[zz]hdfs-over-ftp安装

amy_888

原文见：http://nubetech.co/accessing-hdfs-over-ftp
这个程序是通过hdfs的9000端口访问的。听说还有hadoop自己的拓展包，需要重新编译hadoop。有机会的话安装一次来对比一下效率。
下载压缩包：hdfs-over-ftp-0.20.0.tar.gz（我的hadoop是0.20.2）

1.解压之后在目录下执行
./register-user.sh username password >> users.conf
这会在users.conf中生成新的ftp账户配置。
废话一句，xxxx.homedirectory=/，这里的/就是你的hadoop的根目录。
2.修改hdfs-over-ftp.conf，有两个要注意的地方：
hdfs-uri = hdfs://localhost:9000###确认localhost:9000可以访问到你的hadoop。
superuser = hadoop#####hadoop是不是你起hadoop服务的用户。
3.修改log4j.conf的
log4j.appender.R.File=xxxxxxx

4.启动/关闭：
sudo ./hdfs-over-ftp.sh start
###必须使用sudo，没有sudo权限的话，修改/etc/sudoers ，在root下一行添加。
username    ALL=(ALL)       ALL
sudo ./hdfs-over-ftp.sh stop



　　The Hadoop Distributed File System provides different interfaces so that clients can interact with it. Besides the HDFS shell, the file system exposes itself through WebDAV, Thrift, FTP and FUSE. In this post, we access HDFS over FTP. We have used Hadoop 0.20.2.
　　1. Download the hdfs-over-ftp tar from https://issues.apache.org/jira/secure/attachment/12409518/hdfs-over-ftp-0.20.0.tar.gz
　　2. Untar hdfs-over-ftp-0.20.0.tar.gz.
　　3. We now need to create the configuration with ftp username and password.

./register-user.sh username password >> users.conf
　　 # the username user
ftpserver.user.username.userpassword=0238775C7BD96E2EAB98038AFE0C4279
ftpserver.user.username.homedirectory=/
ftpserver.user.username.enableflag=true
ftpserver.user.username.writepermission=true
ftpserver.user.username.maxloginnumber=0
ftpserver.user.username.maxloginperip=0
ftpserver.user.username.idletime=0
ftpserver.user.username.uploadrate=0
ftpserver.user.username.downloadrate=0
ftpserver.user.username.groups=users

　　4. Configure log4j.conf so that you can diagnose whats happening.
　　5. Now make changes according to your requirement in hdfs-over-ftp.conf

#uncomment this to run ftp server
port = 21
data-ports = 20
　　#uncomment this to run ssl ftp server
#ssl-port = 990
#ssl-data-ports = 989
　　# hdfs uri
hdfs-uri = hdfs://localhost:9000
　　# max number of login
max-logins = 1000
　　# max number of anonymous login
max-anon-logins = 1000
　　 # have to be a user which runs HDFS
# this allows you to start ftp server as a root to use 21 port
# and use hdfs as a superuser
superuser = hadoop

Please provide hdfs-uri according to your requirement.
　　6. Now start ftp server:
  sudo ./hdfs-over-ftp.sh start

7. To login to hdfs as ftp client
  ftp {ip address of namenode machine}
(Note- Use username and password which you registered in user.conf)
　　8. To put file or write in any folder of hdfs you need to provide permission to your user through hadoop ‘chown’ command

bin/hadoop fs -chown -R group:username {path}

10. You can stop the ftp server by
  sudo ./hdfs-over-ftp.sh stop