_message)
if str(expect_http)[0] != "2":
return None
self.tester.assertEquals(resp.status, expect_http)
node = json.loads(resp.body)
node_db = self.db.query(Node).get(node['id'])
self._set_interfaces_if_not_set_in_meta(
node_db.id,
kwargs.get('meta', None))
self.nodes.append(node_db)
else:
node = Node()
node.timestamp = datetime.now()
if 'cluster_id' in node_data:
cluster_id = node_data.pop('cluster_id')
for cluster in self.clusters:
if cluster.id == cluster_id:
node.cluster = cluster
break
else:
node.cluster_id = cluster_id
for key, value in node_data.iteritems():
setattr(node, key, value)
node.attributes = self.create_attributes()
node.attributes.volumes = node.volume_manager.gen_volumes_info()
self.db.add(node)
self.db.commit()
if node.meta and node.meta.get('interfaces'):
self._create_interfaces_from_meta(node)
self.nodes.append(node)
return node
def remove_nodes_resp(cls, **kwargs):
logger.info(
"RPC method remove_nodes_resp received: %s" %
jsonutils.dumps(kwargs)
)
task_uuid = kwargs.get('task_uuid')
nodes = kwargs.get('nodes') or []
error_nodes = kwargs.get('error_nodes') or []
inaccessible_nodes = kwargs.get('inaccessible_nodes') or []
error_msg = kwargs.get('error')
status = kwargs.get('status')
progress = kwargs.get('progress')
if status in [consts.TASK_STATUSES.ready, consts.TASK_STATUSES.error]:
progress = 100
# locking tasks on cluster
task = objects.Task.get_by_uuid(task_uuid, fail_if_not_found=True)
objects.TaskCollection.lock_cluster_tasks(task.cluster_id)
task = objects.Task.get_by_uuid(
task_uuid,
fail_if_not_found=True,
lock_for_update=True
)
# locking cluster
if task.cluster_id is not None:
objects.Cluster.get_by_uid(
task.cluster_id,
fail_if_not_found=True,
lock_for_update=True
)
# locking nodes
all_nodes = itertools.chain(nodes, error_nodes, inaccessible_nodes)
all_nodes_ids = [
node['id'] if 'id' in node else node['uid']
for node in all_nodes
]
locked_nodes = objects.NodeCollection.filter_by_list(
None,
'id',
all_nodes_ids,
order_by='id'
)
objects.NodeCollection.lock_for_update(locked_nodes).all()
def get_node_id(n):
return n.get('id', int(n.get('uid')))
Node.delete_by_ids([get_node_id(n) for n in nodes])
if(len(inaccessible_nodes) > 0):
inaccessible_node_ids = [
get_node_id(n) for n in inaccessible_nodes]
logger.warn(u'Nodes %s not answered by RPC, removing from db',
inaccessible_nodes)
Node.delete_by_ids(inaccessible_node_ids)
for node in error_nodes:
node_db = objects.Node.get_by_uid(node['uid'])
if not node_db:
logger.error(
u"Failed to delete node '%s' marked as error from Astute:"
" node doesn't exist", str(node)
)
else:
node_db.pending_deletion = False
node_db.status = 'error'
db().add(node_db)
node['name'] = node_db.name
db().flush()
success_msg = u"No nodes were removed"
err_msg = u"No errors occurred"
if nodes:
success_msg = u"Successfully removed {0} node(s)".format(
len(nodes)
)
notifier.notify("done", success_msg)
if error_nodes:
err_msg = u"Failed to remove {0} node(s): {1}".format(
len(error_nodes),
', '.join(
[n.get('name') or "ID: {0}".format(n['uid'])
for n in error_nodes])
)
notifier.notify("error", err_msg)
if not error_msg:
error_msg = ". ".join([success_msg, err_msg])
data = {
'status': status,
'progress': progress,
'message': error_msg,
}
objects.Task.update(task, data)
cls._update_action_log_entry(status, task.name, task_uuid, nodes)
def execute(self, nodes_to_delete, mclient_remove=True):
cluster_id = None
if hasattr(self, 'cluster'):
cluster_id = self.cluster.id
objects.TaskCollection.lock_cluster_tasks(cluster_id)
logger.info("Trying to execute node deletion task with nodes %s",
', '.join(str(node.id) for node in nodes_to_delete))
self.verify_nodes_with_cluster(nodes_to_delete)
objects.NodeCollection.lock_nodes(nodes_to_delete)
if cluster_id is None:
# DeletionTask operates on cluster's nodes.
# Nodes that are not in cluster are simply deleted.
Node.delete_by_ids([n.id for n in nodes_to_delete])
db().flush()
task = Task(name=consts.TASK_NAMES.node_deletion,
progress=100,
status=consts.TASK_STATUSES.ready)
db().add(task)
db().flush()
return task
task = Task(name=consts.TASK_NAMES.node_deletion,
cluster=self.cluster)
db().add(task)
for node in nodes_to_delete:
objects.Node.update(node,
{'status': consts.NODE_STATUSES.removing})
db().flush()
nodes_to_deploy = []
objects.Cluster.adjust_nodes_lists_on_controller_removing(
self.cluster, nodes_to_delete, nodes_to_deploy)
if nodes_to_deploy:
objects.NodeCollection.lock_nodes(nodes_to_deploy)
# updating nodes
objects.NodeCollection.update_slave_nodes_fqdn(nodes_to_deploy)
logger.debug("There are nodes to deploy: %s",
" ".join([n.fqdn for n in nodes_to_deploy]))
task_deployment = task.create_subtask(
consts.TASK_NAMES.deployment)
deployment_message = self._call_silently(
task_deployment,
tasks.DeploymentTask,
nodes_to_deploy,
method_name='message'
)
db().flush()
# if failed to generate task message for orchestrator
# then task is already set to error
if task_deployment.status == consts.TASK_STATUSES.error:
return task_deployment
rpc.cast('naily', [deployment_message])
db().commit()
self._call_silently(
task,
tasks.DeletionTask,
nodes=tasks.DeletionTask.prepare_nodes_for_task(
nodes_to_delete, mclient_remove=mclient_remove))
return task
def POST(self):
""":returns: JSONized Node object.
:http: * 201 (cluster successfully created)
* 400 (invalid node data specified)
* 403 (node has incorrect status)
* 409 (node with such parameters already exists)
"""
data = self.checked_data()
if data.get("status", "") != "discover":
error = web.forbidden()
error.data = "Only bootstrap nodes are allowed to be registered."
msg = u"Node with mac '{0}' was not created, " \
u"because request status is '{1}'."\
.format(data[u'mac'], data.get(u'status'))
logger.warning(msg)
raise error
node = Node(name="Untitled (%s)" % data['mac'][-5:],
timestamp=datetime.now())
if "cluster_id" in data:
# FIXME(vk): this part is needed only for tests. Normally,
# nodes are created only by agent and POST requests don't contain
# cluster_id, but our integration and unit tests widely use it.
# We need to assign cluster first
cluster_id = data.pop("cluster_id")
if cluster_id:
node.cluster = db.query(Cluster).get(cluster_id)
for key, value in data.iteritems():
if key == "id":
continue
elif key == "meta":
node.create_meta(value)
else:
setattr(node, key, value)
db().add(node)
db().commit()
node.attributes = NodeAttributes()
try:
node.attributes.volumes = node.volume_manager.gen_volumes_info()
if node.cluster:
node.cluster.add_pending_changes("disks", node_id=node.id)
except Exception as exc:
msg = (u"Failed to generate volumes "
"info for node '{0}': '{1}'").format(
node.name or data.get("mac") or data.get("id"),
str(exc) or "see logs for details")
logger.warning(traceback.format_exc())
notifier.notify("error", msg, node_id=node.id)
db().add(node)
db().commit()
network_manager = NetworkManager
# Add interfaces for node from 'meta'.
if node.meta and node.meta.get('interfaces'):
network_manager.update_interfaces_info(node)
if node.cluster_id:
network_manager = node.cluster.network_manager
network_manager.assign_networks_by_default(node)
try:
# we use multiplier of 1024 because there are no problems here
# with unfair size calculation
ram = str(
round(float(node.meta['memory']['total']) / 1073741824,
1)) + " GB RAM"
except Exception as exc:
logger.warning(traceback.format_exc())
ram = "unknown RAM"
try:
# we use multiplier of 1000 because disk vendors specify HDD size
# in terms of decimal capacity. Sources:
# http://knowledge.seagate.com/articles/en_US/FAQ/172191en
# http://physics.nist.gov/cuu/Units/binary.html
hd_size = round(
float(
sum([d["size"] for d in node.meta["disks"]]) / 1000000000),
1)
# if HDD > 100 GB we show it's size in TB
if hd_size > 100:
hd_size = str(hd_size / 1000) + " TB HDD"
else:
hd_size = str(hd_size) + " GB HDD"
except Exception as exc:
logger.warning(traceback.format_exc())
hd_size = "unknown HDD"
cores = str(node.meta.get('cpu', {}).get('total', "unknown"))
notifier.notify("discover",
"New node is discovered: %s CPUs / %s / %s " %
(cores, ram, hd_size),
node_id=node.id)
raise web.webapi.created(json.dumps(NodeHandler.render(node),
indent=4))