Support using Hadoop's proxyuser mechanism to access kerberos-enabled Hadoop clusters

[iBuddha]

HDFS

比如在集群中添加以下配置
For example, add the following configuration to the cluster

   <property>
     <name>hadoop.proxyuser.linkis.hosts</name>
     <value>*</value>
   </property>
   <property>
     <name>hadoop.proxyuser.linkis.groups</name>
     <value>*</value>
   </property>

In this way, you only need to log in with the keytab and principal of the user linkis, create UGI, and then use UserGroupInformation.createProxyUser(userName, proxyUGI) to create the UGI used when the user accesses HDFS. It is no longer necessary to generate a keytab for each user
这样只需要用linkis这个用户的keytab和principal登录，创建UGI， 然后再用UserGroupInformation.createProxyUser(userName, proxyUGI)来创建用户访问HDFS时候使用的UGI。而不再需要为每个用户生成keytab

YARN

When submitting the Spark program, you can also use Spark's --proxy-user configuration. In this way, the ticket of the user who started linkis can be used to submit the spark program to YARN, but the user who actually runs spark is the identity of the end user. You can avoid creating a keytab for each user.
在提交Spark程序时候，也可以用Spark的--proxy-user配置。这样就可用启动linkis的用户的ticket 来提交spark程序到YARN， 但是实际运行spark的用户身份为终端用户的身份。可以避免为每个用户创建keytab。