[SPARK-26929] table owner should use user instead of principal when use kerberos

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.3.0, 2.3.1, 2.3.2, 2.3.3, 2.4.0
• Fix Version/s: 3.0.0
• Component/s: SQL
• Assignee: hong dongdong
• Reporter: hong dongdong
• Created: 19/Feb/19 09:10
• Updated: 16/Sep/19 19:37
• Resolved: 16/Sep/19 18:08

Description

In kerberos cluster, when use spark-sql or beeline to create table,  the owner will be whole info of principal. the issue  was fixed in SPARK-19970 and modify by SPARK-22846, so it occur again. It will causes some problems when using role., and this time should resolved two issues together.

Use  org.apache.hadoop.hive.shims.Utils.getUGI  directly to get ugi.getShortUserName

instead of use  conf.getUser which return principal info.

Code change

private val userName: String = try {
val ugi = HiveUtils.getUGI
ugi.getShortUserName
} catch {
case e: LoginException => throw new IOException(e)
}

Berfore

scala> sql("create table t(a int)").show
 scala> sql("desc formatted t").show(false)
 ...
|Owner:|[e-mail]| |

After:

 scala> sql("create table t(a int)").show
 scala> sql("desc formatted t").show(false)
 ...
|Owner:|spark| |

Issue Links

• is duplicated by: SPARK-28559 when execute "create table" on ThriftServer ,the hive table owner is wrong
• links to: GitHub Pull Request #23837
• links to: GitHub Pull Request #23952