class IKazooClient(metaclass=Singleton):
# zk_client = None
def __init__(self) -> None:
super().__init__()
self.zk_client = KazooClient(hosts=ZK_PATH)
self.zk_client.start()
def get_kazoo_client(self):
"""这里加异常捕捉 防止本服务不可以用"""
while self.zk_client is None or (self.zk_client.state in [
KeeperState.CLOSED, KeeperState.AUTH_FAILED, KazooState.LOST
]):
try:
if self.zk_client.state in [
KeeperState.CLOSED, KeeperState.AUTH_FAILED,
KazooState.LOST
]:
print("ERROR : zookeeper 尝试重连")
self.zk_client.restart()
# 重新注册服务
# re_register()
time.sleep(1)
except Exception:
print(self.zk_client)
if self.zk_client is not None:
print(self.zk_client.state)
print(
"ERROR : zookeeper 连接失败 socket 节点选取将不可以用!!!!!!!!!!!!!!!!!!!"
)
return self.zk_client
class zkUtils:
"""Zookeeper utility class implemented with Kazoo"""
def __init__(self,
hosts=DEFAULT_ZK,
read_only=True,
logging_level=logging.INFO):
self.zc = KazooClient(hosts=hosts, read_only=read_only)
self.logger = get_logger(LOGGER_NAME, logging_level)
self.logger.info(
"Zookeeper hosts: {}, client util instantiated.".format(hosts))
def get_path_data(self, path="/"):
try:
self.zc.restart()
if self.zc.exists(path):
return self.zc.get(path)[0]
else:
return None
except Exception as e:
self.logger.exception("Get path data exception: {}!".format(
str(e)))
return None
def get_children_list(self, path="/"):
try:
self.zc.restart()
if self.zc.exists(path):
return self.zc.get_children(path)
else:
return None
except Exception as e:
self.logger.exception("Get path data exception: {}!".format(
str(e)))
return None
class ZKBase(object):
"""
scheduler需要定义以下函数：
- callback(KazooState.*)： 处理CONN: LOST,SUSPENDED
-
"""
def __init__(self,conf,scheduler):
self.conf = conf
self.state = "STOPPED"
self.zk = KazooClient(self.conf.processor.zk_url, timeout=self.conf.processor.timeout)
try:
self.zk.start()
except Exception:
LOG.error("Start zk error: %s" % traceback.format_exc(limit=2))
self.zk.add_listener(self.listen)
self.scheduler = scheduler
self.state = "RUNNING" # STOPPED
self.state_changes = False
def listen(self, state):
if state == KazooState.LOST or state == KazooState.SUSPENDED:
LOG.error("Session state change: %s" % state)
if self.state == "STOPPED":
return
self.state_changes = True
self.scheduler.callback(state)
elif state == KazooState.CONNECTED:
LOG.info("Connectied to ZK.")
else:
LOG.error("Session timeout. Cannot conect to ZK.")
def reinit(self,state):
while self.zk.state != KazooState.CONNECTED:
LOG.info("Restart zk connection until connected.")
try:
self.zk.restart()
except Exception:
LOG.error("Reinit: %s" % traceback.format_exc(limit=1))
self.state_changes = False
def create(self,path, value='', acl=None, ephemeral=False, sequence=False, makepath=False):
LOG.info("Create: path %s" % path)
self.zk.create(path=path,
value=value.encode(),
ephemeral=ephemeral,
sequence=sequence,
makepath=makepath)
def get(self,path,watch=None):
self.zk.get(path,watch)
def get_children(self,path,watch=None):
return self.zk.get_children(path,watch=watch)
def terminate(self):
LOG.info("Terminate ZK connection.")
self.state = "STOPPED"
self.zk.stop()
self.zk.close()
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 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 ZooKeeper(AbstractDCS):
def __init__(self, name, config):
super(ZooKeeper, self).__init__(name, config)
hosts = config.get('hosts', [])
if isinstance(hosts, list):
hosts = ','.join(hosts)
self.exhibitor = None
if 'exhibitor' in config:
exhibitor = config['exhibitor']
interval = exhibitor.get('poll_interval', 300)
self.exhibitor = ExhibitorEnsembleProvider(exhibitor['hosts'], exhibitor['port'], poll_interval=interval)
hosts = self.exhibitor.zookeeper_hosts
self._client = KazooClient(hosts=hosts, timeout=(config.get('session_timeout') or 30),
command_retry={'deadline': (config.get('reconnect_timeout') or 10),
'max_delay': 1, 'max_tries': -1},
connection_retry={'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._client.start()
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 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 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
if failover:
failover = 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, leader, self._last_leader_operation, members, failover)
def _load_cluster(self):
if self.exhibitor and self.exhibitor.poll():
self._client.set_hosts(self.exhibitor.zookeeper_hosts)
if self._fetch_cluster or self._cluster is None:
try:
self._client.retry(self._inner_load_cluster)
except:
logger.exception('get_cluster')
self.session_listener(KazooState.LOST)
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):
ret = self._create(self.leader_path, self._name, makepath=True, ephemeral=True)
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 (not index and self._create(self.failover_path, value))
except:
logging.exception('set_failover_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):
cluster = self.cluster
member = cluster and ([m for m in cluster.members if m.name == self._name] or [None])[0]
path = self.member_path
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.retry(self._client.delete, path)
except NoNodeError:
pass
except:
return False
member = None
if member and data == self._my_member_data:
return True
try:
if member:
self._client.retry(self._client.set, path, data)
else:
self._client.retry(self._client.create, path, data, makepath=True, ephemeral=True)
self._my_member_data = data
return True
except NodeExistsError:
try:
self._client.retry(self._client.set, path, data)
self._my_member_data = data
return True
except:
logger.exception('touch_member')
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:
self._last_leader_operation = last_operation
path = self.leader_optime_path
try:
self._client.retry(self._client.set, path, last_operation)
except NoNodeError:
try:
self._client.retry(self._client.create, path, last_operation, makepath=True)
except:
logger.exception('Failed to create %s', path)
except:
logger.exception('Failed to update %s', 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 ServiceRegister(object):
def __init__(self, hosts="127.0.0.1:2181", read_only=True, logger=None):
"""
服务注册
:param hosts: Zookeeper集群地址列表
:param read_only: 是否只读
:param logger: 日志对象
"""
if not logger:
import logging
logging.basicConfig()
self._zk = KazooClient(hosts, read_only=read_only, logger=logger)
self._zk.start()
def restart(self):
self._zk.restart()
def retry_get(self, path, watcher=None):
"""
重读
:param path: 节点路由
:param watcher: 观察者回调函数
:return: 成功：节点值，版本号；失败：异常信息，异常代码。
"""
return self._zk.retry(self.get, path, watcher)
def lock(self, path, identifier, timeout=None):
"""
分布式锁
:param path: 路由
:param identifier: 锁标识
:param timeout: 超时时间
:return: 锁对象
"""
return DLock(self._zk, path, identifier, timeout)
def exist(self, path):
"""
节点是否存在
:param path: 路由
:return: 存在返回True，不存在返回False。
"""
state = self._zk.exists(path)
return state is not None
def create(self, path, value=""):
"""
创建节点
:param path: 节点路由
:param value: 节点值
:return: 节点路由
"""
try:
res_path = self._zk.create(path, value, makepath=True)
except NodeExistsError:
return path
except NoNodeError as e:
return e.message
except ZookeeperError as e:
return e.message
else:
return res_path
def get(self, path, watcher=None):
"""
查节点值
:param path: 节点路由
:param watcher: 观察者回调函数
:return: 成功：节点值，版本号；失败：异常信息，异常代码。
"""
try:
data, state = self._zk.get(path)
self._zk.DataWatch(path, watcher)
except NoNodeError as e:
return e.message, -2
except ZookeeperError as e:
return e.message, -3
else:
return data, state.version
def get_children(self, path, watcher=None):
"""
查子节点列表
:param path: 节点路由
:param watcher: 观察者回调函数
:return: 子节点列表
"""
try:
data = self._zk.get_children(path)
self._zk.DataWatch(path, watcher)
except NoNodeError as e:
return [], -2
except ZookeeperError as e:
return [], -3
else:
return data, 0
def set(self, path, value, version=-1):
"""
改节点值
:param path: 节点路由
:param value: 节点值
:param version: 成功：版本号；失败：异常信息。
"""
try:
state = self._zk.set(path, value, version)
except BadVersionError as e:
return e.message
except NoNodeError as e:
return e.message
except ZookeeperError as e:
return e.message
else:
return state.version
def run(self):
server_info = self.mac_and_ip(os.environ.get('MONITOR_INTERFACE', 'eth0'))
zk_path = '/remote/alive/workstation/%s' % server_info['mac']
zk = KazooClient(hosts=os.environ.get('ZOOKEEPER', 'zookeeper_server:2181'))
sleep_time = 10
while not zk.connected:
try:
zk.start()
except:
pass
time.sleep(sleep_time)
web_keyname, has_error, device_loop, loop = 'api', False, int(10 / sleep_time), 0
server_info[web_keyname] = {}
while True:
try:
if not zk.connected:
zk.restart()
logger.info('Restart zk connection!')
if not zk.exists(zk_path):
value = json.dumps(server_info)
zk.create(zk_path, value, ephemeral=True, makepath=True)
logger.info('Create ZK %s: %s' % (zk_path, value))
server_info[web_keyname].update({'port': int(os.environ.get('MONITOR_PORT', 80)), 'path': '/api'})
url = 'http://%s:%d' % (server_info['ip'], server_info[web_keyname]['port'])
try:
r = requests.get('%s/api/ping' % url)
if r.status_code == 200:
server_info[web_keyname]['status'] = 'up'
else:
logger.error("%s --- %s" % (r.url, r.text))
except Error as e:
logger.error(e)
server_info[web_keyname]['status'] = 'down'
server_info[web_keyname]['devices'] = {}
server_info[web_keyname]['jobs'] = []
else:
if loop == 0: # we want not to query devices so frequently, so...
devices = requests.get('%s/api/0/devices' % url)
if devices.status_code == 200:
server_info[web_keyname]['devices'] = devices.json()
else:
logger.error("%s --- %s" % (devices.url, devices.text))
server_info[web_keyname]['devices'] = {}
jobs = requests.get('%s/api/0/jobs' % url, params={'all': False})
if jobs.status_code == 200:
server_info[web_keyname]['jobs'] = jobs.json()['jobs']
else:
logger.error("%s --- %s" % (jobs.url, jobs.text))
server_info[web_keyname]['jobs'] = []
data, stat = zk.get(zk_path)
if data and json.loads(data) != server_info:
value = json.dumps(server_info)
zk.set(zk_path, value)
logger.info('Update ZK %s: %s' % (zk_path, value))
except:
loop = 0
if not has_error:
logger.error("Connection error!")
has_error = True
else:
has_error = False
loop = (loop + 1) % device_loop
time.sleep(sleep_time)
