hadoop环境描述：
master节点：node1
slave节点：node2，node3，node4
远端服务器（python连接hive）：node29
需求：通过hive查询到cdn日志中指定时间段内url访问次数最多的前10个url
ps：用pig查询可以查询文章：
http://shineforever.blog.51cto.com/1429204/1571124
说明：python操作远程操作需要使用Thrift接口：
hive源码包下面自带Thrift插件：
[root@node1 shell]# ls -l /usr/local/hive-0.8.1/lib/py
total 28
drwxr-xr-x 2 hadoop hadoop 4096 Nov 5 15:29 fb303
drwxr-xr-x 2 hadoop hadoop 4096 Oct 15 10:30 fb303_scripts
drwxr-xr-x 2 hadoop hadoop 4096 Nov 5 15:29 hive_metastore
drwxr-xr-x 2 hadoop hadoop 4096 Oct 15 10:30 hive_serde
drwxr-xr-x 2 hadoop hadoop 4096 Nov 5 15:29 hive_service
drwxr-xr-x 2 hadoop hadoop 4096 Nov 5 15:20 queryplan
drwxr-xr-x 6 hadoop hadoop 4096 Nov 5 15:20 thrift
1)把相关文件scp到远端的node29相关目录下：
scp -r /usr/local/hive-0.8.1/lib/py/* 172.16.41.29:/usr/local/hive_py/.
2) 在node1服务器上开发hive：
[hadoop@node1 py]$ hive --service hiveserver
Starting Hive Thrift Server
3）在node29上编写查询脚本：
#!/usr/bin/env python
#coding:utf-8
#找出cdn日志指定时间段，url访问次数最多的前10个；
import sys
import os
import string
import re
import MySQLdb
#加载hive的python相关库文件；
sys.path.append('/usr/local/hive_py')
from hive_service import ThriftHive
from hive_service.ttypes import HiveServerException
from thrift import Thrift
from thrift.transport import TSocket
from thrift.transport import TTransport
from thrift.protocol import TBinaryProtocol
dbname="default"
hsql="select request,count(request) as counts from cdnlog where time >= '[27/Oct/2014:10:40:00 +0800]' and time <= '[27/Oct/2014:10:49
:59 +0800]' group by request order by counts desc limit 10"
def hiveExe(hsql,dbname):
try:
transport = TSocket.TSocket('172.16.41.151', 10000)
transport = TTransport.TBufferedTransport(transport)
protocol = TBinaryProtocol.TBinaryProtocol(transport)
client = ThriftHive.Client(protocol)
transport.open()
#加载增长表达式支持，必需(以下加载路径为远端hive路径，不是脚本的本地路径！)
client.execute('add jar /usr/local/hive-0.8.1/lib/hive_contrib.jar')
# client.execute("use " + dbname)
# row = client.fetchOne()
client.execute(hsql)
return client.fetchAll() #查询所有数据；
transport.close()
except Thrift.TException, tx:
print '%s' % (tx.message)
if __name__ == '__main__':
results=hiveExe(hsql,dbname)
num=len(results)
for i in range(num):
在node29上执行脚本，输入结果为：
wKioL1RbHoPT0rHZAAHHICCIv7U438.jpg
node1服务器上hive计算过程为：
wKiom1RbHjnhxxvdAAcmfK-RneY166.jpg
©著作权归作者所有：来自51CTO博客作者shine_forever的原创作品，如需转载，请注明出处，否则将追究法律责任