nova-conductor与AMQP(二)

源码版本：H版

一、首先看服务的启动脚本

/usr/bin/nova-conductor

import sys
from nova.cmd.conductor import main
if __name__ == "__main__":
    sys.exit(main())

nova/cmd/conductor.py

def main():
    ...
    server = service.Service.create(binary='nova-conductor',
                                    topic=CONF.conductor.topic,
                                    manager=CONF.conductor.manager)
    service.serve(server, workers=CONF.conductor.workers)
    service.wait()

二、分析RPC服务的创建

nova/service.py

Service类：
@classmethod
def create(cls, host=None, binary=None, topic=None, manager=None,
           report_interval=None, periodic_enable=None,
           periodic_fuzzy_delay=None, periodic_interval_max=None,
           db_allowed=True):
    ...
    service_obj = cls(host, binary, topic, manager,
                      report_interval=report_interval,
                      periodic_enable=periodic_enable,
                      periodic_fuzzy_delay=periodic_fuzzy_delay,
                      periodic_interval_max=periodic_interval_max,
                      db_allowed=db_allowed)
    return service_obj

三、分析服务的运行

nova/service.py

def serve(server, workers=None):
    global _launcher
    if _launcher:
        raise RuntimeError(_('serve() can only be called once'))
    """ service为nova.openstack.common.service"""
    _launcher = service.launch(server, workers=workers)

nova/openstack/common/service.py

def launch(service, workers=None):
    if workers:
        launcher = ProcessLauncher()
        launcher.launch_service(service, workers=workers)
    else:
        launcher = ServiceLauncher()
        launcher.launch_service(service)
    return launcher

　　参考nova-api的服务启动过程（http://www.cnblogs.com/littlebugfish/p/4022907.html）,即启动协程（使用eventlet）运行Service，主要是start函数。注意，如果在配置文件中指明多个workers的话，将有多个进程监听消息队列，取得消息的进程负责处理。接着看start函数的代码，如下：

nova/service.py

Service类：
def start(self):
    ...
    self.manager.pre_start_hook()
    if self.backdoor_port is not None:
        self.manager.backdoor_port = self.backdoor_port
    """创建AMQP连接，方便后面创建Consumer"""
    self.conn = rpc.create_connection(new=True)
    LOG.debug(_("Creating Consumer connection for Service %s") %
              self.topic)
    rpc_dispatcher = self.manager.create_rpc_dispatcher(self.backdoor_port)
    self.conn.create_consumer(self.topic, rpc_dispatcher, fanout=False)
    node_topic = '%s.%s' % (self.topic, self.host)
    self.conn.create_consumer(node_topic, rpc_dispatcher, fanout=False)
    self.conn.create_consumer(self.topic, rpc_dispatcher, fanout=True)
    self.conn.consume_in_thread()
    self.manager.post_start_hook()
 
    LOG.debug(_("Join ServiceGroup membership for this service %s")
              % self.topic)
    self.servicegroup_api.join(self.host, self.topic, self)
 
    if self.periodic_enable:
        if self.periodic_fuzzy_delay:
            initial_delay = random.randint(0, self.periodic_fuzzy_delay)
        else:
            initial_delay = None
        self.tg.add_dynamic_timer(self.periodic_tasks,
                                 initial_delay=initial_delay,
                                 periodic_interval_max=
                                    self.periodic_interval_max)

　　self.conn.create_consumer(node_topic, rpc_dispatcher, fanout=False)。这里的self.conn主要指nova.openstack.common.rpc.impl_qpid.Connection，（此处设AMQP后端实现为qpid，这个可以在nova.conf中进行配置）所以调用的create_consumer代码如下：

def create_consumer(self, topic, proxy, fanout=False):
    proxy_cb = rpc_amqp.ProxyCallback(
        self.conf, proxy,
        rpc_amqp.get_connection_pool(self.conf, Connection))
    self.proxy_callbacks.append(proxy_cb)
 
    if fanout:
        consumer = FanoutConsumer(self.conf, self.session, topic, proxy_cb)
    else:
        consumer = TopicConsumer(self.conf, self.session, topic, proxy_cb)
 
    self._register_consumer(consumer)
 
    return consumer

　　该函数会根据具体的后端AMQP实现，向AMQP服务器发送队列和exchange创建(第一次会创建，后面会复用)和绑定请求，这样就可以将Consumer和具体的队列绑定并进行监听。self.conn.consume_in_thread函数就是负责启动Consumer线程，其使用evelent.spawn创建一个协程一直运行等待消息，在有消息到来时会创建新的协程运行远程调用的函数。当队列有消息到来时，调用proxy_cb进行处理，即ProxyCallback对象的__call__函数，代码如下：

nova/openstack/common/rpc/amqp.py

ProxyCallback类：
def __call__(self, message_data):
    if hasattr(local.store, 'context'):
        del local.store.context
    rpc_common._safe_log(LOG.debug, _('received %s'), message_data)
    self.msg_id_cache.check_duplicate_message(message_data)
    ctxt = unpack_context(self.conf, message_data)
    """解析消息"""
    method = message_data.get('method')
    args = message_data.get('args', {})
    version = message_data.get('version')
    namespace = message_data.get('namespace')
    if not method:
        LOG.warn(_('no method for message: %s') % message_data)
        ctxt.reply(_('No method for message: %s') % message_data,
                   connection_pool=self.connection_pool)
        return
    """处理消息"""
    self.pool.spawn_n(self._process_data, ctxt, version, method,
                      namespace, args)
 
def _process_data(self, ctxt, version, method, namespace, args):
        ctxt.update_store()
    try:
        rval = self.proxy.dispatch(ctxt, version, method, namespace,
                                   **args)
    ...

　　可以看到消息的处理主要是通过新开的协程（使用eventlet）来执行self._process_data函数，而self._process_data函数中主要调用了self.proxy.dispatch函数，那么这个dispatch函数的内容究竟是什么呢？

1、首先找到self.proxy

　　根据对self.proxy的层层追踪，其为上面ProxyCallback构造时传入，即start函数中的 rpc_dispatcher = self.manager.create_rpc_dispatcher(self.backdoor_port)，这里的self.manager在创建该Service的时候就已经设定为ConductorManager。ConductorManager的create_rpc_dispatcher函数代码如下：

def create_rpc_dispatcher(self, *args, **kwargs):
    #self.compute_task_mgr = ComputeTaskManager()
    kwargs['additional_apis'] = [self.compute_task_mgr]
    return super(ConductorManager, self).create_rpc_dispatcher(*args,
            **kwargs)

　　根据类的继承关系：

nova/manager.py

Manager类：
def create_rpc_dispatcher(self, backdoor_port=None, additional_apis=None):
    apis = []
    if additional_apis:
        apis.extend(additional_apis)
    base_rpc = baserpc.BaseRPCAPI(self.service_name, backdoor_port)
    apis.extend([self, base_rpc])
    serializer = objects_base.NovaObjectSerializer()
    return rpc_dispatcher.RpcDispatcher(apis, serializer)

nova/openstack/common/rpc/dispatcher.py

RpcDispatcher类
def __init__(self, callbacks, serializer=None):
    #callbacks为一个list，包含ComputeTaskManager,ConductorManager,BaseRPCAPI
    self.callbacks = callbacks
    if serializer is None:
        serializer = rpc_serializer.NoOpSerializer()
    self.serializer = serializer
    super(RpcDispatcher, self).__init__()

　　所以最后的self.proxy为RpcDispatcher对象。要注意的是RpcDispatcher对象的callbacks属性包含了一系列的Manager类，它们将被用来处理消息中指定的method。参考下图：

2、分析self.proxy.dispatch函数

　　即RpcDispatcher类的dispatch函数，代码如下：

def dispatch(self, ctxt, version, method, namespace, **kwargs):
    if not version:
        version = '1.0'
    had_compatible = False
    for proxyobj in self.callbacks:
        try:
            cb_namespace = proxyobj.RPC_API_NAMESPACE
        except AttributeError:
            cb_namespace = None
        if namespace != cb_namespace:
            continue
        ...
        if not hasattr(proxyobj, method):
            continue
        if is_compatible:
            kwargs = self._deserialize_args(ctxt, kwargs)
            result = getattr(proxyobj, method)(ctxt, **kwargs)
            return self.serializer.serialize_entity(ctxt, result)
    ...

　　该函数会根据namespace匹配Manager，然后分析Manager类中的函数是否和消息中的函数匹配，如果匹配即调用Manager类中对应的函数进行处理。经过观察，其实对外提供rpc服务的组件的rpcapi.py和manager.py文件中的类是对应的，它们通过namespace进行匹配查找，在命名上具有相同的前缀，所以如果要跟踪rpcapi.py中函数的后续实现只需查看该组件的manager.py文件里对应的Manager类的对应函数就可以了。

参考文章：

http://bingotree.cn/?p=242

http://blog.csdn.net/gaoxingnengjisuan/article/details/12231633