254 lines
10 KiB
Python
254 lines
10 KiB
Python
#
|
|
# 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 re
|
|
|
|
import eventlet
|
|
import netaddr
|
|
from neutron_lib import exceptions
|
|
from oslo_concurrency import lockutils
|
|
from oslo_log import log as logging
|
|
|
|
from neutron.agent.linux import utils as linux_utils
|
|
from neutron.common import constants as n_const
|
|
|
|
LOG = logging.getLogger(__name__)
|
|
CONTRACK_MGRS = {}
|
|
MAX_CONNTRACK_ZONES = 65535
|
|
ZONE_START = 4097
|
|
|
|
WORKERS = 8
|
|
|
|
|
|
class IpConntrackUpdate(object):
|
|
"""Encapsulates a conntrack update
|
|
|
|
An instance of this object carries the information necessary to
|
|
process a request to update the conntrack table.
|
|
"""
|
|
def __init__(self, device_info_list, rule, remote_ips):
|
|
self.device_info_list = device_info_list
|
|
self.rule = rule
|
|
self.remote_ips = remote_ips
|
|
|
|
def __repr__(self):
|
|
return ('<IpConntrackUpdate(device_info_list=%s, rule=%s, '
|
|
'remote_ips=%s>' % (self.device_info_list, self.rule,
|
|
self.remote_ips))
|
|
|
|
|
|
@lockutils.synchronized('conntrack')
|
|
def get_conntrack(get_rules_for_table_func, filtered_ports, unfiltered_ports,
|
|
execute=None, namespace=None, zone_per_port=False):
|
|
try:
|
|
return CONTRACK_MGRS[namespace]
|
|
except KeyError:
|
|
ipconntrack = IpConntrackManager(get_rules_for_table_func,
|
|
filtered_ports, unfiltered_ports,
|
|
execute, namespace, zone_per_port)
|
|
CONTRACK_MGRS[namespace] = ipconntrack
|
|
return CONTRACK_MGRS[namespace]
|
|
|
|
|
|
class IpConntrackManager(object):
|
|
"""Smart wrapper for ip conntrack."""
|
|
|
|
def __init__(self, get_rules_for_table_func, filtered_ports,
|
|
unfiltered_ports, execute=None, namespace=None,
|
|
zone_per_port=False):
|
|
self.get_rules_for_table_func = get_rules_for_table_func
|
|
self.execute = execute or linux_utils.execute
|
|
self.namespace = namespace
|
|
self.filtered_ports = filtered_ports
|
|
self.unfiltered_ports = unfiltered_ports
|
|
self.zone_per_port = zone_per_port # zone per port vs per network
|
|
self._populate_initial_zone_map()
|
|
self._queue = eventlet.queue.LightQueue()
|
|
self._start_process_queue()
|
|
|
|
def _start_process_queue(self):
|
|
LOG.debug("Starting ip_conntrack _process_queue_worker() threads")
|
|
pool = eventlet.GreenPool(size=WORKERS)
|
|
for i in range(WORKERS):
|
|
pool.spawn_n(self._process_queue_worker)
|
|
|
|
def _process_queue_worker(self):
|
|
# While it's technically not necessary to have this method, the
|
|
# 'while True' could just be in _process_queue(), the tests have
|
|
# to be able to drain the queue without blocking, so _process_queue()
|
|
# is made standalone.
|
|
while True:
|
|
self._process_queue()
|
|
|
|
def _process_queue(self):
|
|
update = None
|
|
try:
|
|
# this will block until an entry gets added to the queue
|
|
update = self._queue.get()
|
|
if update.remote_ips:
|
|
for remote_ip in update.remote_ips:
|
|
self._delete_conntrack_state(
|
|
update.device_info_list, update.rule, remote_ip)
|
|
else:
|
|
self._delete_conntrack_state(
|
|
update.device_info_list, update.rule)
|
|
except Exception:
|
|
LOG.exception("Failed to process ip_conntrack queue entry: %s",
|
|
update)
|
|
|
|
def _process(self, device_info_list, rule, remote_ips=None):
|
|
# queue the update to allow the caller to resume its work
|
|
update = IpConntrackUpdate(device_info_list, rule, remote_ips)
|
|
self._queue.put(update)
|
|
|
|
@staticmethod
|
|
def _generate_conntrack_cmd_by_rule(rule, namespace):
|
|
ethertype = rule.get('ethertype')
|
|
protocol = rule.get('protocol')
|
|
direction = rule.get('direction')
|
|
cmd = ['conntrack', '-D']
|
|
if protocol:
|
|
cmd.extend(['-p', str(protocol)])
|
|
cmd.extend(['-f', str(ethertype).lower()])
|
|
cmd.append('-d' if direction == 'ingress' else '-s')
|
|
cmd_ns = []
|
|
if namespace:
|
|
cmd_ns.extend(['ip', 'netns', 'exec', namespace])
|
|
cmd_ns.extend(cmd)
|
|
return cmd_ns
|
|
|
|
def _get_conntrack_cmds(self, device_info_list, rule, remote_ip=None):
|
|
conntrack_cmds = set()
|
|
cmd = self._generate_conntrack_cmd_by_rule(rule, self.namespace)
|
|
ethertype = rule.get('ethertype')
|
|
for device_info in device_info_list:
|
|
zone_id = self.get_device_zone(device_info, create=False)
|
|
if not zone_id:
|
|
LOG.debug("No zone for device %(dev)s. Will not try to "
|
|
"clear conntrack state. Zone map: %(zm)s",
|
|
{'dev': device_info['device'],
|
|
'zm': self._device_zone_map})
|
|
continue
|
|
ips = device_info.get('fixed_ips', [])
|
|
for ip in ips:
|
|
net = netaddr.IPNetwork(ip)
|
|
if str(net.version) not in ethertype:
|
|
continue
|
|
ip_cmd = [str(net.ip), '-w', zone_id]
|
|
if remote_ip and str(
|
|
netaddr.IPNetwork(remote_ip).version) in ethertype:
|
|
if rule.get('direction') == 'ingress':
|
|
direction = '-s'
|
|
else:
|
|
direction = '-d'
|
|
ip_cmd.extend([direction, str(remote_ip)])
|
|
conntrack_cmds.add(tuple(cmd + ip_cmd))
|
|
return conntrack_cmds
|
|
|
|
def _delete_conntrack_state(self, device_info_list, rule, remote_ip=None):
|
|
conntrack_cmds = self._get_conntrack_cmds(device_info_list,
|
|
rule, remote_ip)
|
|
for cmd in conntrack_cmds:
|
|
try:
|
|
self.execute(list(cmd), run_as_root=True,
|
|
check_exit_code=True,
|
|
extra_ok_codes=[1])
|
|
except RuntimeError:
|
|
LOG.exception("Failed execute conntrack command %s", cmd)
|
|
|
|
def delete_conntrack_state_by_rule(self, device_info_list, rule):
|
|
self._process(device_info_list, rule)
|
|
|
|
def delete_conntrack_state_by_remote_ips(self, device_info_list,
|
|
ethertype, remote_ips):
|
|
for direction in ['ingress', 'egress']:
|
|
rule = {'ethertype': str(ethertype).lower(),
|
|
'direction': direction}
|
|
self._process(device_info_list, rule, remote_ips)
|
|
|
|
def _populate_initial_zone_map(self):
|
|
"""Setup the map between devices and zones based on current rules."""
|
|
self._device_zone_map = {}
|
|
rules = self.get_rules_for_table_func('raw')
|
|
for rule in rules:
|
|
match = re.match(r'.* --physdev-in (?P<dev>[a-zA-Z0-9\-]+)'
|
|
r'.* -j CT --zone (?P<zone>\d+).*', rule)
|
|
if match:
|
|
# strip off any prefix that the interface is using
|
|
short_port_id = (match.group('dev')
|
|
[n_const.LINUX_DEV_PREFIX_LEN:])
|
|
self._device_zone_map[short_port_id] = int(match.group('zone'))
|
|
LOG.debug("Populated conntrack zone map: %s", self._device_zone_map)
|
|
|
|
def _device_key(self, port):
|
|
# we have to key the device_zone_map based on the fragment of the
|
|
# UUID that shows up in the interface name. This is because the initial
|
|
# map is populated strictly based on interface names that we don't know
|
|
# the full UUID of.
|
|
if self.zone_per_port:
|
|
identifier = port['device'][n_const.LINUX_DEV_PREFIX_LEN:]
|
|
else:
|
|
identifier = port['network_id']
|
|
return identifier[:(n_const.LINUX_DEV_LEN -
|
|
n_const.LINUX_DEV_PREFIX_LEN)]
|
|
|
|
def get_device_zone(self, port, create=True):
|
|
device_key = self._device_key(port)
|
|
try:
|
|
return self._device_zone_map[device_key]
|
|
except KeyError:
|
|
if create:
|
|
return self._generate_device_zone(device_key)
|
|
|
|
def _free_zones_from_removed_ports(self):
|
|
"""Clears any entries from the zone map of removed ports."""
|
|
existing_ports = [
|
|
self._device_key(port)
|
|
for port in (list(self.filtered_ports.values()) +
|
|
list(self.unfiltered_ports.values()))
|
|
]
|
|
removed = set(self._device_zone_map) - set(existing_ports)
|
|
for dev in removed:
|
|
self._device_zone_map.pop(dev, None)
|
|
|
|
def _generate_device_zone(self, short_device_id):
|
|
"""Generates a unique conntrack zone for the passed in ID."""
|
|
try:
|
|
zone = self._find_open_zone()
|
|
except exceptions.CTZoneExhaustedError:
|
|
# Free some zones and try again, repeat failure will not be caught
|
|
self._free_zones_from_removed_ports()
|
|
zone = self._find_open_zone()
|
|
|
|
self._device_zone_map[short_device_id] = zone
|
|
LOG.debug("Assigned CT zone %(z)s to device %(dev)s.",
|
|
{'z': zone, 'dev': short_device_id})
|
|
return self._device_zone_map[short_device_id]
|
|
|
|
def _find_open_zone(self):
|
|
# call set to dedup because old ports may be mapped to the same zone.
|
|
zones_in_use = sorted(set(self._device_zone_map.values()))
|
|
if not zones_in_use:
|
|
return ZONE_START
|
|
# attempt to increment onto the highest used zone first. if we hit the
|
|
# end, go back and look for any gaps left by removed devices.
|
|
last = zones_in_use[-1]
|
|
if last < MAX_CONNTRACK_ZONES:
|
|
return max(last + 1, ZONE_START)
|
|
for index, used in enumerate(zones_in_use):
|
|
if used - index != ZONE_START:
|
|
# gap found, let's use it!
|
|
return index + ZONE_START
|
|
# conntrack zones exhausted :( :(
|
|
raise exceptions.CTZoneExhaustedError()
|