You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@hive.apache.org by "Sergey Shelukhin (JIRA)" <ji...@apache.org> on 2017/10/04 19:48:00 UTC
[jira] [Updated] (HIVE-17694) restrict access to sysdb in the
non-Ranger case
[ https://issues.apache.org/jira/browse/HIVE-17694?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Sergey Shelukhin updated HIVE-17694:
------------------------------------
Summary: restrict access to sysdb in the non-Ranger case (was: restrict access to sysdb in a non-Ranger case)
> restrict access to sysdb in the non-Ranger case
> -----------------------------------------------
>
> Key: HIVE-17694
> URL: https://issues.apache.org/jira/browse/HIVE-17694
> Project: Hive
> Issue Type: Bug
> Reporter: Sergey Shelukhin
> Assignee: Gunther Hagleitner
>
> With ranger, sysdb access is restricted by ranger based on the database type in table properties (as far as I understand, that already works).
> It might make sense to restrict access w/o ranger, given that one can probably inject malicious query, or at least gain access to some real data like partition key-values, values from column stats, etc, thru the sql string.
> Not sure if a doAs mechanism is possible or easy to implement... some Hadoop features allow specifying a set of users or groups in configuration (e.g. IPC acls), so we can do that with a restricted config, and check the UGI. Alternatively we can just add an off switch and restrict access, however it's not very convenient with a restricted config, cause the only way to make changes for the admin would be to temporarily give access to everyone (or again everyone would be able to change it for their session).
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)