python读取hdfs上的parquet文件以及python访问hdfs HA的三种方法

python读取hdfs上的parquet文件

在使用python做大数据和机器学习处理过程中,首先需要读取hdfs数据,对于常用格式数据一般比较容易读取,parquet略微特殊。从hdfs上使用python获取parquet格式数据的方法(当然也可以先把文件拉到本地再读取也可以):

1、安装anaconda环境。

2、安装hdfs3。
      conda install hdfs3

3、安装fastparquet。
      conda install fastparquet

4、安装python-snappy。
      conda install python-snappy

5、读取文件

##namenode mode: from hdfs3 import HDFileSystem from fastparquet import ParquetFile  hdfs = HDFileSystem(host=IP, port=8020) sc = hdfs.open  pf = ParquetFile(filename, open_with=sc) df = pf.to_pandas()  ##返回pandas的DataFrame类型   ##HA mode: from hdfs3 import HDFileSystem from fastparquet import ParquetFile  host = "nameservice1" conf = {         "dfs.nameservices":"nameservice1",         ......  } hdfs = HDFileSystem(host = host, pars = conf) ......

python访问HDFS HA的三种方法

python访问hdfs常用的包有三个,如下:

1、hdfs3

其实从安装便捷性和使用上来说,并不推荐hdfs3,因为他的系统依赖和网络要求较高,但是某些情况下使用hdfs3会比较方便,官网资料点这里。如上面介绍,IP直接访问namenode:

from hdfs3 import HDFileSystem  hdfs = HDFileSystem(host=namenode, port=8020) hdfs.ls('/tmp')

HA访问:

host = "nameservice1" conf = {"dfs.nameservices": "nameservice1",         "dfs.ha.namenodes.nameservice1": "namenode113,namenode188",         "dfs.namenode.rpc-address.nameservice1.namenode113": "hostname_of_server1:8020",         "dfs.namenode.rpc-address.nameservice1.namenode188": "hostname_of_server2:8020",         "dfs.namenode.http-address.nameservice1.namenode188": "hostname_of_server1:50070",         "dfs.namenode.http-address.nameservice1.namenode188": "hostname_of_server2:50070",         "hadoop.security.authentication": "kerberos" } fs = HDFileSystem(host=host, pars=conf)   ##或者下面这种配置 host = "ns1" conf = {         "dfs.nameservices":"ns1", 		"dfs.ha.namenodes.ns1":"namenode122,namenode115", 		"dfs.namenode.rpc-address.ns1.namenode122":"nnlab01:8020", 		"dfs.namenode.servicerpc-address.ns1.namenode122":"nnlab01:8022", 		"dfs.namenode.http-address.ns1.namenode122":"nnlab01:50070", 		"dfs.namenode.https-address.ns1.namenode122":"nnlab01:50470", 		"dfs.namenode.rpc-address.ns1.namenode115":"nnlab02:8020", 		"dfs.namenode.servicerpc-address.ns1.namenode115":"nnlab02:8022", 		"dfs.namenode.http-address.ns1.namenode115":"nnlab02:50070", 		"dfs.namenode.https-address.ns1.namenode115":"nnlab02:50470", }  hdfs = HDFileSystem(host = host, pars = conf)

2、hdfs

这种方法在使用的时候配置比较简单,官网资料也比较丰富,但是需要注意的是该API可以模拟用户访问,权限较大。IP直接访问:

import hdfs  client = hdfs.client.InsecureClient(url="http://namenode:50070", user="hdfs")

HA访问:

import hdfs  client = hdfs.client.InsecureClient(url="http://namenode1:50070;http://namenode2:50070", user="hdfs")

3、pyhdfs

安装命令:pip install PyHDFS

官网地址,直接访问:

import pyhdfs  client = pyhdfs.HdfsClient(hosts="namenode:50070",user_name="hdfs")

HA访问

import pyhdfs  client = pyhdfs.HdfsClient(hosts=["namenode1:50070","namenode2:50070"],user_name="hdfs")