因为报表前端用dot NET MVC写的,要想从HIVE中获取详细数据,目前的方案是使用hive thriftserver。
1. 生成HIVE Thrift对应的C#类文件
遇到的问题是找不到thriftserver.thrift中引用的fb303.thrift文件。解决方案:从thrift源码目录中复制一份过来。然后利用thriftserver.thrift生成了4个类文件,复制到项目中编译报错,不得不将引用的其他三个文件分别利用thrift生成C#类,最后将几十个类文件复制到项目中,修改项目属性,将framework 4 client profile修改为framework 4,编译通过。
client代码如下:
static void Main(string[] args)
{
TTransport transport = new TSocket("192.168.1.1", 10000);
TProtocol protocol = new TBinaryProtocol(transport);
ThriftHive.Client client = new ThriftHive.Client(protocol);
transport.Open();
//client.execute("add file /data/home/script/ad_resolve2.py;");
client.execute("select * from web_pv_log_detail3 where dt = '2012-09-10' limit 10");
Console.WriteLine("the result is:");
var items = client.fetchAll();
foreach (var item in items)
{
Console.WriteLine(item);
}
transport.Close();
Console.ReadLine();
}
{
TTransport transport = new TSocket("192.168.1.1", 10000);
TProtocol protocol = new TBinaryProtocol(transport);
ThriftHive.Client client = new ThriftHive.Client(protocol);
transport.Open();
//client.execute("add file /data/home/script/ad_resolve2.py;");
client.execute("select * from web_pv_log_detail3 where dt = '2012-09-10' limit 10");
Console.WriteLine("the result is:");
var items = client.fetchAll();
foreach (var item in items)
{
Console.WriteLine(item);
}
transport.Close();
Console.ReadLine();
}
2. 启动服务端服务:
hive --service hiveserver
3. 测试。发现多次调用execute之后调用fetch,每次会获取到最后一个execute的结果,对于多线程调用的可用性持怀疑态度。