通过Thrift source向Flume发送数据的Python实现

来源:互联网 发布:优酷福利待遇 知乎 编辑:程序博客网 时间:2024/05/17 18:01

目前Flume支持Thrift source,即通过一个Thrift服务来收集数据(这一点和scribe是一样的),然后通过相应的channel发送到sink中去。以下是具体的实现过程:


环境:Python 2.7.5/CDH4.3 Flume 1.3/Thrift 0.9/
首先,我们需要一个Thrift协议的Python Flume客户端的模块,这个模块可以根据Thrift的定义自动生成。你应该先从Cloudera的网站上下载到CDH4.3中的Flume tarball :

wget http://archive.cloudera.com/cdh4/cdh/4/



下载到本地之后解压,在目录flume-ng-sdk\src\main\thrift下有Thrift对应的定义文件,并用它来生成对应的客户端模块:
tar xzvf flume-ng-1.3.0-cdh4.3.0.tar.gzcd apache-flume-1.3.0-cdh4.3.0-bin\flume-ng-sdk\src\main\thriftthrift --gen py flume.thrit



你会在当前目录下得到一个叫做gen-py的目录,我们将其更名为genpy之后,放到Python的系统模块路径中去:
mv gen-py/ /usr/local/lib/python2.7/site-packages/genpy




此时,你就可以通过以下过程来引用这个模块了:
[GCC 4.4.7 20120313 (Red Hat 4.4.7-3)] on linux2Type "help", "copyright", "credits" or "license" for more information.>>> from genpy import flume>>> dir(flume)['__all__', '__builtins__', '__doc__', '__file__', '__name__', '__package__', '__path__']>>> 



下面,就是利用这模块来封装一个客户端模块,注意Flume的Thrift source服务端使用的协议是继承自TCompactProtocol的TTupleProtocol:
public final class TTupleProtocol extends TCompactProtocol {...


在Thrift Python模块中,只有两种可选协议:TCompactProtocol, TBinaryProtocol, 很显然我们需要使用前一种协议,如果使用TBinaryProtocol,会在服务器端报以下错误:
18 Jul 2013 18:25:29,447 ERROR [pool-5-thread-4] (org.apache.thrift.server.TThreadPoolServer$WorkerProcess.run:213)  - Thrift error occurred during processing of message.org.apache.thrift.protocol.TProtocolException: Expected protocol id ffffff82 but got ffffff80        at org.apache.thrift.protocol.TCompactProtocol.readMessageBegin(TCompactProtocol.java:472)        at org.apache.thrift.TBaseProcessor.process(TBaseProcessor.java:27)        at org.apache.thrift.server.TThreadPoolServer$WorkerProcess.run(TThreadPoolServer.java:206)        at java.util.concurrent.ThreadPoolExecutor$Worker.runTask(ThreadPoolExecutor.java:895)        at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:918)        at java.lang.Thread.run(Thread.java:662)




在客户端会有以下的报错
Traceback (most recent call last):  File "pyflume.py", line 65, in <module>    flume_client.send({'a':'hello', 'b':'world'}, 'events under hello world')  File "pyflume.py", line 53, in send    self.client.append(event)  File "/usr/local/lib/python2.7/site-packages/genpy/flume/ThriftSourceProtocol.py", line 49, in append    return self.recv_append()  File "/usr/local/lib/python2.7/site-packages/genpy/flume/ThriftSourceProtocol.py", line 60, in recv_append    (fname, mtype, rseqid) = self._iprot.readMessageBegin()  File "/usr/local/lib/python2.7/site-packages/thrift/protocol/TBinaryProtocol.py", line 126, in readMessageBegin    sz = self.readI32()  File "/usr/local/lib/python2.7/site-packages/thrift/protocol/TBinaryProtocol.py", line 206, in readI32    buff = self.trans.readAll(4)  File "/usr/local/lib/python2.7/site-packages/thrift/transport/TTransport.py", line 58, in readAll    chunk = self.read(sz - have)  File "/usr/local/lib/python2.7/site-packages/thrift/transport/TTransport.py", line 271, in read    self.readFrame()  File "/usr/local/lib/python2.7/site-packages/thrift/transport/TTransport.py", line 275, in readFrame    buff = self.__trans.readAll(4)  File "/usr/local/lib/python2.7/site-packages/thrift/transport/TTransport.py", line 58, in readAll    chunk = self.read(sz - have)  File "/usr/local/lib/python2.7/site-packages/thrift/transport/TSocket.py", line 118, in read    message='TSocket read 0 bytes')thrift.transport.TTransport.TTransportException: TSocket read 0 bytes




以下就是具体的实现代码,大家可以参考修改:
#coding=utf-8'''Created on 2013-07-18@author: Felix'''from genpy.flume import ThriftSourceProtocolfrom genpy.flume.ttypes import ThriftFlumeEventfrom thrift.transport import TTransport, TSocketfrom thrift.protocol import TCompactProtocolclass _Transport(object):    def __init__(self, thrift_host, thrift_port, timeout=None, unix_socket=None):        self.thrift_host = thrift_host        self.thrift_port = thrift_port        self.timeout = timeout        self.unix_socket = unix_socket                self._socket = TSocket.TSocket(self.thrift_host, self.thrift_port, self.unix_socket)        self._transport_factory = TTransport.TFramedTransportFactory()        self._transport = self._transport_factory.getTransport(self._socket)            def connect(self):        try:            if self.timeout:                self._socket.setTimeout(self.timeout)            if not self.is_open():                self._transport = self._transport_factory.getTransport(self._socket)                self._transport.open()        except Exception, e:            print(e)            self.close()        def is_open(self):        return self._transport.isOpen()        def get_transport(self):        return self._transport        def close(self):        self._transport.close()        class FlumeClient(object):    def __init__(self, thrift_host, thrift_port, timeout=None, unix_socket=None):        self._transObj = _Transport(thrift_host, thrift_port, timeout=timeout, unix_socket=unix_socket)        self._protocol = TCompactProtocol.TCompactProtocol(trans=self._transObj.get_transport())        self.client = ThriftSourceProtocol.Client(iprot=self._protocol, oprot=self._protocol)        self._transObj.connect()            def send(self, event):        try:            self.client.append(event)        except Exception, e:            print(e)        finally:            self._transObj.connect()        def send_batch(self, events):        try:            self.client.appendBatch(events)        except Exception, e:            print(e)        finally:            self._transObj.connect()        def close(self):        self._transObj.close()    if __name__ == '__main__':    import random    flume_client = FlumeClient('192.168.1.141', 4141)    event = ThriftFlumeEvent({'a':'hello', 'b':'world'}, 'events under hello world2')    events = [ThriftFlumeEvent({'a':'hello', 'b':'world'}, 'events under hello world%s' % random.randint(0, 1000)) for _ in range(100)]        flume_client.send(event)    flume_client.send_batch(events)    flume_client.close()


    
        


以上代码也在github上:https://github.com/sinolambda/pyflume

原创粉丝点击