Files
vmware-nsx/neutron/db/agents_db.py
Roman Podoliaka 5529071bf1 Fix a race condition in agents status update code
Code handling agents status updates coming via RPC checks,
if a corresponding entry for the given (agent_type, host)
pair already exists in DB and updates it. And if it doesn't
exist, a new entry is created.

Without a unique constraint this can cause a race condition
resulting in adding of two agent entries having the same value
of (agent_type, host) pair.

Note, that it's already not allowed to have multiple agents of
the same type having the same host value, but currently it's
enforced only at code level, not at DB schema level, which
effectively makes race conditions possible.

Closes-Bug: #1254246

Change-Id: I1ebaa111154b3d6b34074705b579097ab730594c
2013-12-09 15:18:34 +02:00

214 lines
8.7 KiB
Python

# vim: tabstop=4 shiftwidth=4 softtabstop=4
# Copyright (c) 2013 OpenStack Foundation.
# 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.
from eventlet import greenthread
from oslo.config import cfg
import sqlalchemy as sa
from sqlalchemy.orm import exc
from neutron.db import model_base
from neutron.db import models_v2
from neutron.extensions import agent as ext_agent
from neutron import manager
from neutron.openstack.common.db import exception as db_exc
from neutron.openstack.common import jsonutils
from neutron.openstack.common import log as logging
from neutron.openstack.common import timeutils
LOG = logging.getLogger(__name__)
cfg.CONF.register_opt(
cfg.IntOpt('agent_down_time', default=9,
help=_("Seconds to regard the agent is down; should be at "
"least twice report_interval, to be sure the "
"agent is down for good.")))
class Agent(model_base.BASEV2, models_v2.HasId):
"""Represents agents running in neutron deployments."""
__table_args__ = (
sa.UniqueConstraint('agent_type', 'host',
name='uniq_agents0agent_type0host'),
)
# L3 agent, DHCP agent, OVS agent, LinuxBridge
agent_type = sa.Column(sa.String(255), nullable=False)
binary = sa.Column(sa.String(255), nullable=False)
# TOPIC is a fanout exchange topic
topic = sa.Column(sa.String(255), nullable=False)
# TOPIC.host is a target topic
host = sa.Column(sa.String(255), nullable=False)
admin_state_up = sa.Column(sa.Boolean, default=True,
nullable=False)
# the time when first report came from agents
created_at = sa.Column(sa.DateTime, nullable=False)
# the time when first report came after agents start
started_at = sa.Column(sa.DateTime, nullable=False)
# updated when agents report
heartbeat_timestamp = sa.Column(sa.DateTime, nullable=False)
# description is note for admin user
description = sa.Column(sa.String(255))
# configurations: a json dict string, I think 4095 is enough
configurations = sa.Column(sa.String(4095), nullable=False)
class AgentDbMixin(ext_agent.AgentPluginBase):
"""Mixin class to add agent extension to db_plugin_base_v2."""
def _get_agent(self, context, id):
try:
agent = self._get_by_id(context, Agent, id)
except exc.NoResultFound:
raise ext_agent.AgentNotFound(id=id)
return agent
@classmethod
def is_agent_down(cls, heart_beat_time):
return timeutils.is_older_than(heart_beat_time,
cfg.CONF.agent_down_time)
def get_configuration_dict(self, agent_db):
try:
conf = jsonutils.loads(agent_db.configurations)
except Exception:
msg = _('Configuration for agent %(agent_type)s on host %(host)s'
' is invalid.')
LOG.warn(msg, {'agent_type': agent_db.agent_type,
'host': agent_db.host})
conf = {}
return conf
def _make_agent_dict(self, agent, fields=None):
attr = ext_agent.RESOURCE_ATTRIBUTE_MAP.get(
ext_agent.RESOURCE_NAME + 's')
res = dict((k, agent[k]) for k in attr
if k not in ['alive', 'configurations'])
res['alive'] = not AgentDbMixin.is_agent_down(
res['heartbeat_timestamp'])
res['configurations'] = self.get_configuration_dict(agent)
return self._fields(res, fields)
def delete_agent(self, context, id):
with context.session.begin(subtransactions=True):
agent = self._get_agent(context, id)
context.session.delete(agent)
def update_agent(self, context, id, agent):
agent_data = agent['agent']
with context.session.begin(subtransactions=True):
agent = self._get_agent(context, id)
agent.update(agent_data)
return self._make_agent_dict(agent)
def get_agents_db(self, context, filters=None):
query = self._get_collection_query(context, Agent, filters=filters)
return query.all()
def get_agents(self, context, filters=None, fields=None):
return self._get_collection(context, Agent,
self._make_agent_dict,
filters=filters, fields=fields)
def _get_agent_by_type_and_host(self, context, agent_type, host):
query = self._model_query(context, Agent)
try:
agent_db = query.filter(Agent.agent_type == agent_type,
Agent.host == host).one()
return agent_db
except exc.NoResultFound:
raise ext_agent.AgentNotFoundByTypeHost(agent_type=agent_type,
host=host)
except exc.MultipleResultsFound:
raise ext_agent.MultipleAgentFoundByTypeHost(agent_type=agent_type,
host=host)
def get_agent(self, context, id, fields=None):
agent = self._get_agent(context, id)
return self._make_agent_dict(agent, fields)
def _create_or_update_agent(self, context, agent):
with context.session.begin(subtransactions=True):
res_keys = ['agent_type', 'binary', 'host', 'topic']
res = dict((k, agent[k]) for k in res_keys)
configurations_dict = agent.get('configurations', {})
res['configurations'] = jsonutils.dumps(configurations_dict)
current_time = timeutils.utcnow()
try:
agent_db = self._get_agent_by_type_and_host(
context, agent['agent_type'], agent['host'])
res['heartbeat_timestamp'] = current_time
if agent.get('start_flag'):
res['started_at'] = current_time
greenthread.sleep(0)
agent_db.update(res)
except ext_agent.AgentNotFoundByTypeHost:
greenthread.sleep(0)
res['created_at'] = current_time
res['started_at'] = current_time
res['heartbeat_timestamp'] = current_time
res['admin_state_up'] = True
agent_db = Agent(**res)
greenthread.sleep(0)
context.session.add(agent_db)
greenthread.sleep(0)
def create_or_update_agent(self, context, agent):
"""Create or update agent according to report."""
try:
return self._create_or_update_agent(context, agent)
except db_exc.DBDuplicateEntry as e:
if e.columns == ['agent_type', 'host']:
# It might happen that two or more concurrent transactions are
# trying to insert new rows having the same value of
# (agent_type, host) pair at the same time (if there has been
# no such entry in the table and multiple agent status updates
# are being processed at the moment). In this case having a
# unique constraint on (agent_type, host) columns guarantees
# that only one transaction will succeed and insert a new agent
# entry, others will fail and be rolled back. That means we
# must retry them one more time: no INSERTs will be issued,
# because _get_agent_by_type_and_host() will return the
# existing agent entry, which will be updated multiple times
return self._create_or_update_agent(context, agent)
raise
class AgentExtRpcCallback(object):
"""Processes the rpc report in plugin implementations."""
RPC_API_VERSION = '1.0'
START_TIME = timeutils.utcnow()
def __init__(self, plugin=None):
self.plugin = plugin
def report_state(self, context, **kwargs):
"""Report state from agent to server."""
time = kwargs['time']
time = timeutils.parse_strtime(time)
if self.START_TIME > time:
LOG.debug(_("Message with invalid timestamp received"))
return
agent_state = kwargs['agent_state']['agent_state']
if not self.plugin:
self.plugin = manager.NeutronManager.get_plugin()
self.plugin.create_or_update_agent(context, agent_state)