[
https://issues.apache.org/jira/browse/KUDU-2973?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17034844#comment-17034844
]
ASF subversion and git services commented on KUDU-2973:
-------------------------------------------------------
Commit f392503972e41fe57d80fefb3779a82c9a92229f in kudu's branch
refs/heads/master from Attila Bukor
[ https://gitbox.apache.org/repos/asf?p=kudu.git;h=f392503 ]
KUDU-2973 Add semi-database support for Ranger
The table names are "normalized" if HMS integration is enabled so that
they can be synchronized to HMS. Hive's table/database identifiers are
pretty restrictive (alphanumerical ASCII characters + underscore and
forward slash) and it's case-insensitive. As Kudu doesn't support
databases it imposes a further restriction: the table identifiers to be
synchronized with HMS have to contain exactly one "." character which
separates the database name from the table name in HMS.
Ranger's restrictions regarding identifiers are more lax, can be case
sensitive (will not be used for Kudu) and is not limited to ASCII.
This commit introduces a method to parse database and table name for
Ranger from a table identifier that works slightly different from Hive's
table name parser, as it allows non-ASCII characters and provides a
configurable default database name when the table name doesn't contain a
period.
Change-Id: I11431ff5bc75540edff56ef3d4ad384fa37d33d5
Reviewed-on: http://gerrit.cloudera.org:8080/15018
Tested-by: Kudu Jenkins
Reviewed-by: Alexey Serbin <[email protected]>
Reviewed-by: Hao Hao <[email protected]>
> Support semi-database concept even without HMS integration
> ----------------------------------------------------------
>
> Key: KUDU-2973
> URL: https://issues.apache.org/jira/browse/KUDU-2973
> Project: Kudu
> Issue Type: Sub-task
> Affects Versions: 1.11.0
> Reporter: Hao Hao
> Assignee: Attila Bukor
> Priority: Major
>
> For Ranger integration, we need to continue to have semi-database support.
> And currently it is tied to the HMS integration, which needs to be separated
> out. This includes to extract database information from table name for
> retrieving corresponding Ranger policies. For example, "db.table" belongs to
> 'db'. And as Kudu table name is case sensitive and can have special
> character, database "table" will be considered as 'table' that belongs to
> 'default' database, and "db.table.abc" will be considered as 'table.abc' that
> belongs to 'db' database.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
