def cleanup(args):
now = dt.utcnow()
server = '{server}:{port}'.format(server=args.server, port=args.port)
logging.info('Connecting to {}'.format(server))
zk = KazooClient(hosts=server)
zk.start()
for path in args.zk_paths:
zk_path = '{}/{}'.format(args.zk_root_path, path)
nodes = zk.get_children(zk_path)
logging.info("Found {} nodes under {}".format(len(nodes), zk_path))
deleted = 0
for node in nodes:
node_path = '{}/{}'.format(zk_path, node)
data, stat = zk.get(node_path)
last_modified = dt.fromtimestamp(stat.mtime/1000.0)
if ((now - last_modified).days > args.age) or (args.inclusive and (now - last_modified).days >= args.age):
if not args.dry_run:
# Kazoo does not support recursive async deletes
if stat.children_count == 0:
res = zk.delete_async(node_path)
else:
zk.delete(node_path, recursive=True)
deleted += 1
logging.info("Deleted {} nodes".format(deleted))
zk.stop()
class ZooKeeper(AbstractDCS):
def __init__(self, config):
super(ZooKeeper, self).__init__(config)
hosts = config.get('hosts', [])
if isinstance(hosts, list):
hosts = ','.join(hosts)
self._client = KazooClient(hosts, handler=PatroniSequentialThreadingHandler(config['retry_timeout']),
timeout=config['ttl'], connection_retry={'max_delay': 1, 'max_tries': -1},
command_retry={'deadline': config['retry_timeout'], 'max_delay': 1, 'max_tries': -1})
self._client.add_listener(self.session_listener)
self._my_member_data = None
self._fetch_cluster = True
self._last_leader_operation = 0
self._orig_kazoo_connect = self._client._connection._connect
self._client._connection._connect = self._kazoo_connect
self._client.start()
def _kazoo_connect(self, host, port):
"""Kazoo is using Ping's to determine health of connection to zookeeper. If there is no
response on Ping after Ping interval (1/2 from read_timeout) it will consider current
connection dead and try to connect to another node. Without this "magic" it was taking
up to 2/3 from session timeout (ttl) to figure out that connection was dead and we had
only small time for reconnect and retry.
This method is needed to return different value of read_timeout, which is not calculated
from negotiated session timeout but from value of `loop_wait`. And it is 2 sec smaller
than loop_wait, because we can spend up to 2 seconds when calling `touch_member()` and
`write_leader_optime()` methods, which also may hang..."""
ret = self._orig_kazoo_connect(host, port)
return max(self.loop_wait - 2, 2)*1000, ret[1]
def session_listener(self, state):
if state in [KazooState.SUSPENDED, KazooState.LOST]:
self.cluster_watcher(None)
def cluster_watcher(self, event):
self._fetch_cluster = True
self.event.set()
def reload_config(self, config):
self.set_retry_timeout(config['retry_timeout'])
loop_wait = config['loop_wait']
loop_wait_changed = self._loop_wait != loop_wait
self._loop_wait = loop_wait
self._client.handler.set_connect_timeout(loop_wait)
# We need to reestablish connection to zookeeper if we want to change
# read_timeout (and Ping interval respectively), because read_timeout
# is calculated in `_kazoo_connect` method. If we are changing ttl at
# the same time, set_ttl method will reestablish connection and return
# `!True`, otherwise we will close existing connection and let kazoo
# open the new one.
if not self.set_ttl(int(config['ttl'] * 1000)) and loop_wait_changed:
self._client._connection._socket.close()
def set_ttl(self, ttl):
"""It is not possible to change ttl (session_timeout) in zookeeper without
destroying old session and creating the new one. This method returns `!True`
if session_timeout has been changed (`restart()` has been called)."""
if self._client._session_timeout != ttl:
self._client._session_timeout = ttl
self._client.restart()
return True
def set_retry_timeout(self, retry_timeout):
self._client._retry.deadline = retry_timeout
def get_node(self, key, watch=None):
try:
ret = self._client.get(key, watch)
return (ret[0].decode('utf-8'), ret[1])
except NoNodeError:
return None
@staticmethod
def member(name, value, znode):
return Member.from_node(znode.version, name, znode.ephemeralOwner, value)
def get_children(self, key, watch=None):
try:
return self._client.get_children(key, watch)
except NoNodeError:
return []
def load_members(self):
members = []
for member in self.get_children(self.members_path, self.cluster_watcher):
data = self.get_node(self.members_path + member)
if data is not None:
members.append(self.member(member, *data))
return members
def _inner_load_cluster(self):
self._fetch_cluster = False
self.event.clear()
nodes = set(self.get_children(self.client_path(''), self.cluster_watcher))
if not nodes:
self._fetch_cluster = True
# get initialize flag
initialize = (self.get_node(self.initialize_path) or [None])[0] if self._INITIALIZE in nodes else None
# get global dynamic configuration
config = self.get_node(self.config_path, watch=self.cluster_watcher) if self._CONFIG in nodes else None
config = config and ClusterConfig.from_node(config[1].version, config[0], config[1].mzxid)
# get list of members
members = self.load_members() if self._MEMBERS[:-1] in nodes else []
# get leader
leader = self.get_node(self.leader_path) if self._LEADER in nodes else None
if leader:
client_id = self._client.client_id
if leader[0] == self._name and client_id is not None and client_id[0] != leader[1].ephemeralOwner:
logger.info('I am leader but not owner of the session. Removing leader node')
self._client.delete(self.leader_path)
leader = None
if leader:
member = Member(-1, leader[0], None, {})
member = ([m for m in members if m.name == leader[0]] or [member])[0]
leader = Leader(leader[1].version, leader[1].ephemeralOwner, member)
self._fetch_cluster = member.index == -1
# failover key
failover = self.get_node(self.failover_path, watch=self.cluster_watcher) if self._FAILOVER in nodes else None
failover = failover and Failover.from_node(failover[1].version, failover[0])
# get last leader operation
optime = self.get_node(self.leader_optime_path) if self._OPTIME in nodes and self._fetch_cluster else None
self._last_leader_operation = 0 if optime is None else int(optime[0])
self._cluster = Cluster(initialize, config, leader, self._last_leader_operation, members, failover)
def _load_cluster(self):
if self._fetch_cluster or self._cluster is None:
try:
self._client.retry(self._inner_load_cluster)
except Exception:
logger.exception('get_cluster')
self.cluster_watcher(None)
raise ZooKeeperError('ZooKeeper in not responding properly')
def _create(self, path, value, **kwargs):
try:
self._client.retry(self._client.create, path, value.encode('utf-8'), **kwargs)
return True
except:
return False
def attempt_to_acquire_leader(self, permanent=False):
ret = self._create(self.leader_path, self._name, makepath=True, ephemeral=not permanent)
if not ret:
logger.info('Could not take out TTL lock')
return ret
def set_failover_value(self, value, index=None):
try:
self._client.retry(self._client.set, self.failover_path, value.encode('utf-8'), version=index or -1)
return True
except NoNodeError:
return value == '' or (index is None and self._create(self.failover_path, value))
except:
logging.exception('set_failover_value')
return False
def set_config_value(self, value, index=None):
try:
self._client.retry(self._client.set, self.config_path, value.encode('utf-8'), version=index or -1)
return True
except NoNodeError:
return index is None and self._create(self.config_path, value)
except Exception:
logging.exception('set_config_value')
return False
def initialize(self, create_new=True, sysid=""):
return self._create(self.initialize_path, sysid, makepath=True) if create_new \
else self._client.retry(self._client.set, self.initialize_path, sysid.encode("utf-8"))
def touch_member(self, data, ttl=None, permanent=False):
cluster = self.cluster
member = cluster and ([m for m in cluster.members if m.name == self._name] or [None])[0]
data = data.encode('utf-8')
if member and self._client.client_id is not None and member.session != self._client.client_id[0]:
try:
self._client.delete_async(self.member_path).get(timeout=1)
except NoNodeError:
pass
except:
return False
member = None
if member:
if data == self._my_member_data:
return True
else:
try:
self._client.create_async(self.member_path, data, makepath=True, ephemeral=not permanent).get(timeout=1)
self._my_member_data = data
return True
except Exception as e:
if not isinstance(e, NodeExistsError):
logger.exception('touch_member')
return False
try:
self._client.set_async(self.member_path, data).get(timeout=1)
self._my_member_data = data
return True
except:
logger.exception('touch_member')
return False
def take_leader(self):
return self.attempt_to_acquire_leader()
def write_leader_optime(self, last_operation):
last_operation = last_operation.encode('utf-8')
if last_operation != self._last_leader_operation:
try:
self._client.set_async(self.leader_optime_path, last_operation).get(timeout=1)
self._last_leader_operation = last_operation
except NoNodeError:
try:
self._client.create_async(self.leader_optime_path, last_operation, makepath=True).get(timeout=1)
self._last_leader_operation = last_operation
except:
logger.exception('Failed to create %s', self.leader_optime_path)
except:
logger.exception('Failed to update %s', self.leader_optime_path)
def update_leader(self):
return True
def delete_leader(self):
self._client.restart()
self._my_member_data = None
return True
def _cancel_initialization(self):
node = self.get_node(self.initialize_path)
if node:
self._client.delete(self.initialize_path, version=node[1].version)
def cancel_initialization(self):
try:
self._client.retry(self._cancel_initialization)
except:
logger.exception("Unable to delete initialize key")
def delete_cluster(self):
try:
return self._client.retry(self._client.delete, self.client_path(''), recursive=True)
except NoNodeError:
return True
def watch(self, timeout):
if super(ZooKeeper, self).watch(timeout):
self._fetch_cluster = True
return self._fetch_cluster
class ZookeeperServiceRegistry(BaseServiceRegistry):
def __init__(self, hosts=DEFAULT_HOSTS, chroot=DEFAULT_CHROOT):
super(ZookeeperServiceRegistry, self).__init__()
self.chroot = chroot
self.client = KazooClient(
hosts=hosts,
handler=SequentialGeventHandler(),
)
self.client.add_listener(self.on_kazoo_state_change)
self.start_count = 0
@classmethod
def from_config(cls, config, **kwargs):
return cls(hosts=config.get('hosts', DEFAULT_HOSTS),
chroot=config.get('chroot', DEFAULT_CHROOT),
**kwargs)
def on_start(self, timeout=10):
self.start_count += 1
if self.start_count > 1:
return
started = self.client.start_async()
started.wait(timeout=timeout)
if not self.client.connected:
raise RuntimeError('could not connect to zookeeper')
logger.debug('connected to zookeeper (version=%s)',
'.'.join(map(str, self.client.server_version())))
def on_stop(self):
self.start_count -= 1
if self.start_count != 0:
return
self.client.stop()
def on_kazoo_state_change(self, state):
logger.info('kazoo connection state changed to %s', state)
def on_service_type_watch(self, service, event):
try:
if event.type == EventType.CHILD:
# FIXME: figure out proper retry strategy
self.client.retry(self.lookup, service.container, service)
except Exception:
logger.exception('error in service type watcher')
def on_service_watch(self, service, event):
try:
prefix, service_type, identity = event.path.rsplit('/', 2)
if event.type == EventType.DELETED:
service.remove(identity)
except Exception:
logger.exception('error in service watcher')
def _get_service_znode(self, service, service_type, identity):
path = self._get_zk_path(service_type, identity)
result = self.client.get_async(path,
watch=functools.partial(
self.on_service_watch, service))
value, znode = result.get()
items = six.iteritems(json.loads(value.decode('utf-8')))
return {str(k): str(v) for k, v in items}
def discover(self, container):
result = self.client.get_children_async(path='%s/services' %
self.chroot, )
return list(result.get())
def lookup(self, container, service, watch=True, timeout=1):
def child_watch(event):
print(event)
service_type = service.service_type
result = self.client.get_children_async(
path='%s/services/%s' % (self.chroot, service_type),
watch=functools.partial(self.on_service_type_watch, service),
)
try:
names = result.get(timeout=timeout)
except NoNodeError:
raise LookupFailure(None,
"failed to resolve %s" % service.service_type)
logger.info("lookup %s %r", service_type, names)
identities = set(service.identities())
for name in names:
kwargs = self._get_service_znode(service, service_type, name)
identity = kwargs.pop('identity')
service.update(identity, **kwargs)
try:
identities.remove(identity)
except KeyError:
pass
for identity in identities:
service.remove(identity)
return service
def _get_zk_path(self, service_type, identity):
return '%s/services/%s/%s' % (self.chroot, service_type, identity)
def register(self, container, service_type, timeout=1):
path = self._get_zk_path(service_type, container.identity)
value = json.dumps({
'endpoint': container.endpoint,
'identity': container.identity,
'log_endpoint': container.log_endpoint,
})
result = self.client.create_async(path,
value.encode('utf-8'),
ephemeral=True,
makepath=True)
# FIXME: result.set_exception(RegistrationFailure())
result.get(timeout=timeout)
def unregister(self, container, service_type, timeout=1):
path = self._get_zk_path(service_type, container.identity)
result = self.client.delete_async(path)
result.set_exception(RegistrationFailure())
result.get(timeout=timeout)
class ZooKeeper(AbstractDCS):
def __init__(self, config):
super(ZooKeeper, self).__init__(config)
hosts = config.get('hosts', [])
if isinstance(hosts, list):
hosts = ','.join(hosts)
mapping = {
'use_ssl': 'use_ssl',
'verify': 'verify_certs',
'cacert': 'ca',
'cert': 'certfile',
'key': 'keyfile',
'key_password': '******'
}
kwargs = {v: config[k] for k, v in mapping.items() if k in config}
self._client = KazooClient(
hosts,
handler=PatroniSequentialThreadingHandler(config['retry_timeout']),
timeout=config['ttl'],
connection_retry=KazooRetry(max_delay=1,
max_tries=-1,
sleep_func=time.sleep),
command_retry=KazooRetry(deadline=config['retry_timeout'],
max_delay=1,
max_tries=-1,
sleep_func=time.sleep),
**kwargs)
self._client.add_listener(self.session_listener)
self._fetch_cluster = True
self._fetch_optime = True
self._orig_kazoo_connect = self._client._connection._connect
self._client._connection._connect = self._kazoo_connect
self._client.start()
def _kazoo_connect(self, *args):
"""Kazoo is using Ping's to determine health of connection to zookeeper. If there is no
response on Ping after Ping interval (1/2 from read_timeout) it will consider current
connection dead and try to connect to another node. Without this "magic" it was taking
up to 2/3 from session timeout (ttl) to figure out that connection was dead and we had
only small time for reconnect and retry.
This method is needed to return different value of read_timeout, which is not calculated
from negotiated session timeout but from value of `loop_wait`. And it is 2 sec smaller
than loop_wait, because we can spend up to 2 seconds when calling `touch_member()` and
`write_leader_optime()` methods, which also may hang..."""
ret = self._orig_kazoo_connect(*args)
return max(self.loop_wait - 2, 2) * 1000, ret[1]
def session_listener(self, state):
if state in [KazooState.SUSPENDED, KazooState.LOST]:
self.cluster_watcher(None)
def optime_watcher(self, event):
self._fetch_optime = True
self.event.set()
def cluster_watcher(self, event):
self._fetch_cluster = True
self.optime_watcher(event)
def reload_config(self, config):
self.set_retry_timeout(config['retry_timeout'])
loop_wait = config['loop_wait']
loop_wait_changed = self._loop_wait != loop_wait
self._loop_wait = loop_wait
self._client.handler.set_connect_timeout(loop_wait)
# We need to reestablish connection to zookeeper if we want to change
# read_timeout (and Ping interval respectively), because read_timeout
# is calculated in `_kazoo_connect` method. If we are changing ttl at
# the same time, set_ttl method will reestablish connection and return
# `!True`, otherwise we will close existing connection and let kazoo
# open the new one.
if not self.set_ttl(int(config['ttl'] * 1000)) and loop_wait_changed:
self._client._connection._socket.close()
def set_ttl(self, ttl):
"""It is not possible to change ttl (session_timeout) in zookeeper without
destroying old session and creating the new one. This method returns `!True`
if session_timeout has been changed (`restart()` has been called)."""
if self._client._session_timeout != ttl:
self._client._session_timeout = ttl
self._client.restart()
return True
@property
def ttl(self):
return self._client._session_timeout
def set_retry_timeout(self, retry_timeout):
retry = self._client.retry if isinstance(
self._client.retry, KazooRetry) else self._client._retry
retry.deadline = retry_timeout
def get_node(self, key, watch=None):
try:
ret = self._client.get(key, watch)
return (ret[0].decode('utf-8'), ret[1])
except NoNodeError:
return None
def get_leader_optime(self, leader):
watch = self.optime_watcher if not leader or leader.name != self._name else None
optime = self.get_node(self.leader_optime_path, watch)
self._fetch_optime = False
return optime and int(optime[0]) or 0
@staticmethod
def member(name, value, znode):
return Member.from_node(znode.version, name, znode.ephemeralOwner,
value)
def get_children(self, key, watch=None):
try:
return self._client.get_children(key, watch)
except NoNodeError:
return []
def load_members(self, sync_standby):
members = []
for member in self.get_children(self.members_path,
self.cluster_watcher):
watch = member in sync_standby and self.cluster_watcher or None
data = self.get_node(self.members_path + member, watch)
if data is not None:
members.append(self.member(member, *data))
return members
def _inner_load_cluster(self):
self._fetch_cluster = False
self.event.clear()
nodes = set(
self.get_children(self.client_path(''), self.cluster_watcher))
if not nodes:
self._fetch_cluster = True
# get initialize flag
initialize = (self.get_node(self.initialize_path)
or [None])[0] if self._INITIALIZE in nodes else None
# get global dynamic configuration
config = self.get_node(
self.config_path,
watch=self.cluster_watcher) if self._CONFIG in nodes else None
config = config and ClusterConfig.from_node(config[1].version,
config[0], config[1].mzxid)
# get timeline history
history = self.get_node(
self.history_path,
watch=self.cluster_watcher) if self._HISTORY in nodes else None
history = history and TimelineHistory.from_node(
history[1].mzxid, history[0])
# get synchronization state
sync = self.get_node(
self.sync_path,
watch=self.cluster_watcher) if self._SYNC in nodes else None
sync = SyncState.from_node(sync and sync[1].version, sync and sync[0])
# get list of members
sync_standby = sync.leader == self._name and sync.members or []
members = self.load_members(
sync_standby) if self._MEMBERS[:-1] in nodes else []
# get leader
leader = self.get_node(
self.leader_path) if self._LEADER in nodes else None
if leader:
client_id = self._client.client_id
if not self._ctl and leader[0] == self._name and client_id is not None \
and client_id[0] != leader[1].ephemeralOwner:
logger.info(
'I am leader but not owner of the session. Removing leader node'
)
self._client.delete(self.leader_path)
leader = None
if leader:
member = Member(-1, leader[0], None, {})
member = ([m for m in members if m.name == leader[0]]
or [member])[0]
leader = Leader(leader[1].version, leader[1].ephemeralOwner,
member)
self._fetch_cluster = member.index == -1
# get last leader operation
last_leader_operation = self._OPTIME in nodes and self.get_leader_optime(
leader)
# failover key
failover = self.get_node(
self.failover_path,
watch=self.cluster_watcher) if self._FAILOVER in nodes else None
failover = failover and Failover.from_node(failover[1].version,
failover[0])
return Cluster(initialize, config, leader, last_leader_operation,
members, failover, sync, history)
def _load_cluster(self):
cluster = self.cluster
if self._fetch_cluster or cluster is None:
try:
cluster = self._client.retry(self._inner_load_cluster)
except Exception:
logger.exception('get_cluster')
self.cluster_watcher(None)
raise ZooKeeperError('ZooKeeper in not responding properly')
# Optime ZNode was updated or doesn't exist and we are not leader
elif (self._fetch_optime and not self._fetch_cluster or not cluster.last_leader_operation) and\
not (cluster.leader and cluster.leader.name == self._name):
try:
optime = self.get_leader_optime(cluster.leader)
cluster = Cluster(cluster.initialize, cluster.config,
cluster.leader, optime, cluster.members,
cluster.failover, cluster.sync,
cluster.history)
except Exception:
pass
return cluster
def _bypass_caches(self):
self._fetch_cluster = True
def _create(self, path, value, retry=False, ephemeral=False):
try:
if retry:
self._client.retry(self._client.create,
path,
value,
makepath=True,
ephemeral=ephemeral)
else:
self._client.create_async(path,
value,
makepath=True,
ephemeral=ephemeral).get(timeout=1)
return True
except Exception:
logger.exception('Failed to create %s', path)
return False
def attempt_to_acquire_leader(self, permanent=False):
ret = self._create(self.leader_path,
self._name.encode('utf-8'),
retry=True,
ephemeral=not permanent)
if not ret:
logger.info('Could not take out TTL lock')
return ret
def _set_or_create(self,
key,
value,
index=None,
retry=False,
do_not_create_empty=False):
value = value.encode('utf-8')
try:
if retry:
self._client.retry(self._client.set,
key,
value,
version=index or -1)
else:
self._client.set_async(key, value, version=index
or -1).get(timeout=1)
return True
except NoNodeError:
if do_not_create_empty and not value:
return True
elif index is None:
return self._create(key, value, retry)
else:
return False
except Exception:
logger.exception('Failed to update %s', key)
return False
def set_failover_value(self, value, index=None):
return self._set_or_create(self.failover_path, value, index)
def set_config_value(self, value, index=None):
return self._set_or_create(self.config_path, value, index, retry=True)
def initialize(self, create_new=True, sysid=""):
sysid = sysid.encode('utf-8')
return self._create(self.initialize_path, sysid, retry=True) if create_new \
else self._client.retry(self._client.set, self.initialize_path, sysid)
def touch_member(self, data, permanent=False):
cluster = self.cluster
member = cluster and cluster.get_member(self._name,
fallback_to_leader=False)
encoded_data = json.dumps(data, separators=(',', ':')).encode('utf-8')
if member and (self._client.client_id is not None
and member.session != self._client.client_id[0] or
not (deep_compare(member.data.get('tags', {}),
data.get('tags', {})) and
member.data.get('version') == data.get('version')
and member.data.get('checkpoint_after_promote')
== data.get('checkpoint_after_promote'))):
try:
self._client.delete_async(self.member_path).get(timeout=1)
except NoNodeError:
pass
except Exception:
return False
member = None
if member:
if deep_compare(data, member.data):
return True
else:
try:
self._client.create_async(
self.member_path,
encoded_data,
makepath=True,
ephemeral=not permanent).get(timeout=1)
return True
except Exception as e:
if not isinstance(e, NodeExistsError):
logger.exception('touch_member')
return False
try:
self._client.set_async(self.member_path,
encoded_data).get(timeout=1)
return True
except Exception:
logger.exception('touch_member')
return False
def take_leader(self):
return self.attempt_to_acquire_leader()
def _write_leader_optime(self, last_operation):
return self._set_or_create(self.leader_optime_path, last_operation)
def _update_leader(self):
return True
def _delete_leader(self):
self._client.restart()
return True
def _cancel_initialization(self):
node = self.get_node(self.initialize_path)
if node:
self._client.delete(self.initialize_path, version=node[1].version)
def cancel_initialization(self):
try:
self._client.retry(self._cancel_initialization)
except Exception:
logger.exception("Unable to delete initialize key")
def delete_cluster(self):
try:
return self._client.retry(self._client.delete,
self.client_path(''),
recursive=True)
except NoNodeError:
return True
def set_history_value(self, value):
return self._set_or_create(self.history_path, value)
def set_sync_state_value(self, value, index=None):
return self._set_or_create(self.sync_path,
value,
index,
retry=True,
do_not_create_empty=True)
def delete_sync_state(self, index=None):
return self.set_sync_state_value("{}", index)
def watch(self, leader_index, timeout):
if super(ZooKeeper, self).watch(leader_index,
timeout) and not self._fetch_optime:
self._fetch_cluster = True
return self._fetch_cluster
class ZookeeperServiceRegistry(BaseServiceRegistry):
def __init__(self, hosts=DEFAULT_HOSTS, chroot=DEFAULT_CHROOT):
super(ZookeeperServiceRegistry, self).__init__()
self.chroot = chroot
self.client = KazooClient(
hosts=hosts,
handler=SequentialGeventHandler(),
)
self.client.add_listener(self.on_kazoo_state_change)
self.start_count = 0
@classmethod
def from_config(cls, config, **kwargs):
return cls(
hosts=config.get('hosts', DEFAULT_HOSTS),
chroot=config.get('chroot', DEFAULT_CHROOT),
**kwargs
)
def on_start(self, timeout=10):
self.start_count += 1
if self.start_count > 1:
return
started = self.client.start_async()
started.wait(timeout=timeout)
if not self.client.connected:
raise RuntimeError('could not connect to zookeeper')
logger.debug('connected to zookeeper (version=%s)', '.'.join(map(str, self.client.server_version())))
def on_stop(self):
self.start_count -= 1
if self.start_count != 0:
return
self.client.stop()
def on_kazoo_state_change(self, state):
logger.info('kazoo connection state changed to %s', state)
def on_service_type_watch(self, service, event):
try:
if event.type == EventType.CHILD:
# FIXME: figure out proper retry strategy
self.client.retry(self.lookup, service.container, service)
except Exception:
logger.exception('error in service type watcher')
def on_service_watch(self, service, event):
try:
prefix, service_type, identity = event.path.rsplit('/', 2)
if event.type == EventType.DELETED:
service.remove(identity)
except Exception:
logger.exception('error in service watcher')
def _get_service_znode(self, service, service_type, identity):
path = self._get_zk_path(service_type, identity)
result = self.client.get_async(
path, watch=functools.partial(self.on_service_watch, service))
value, znode = result.get()
items = six.iteritems(json.loads(value.decode('utf-8')))
return {str(k): str(v) for k, v in items}
def discover(self, container):
result = self.client.get_children_async(
path='%s/services' % self.chroot,
)
return list(result.get())
def lookup(self, container, service, watch=True, timeout=1):
def child_watch(event):
print(event)
service_type = service.service_type
result = self.client.get_children_async(
path='%s/services/%s' % (self.chroot, service_type),
watch=functools.partial(self.on_service_type_watch, service),
)
try:
names = result.get(timeout=timeout)
except NoNodeError:
raise LookupFailure(None, "failed to resolve %s" % service.service_type)
logger.info("lookup %s %r", service_type, names)
identities = set(service.identities())
for name in names:
kwargs = self._get_service_znode(service, service_type, name)
identity = kwargs.pop('identity')
service.update(identity, **kwargs)
try:
identities.remove(identity)
except KeyError:
pass
for identity in identities:
service.remove(identity)
return service
def _get_zk_path(self, service_type, identity):
return '%s/services/%s/%s' % (self.chroot, service_type, identity)
def register(self, container, service_type, timeout=1):
path = self._get_zk_path(service_type, container.identity)
value = json.dumps({
'endpoint': container.endpoint,
'identity': container.identity,
'log_endpoint': container.log_endpoint,
})
result = self.client.create_async(
path,
value.encode('utf-8'),
ephemeral=True, makepath=True)
# FIXME: result.set_exception(RegistrationFailure())
result.get(timeout=timeout)
def unregister(self, container, service_type, timeout=1):
path = self._get_zk_path(service_type, container.identity)
result = self.client.delete_async(path)
result.set_exception(RegistrationFailure())
result.get(timeout=timeout)
class Zookeeper:
def __init__(self, hosts):
self.zk = KazooClient(hosts=hosts,
handler=SequentialGeventHandler(),
logger=logger)
# returns immediately
event = self.zk.start_async()
# Wait for 30 seconds and see if we're connected
event.wait(timeout=30)
try:
if not self.zk.connected:
# Not connected, stop trying to connect
self.zk.stop()
except (ConnectionLossException, NoAuthException) as error:
raise error
except Exception as error:
raise error
@coroutine
def get_children(self, node):
try:
children = self.zk.get_children_async(node)
raise Return(children.get())
except Exception as error:
raise error
@coroutine
def get_node(self, node):
try:
data = self.zk.get_async(node)
raise Return(data.get())
except Exception as error:
raise error
@coroutine
def check_path_exist(self, path):
try:
result = self.zk.exists(path)
if result:
raise Return(True)
else:
raise Return(False)
except Exception as error:
raise error
@coroutine
def create_path(self, path):
try:
result = self.zk.ensure_path_async(path)
raise Return(result.get())
except Exception as error:
raise error
@coroutine
def create_node(self, path, value):
try:
result = self.zk.create_async(path=path,
value=value,
acl=None,
ephemeral=True)
raise Return(result.get())
except Exception as error:
raise error
@coroutine
def update_node(self, path, value, version=-1):
try:
result = self.zk.set_async(path, value, version)
raise Return(result.get())
except Exception as error:
raise error
@coroutine
def update_node(self, path, value, version=-1):
try:
result = self.zk.set_async(path, value, version)
raise Return(result.get())
except Exception as error:
raise error
@coroutine
def del_node(self, node):
try:
node_info = self.zk.delete_async(node)
raise Return(node_info.get())
except Exception as error:
raise error
def close(self):
self.zk.stop()