Status of Hive Authorization before hive 0.13

The default authorization in hive is not designed with the intent to protect against malicious users from accessing data they should not be accessing. It only helps in preventing users from accidentally doing operations they are not supposed to. It is also incomplete because it does not have authorization checks for many operations including the grant statement. The authorization checks happen during hive query compilation. But as the user is allowed to execute dfs commands, user defined functions and shell commands, it is possible to bypass the client security checks.

Hive also has support for storage based authorization, which is commonly used to add authorization to metastore server api calls. It can now (as of hive 0.12?) be used on the client side as well. While it can protect metastore against changes by malicious users, it does not support fine grained access control (column or row level).

The default authorization model in hive can be used to provide fine grained access control by creating views and granting access to views instead of the underlying table.

SQL Standards based hive authorization

SQL standards based authorization option (introduced in hive 0.13) provides a third option for authorization in hive. This is recommended because it allows Hive to be fully SQL compliant in its authorization model without causing backward compatibility issues for current users. As users migrate to this more secure model, the current default authorization could be deprecated. This authorization mode can be used in conjunction with storage based authorization on the metastore server. Like the current default authorization in hive, this will also be enforced at query compilation time. To provide security through this option, the client will have to be secured. This can be done by allowing users access only through hive server2, and by restricting the user code and non sql commands that can be run. The checks will happen against the user who submits the request, but the query will run as the hive server user. The directories and files for input data would have read access for this hive server user. For users who don’t have the need to protect against malicious users, this could potentially be supported through the hive command line as well.

The goal of this work has been to comply with SQL standard as far as possible, but there are deviations from the standard in the implementation. Some deviations were made to make it easier for existing hive users to migrate to this authorization model, some were made considering ease of use (in such cases we also looked at what many widely used databases do).

Under this authorization model, users who have access to hive-cli, hdfs commands, pig commandline, 'hadoop jar' command etc are considered privileged users. In an organization, it is typically only the teams that work on ETL workloads that need such access. These tools don't access the data through HiveServer2, and as a result their access is not authorized through this model. For hive-cli, pig and mapreduce users, access to hive tables can be controlled using storage based authorization enabled on the metastore server.

Most users such as business analysts tend to use SQL and odbc/jdbc through HiveServer2 and their access can be controlled using this authorization model.

Restrictions on hive commands,statements

Commands such as dfs,add,delete,compile,reset are disabled when this authorization is enabled.

The set commands used to change hive configuration are restricted to a smaller safe set. This is controlled using hive.security.authorization.sqlstd.confwhitelist configuration parameter. If this set needs to be customized, the HiveServer2 admin can set a value for this configuration parameter in its hive-site.xml.

Privilege to add/drop functions and macros are restricted to the admin role.

To enable users to use functions, the ability to create permanent functions has been added. A user in admin role can run commands to create these functions, that all users can then use.

Privileges

● SELECT privilege - gives read access to object

● INSERT privilege - gives ability to add data to object (table)

● UPDATE privilege - gives ability to run update queries on object (table)

● DELETE privilege - gives ability to delete data in object (table)

● ALL PRIVILEGES - gives all privileges

Objects

● The privileges will apply to table and views. The above privileges are not supported on databases.

Object ownership

For certain actions, the ownership of the object (table/view/database) determines if you are authorized to perform the action.

Users and Roles

Privileges can be granted to users as well as roles.
Users can belong to one or more roles.

There are two roles with special meaning - *public* and *admin*.
All users belong to the *public* role. You use this role in your grant statement to grant the privilege to all users.

When a user runs a hive query/command, the privileges granted to the user and her ‘current roles’ are checked. The current roles can be seen using the ‘show current roles;’ command. All roles except for the *admin* role will be in the current roles by default.
To set a specific role as the current role, you need to use the ‘set role’ command. See the command description for details.

Users who do the work of a ‘data base administrator’ are expected to be added to the *admin* role.
They have privileges for running additional commands such as ‘create role’ and ‘drop role’. They can also access objects that they haven’t been given explicit access to. However, a user who belongs to the admin role needs to run the ’SET ROLE’ command before getting the privileges of the admin role, as this role is not in current roles by default.

Role names are case insensitive. ie, “marketing” and “MarkEting” refer to same role.
User names are case sensitive. This is because, unlike role names, user names are not managed within hive.

Role management commands

CREATE ROLE <rolename>;
Creates a new role. Only admin role has privilege for this.
The role names ALL, DEFAULT and NONE are reserved.

DROP ROLE <rolename>;
Drops the given role. Only admin role has privilege for this.

SHOW CURRENT ROLES;
Shows the list of current roles. All actions of the user are authorized by looking at the privileges of the user and all current roles of the user.
The default current roles has all roles for the user except for the admin role (even if the user belongs to the admin role as well).
Any user can run this command.

SET ROLE (<rolename>|ALL)
If a <rolename> is specified, then that role becomes the only role in current roles.
Setting role name to ALL refereshes the list of current roles (in case new roles were granted to the user) and sets them to default list of roles.

If a role the user does not belong to is specified as the <rolename>, it will result in an error.

SHOW ROLES
List all currently existing roles.
Only admin role has privilege for this.

GRANT ROLE <role_name> [, <role_name>] ...
TO principal_specification [, principal_specification] ...
[ WITH ADMIN OPTION ]

principal_specification
: USER user
| ROLE role

Grant one or more roles to other roles or users.
If “WITH ADMIN OPTION” is specified, then the user gets privileges to grant the role to other users/roles.
If the grant statement ends up creating a cycling relationship between roles, the command will fail with an error.

REVOKE ROLE <role_name> [, <role_name>] ...
FROM principal_specification [, principal_specification] ...

principal_specification
: USER user
| ROLE role

Revokes the membership of the roles from the user/roles in the FROM clause.
Currently revoking just the ADMIN OPTION is not possible. You need to revoke the principal’s role grant, and then grant the role to principal without admin option.

SHOW ROLE GRANT (USER|ROLE) <principalname>
Where principal name is the name of a user or role.

Lists all roles the given user or role has been granted.
Currently any user can run this command. But this is likely to change in future to allow users to see only their own role grants, and additional privileges would be needed to see role grants of other users.

0: jdbc:hive2://localhost:10000> GRANT role1 TO USER user1 ;
No rows affected (0.058 seconds)

0: jdbc:hive2://localhost:10000> SHOW ROLE GRANT USER user1
+---------+---------------+----------------+----------+
| role   | grant_option |   grant_time   | grantor |
+---------+---------------+----------------+----------+
| public | false         | 0              |          |
| role1   | false         | 1398284083000 | uadmin   |
+---------+---------------+----------------+----------+

SHOW PRINCIPALS <rolename>
Lists all roles and users who belong to this role.
Only admin role has privilege for this.

0: jdbc:hive2://localhost:10000> SHOW PRINCIPALS role1;
+-----------------+-----------------+---------------+----------+---------------+----------------+
| principal_name | principal_type | grant_option | grantor | grantor_type |   grant_time   |
+-----------------+-----------------+---------------+----------+---------------+----------------+
| role2           | ROLE            | false         | uadmin   | USER          | 1398285926000 |
| role3           | ROLE            | true          | uadmin   | USER          | 1398285946000 |
| user1           | USER            | false         | uadmin   | USER          | 1398285977000 |
+-----------------+-----------------+---------------+----------+---------------+----------------+

Managing object privileges

GRANT

priv_type [, priv_type ] ...

ON <table_or_view_name>

TO principal_specification [, principal_specification] ...

[WITH GRANT OPTION]

REVOKE

priv_type [, priv_type ] ...

ON <table_or_view_name>

FROM principal_specification [, principal_specification] ...


principal_specification
  : USER user
  | ROLE role

If a user is granted a privilege WITH GRANT OPTION on a user or a role, then the user can also grant other users the privileges it has on those tables.

Note that in case of the REVOKE statement, the DROP-BEHAVIOR option of CASCADE is not currently supported (which is in SQL standard). As a result, the revoke statement will not drop any dependent privileges. For details on CASCADE behavior, you can check the postgres revoke documentation.

Configuration

Set the following to hive-site.xml
hive.server2.enable.doAs to false.
hive.users.in.admin.role to the list of comma separated users who need to be added to admin role

Start hiveserver2 with the following additional commandline options -

-hiveconf hive.security.authorization.manager=org.apache.hadoop.hive.ql.security.authorization.plugin.sqlstd.SQLStdHiveAuthorizerFactory

-hiveconf hive.security.authorization.enabled=true

-hiveconf hive.security.authenticator.manager=org.apache.hadoop.hive.ql.security.SessionStateUserAuthenticator.

-hiveconf hive.metastore.uris=' '

References

For information on the SQL standard for security see:

● ISO 9075 Part 1 Framework sections 4.2.6, 4.6.11

● ISO 9075 Part 2 Foundation sections 4.35 and 12

Space shortcuts

Child pages

Status of Hive Authorization before hive 0.13

SQL Standards based hive authorization

Restrictions on hive commands,statements

Privileges

Objects

Object ownership

Users and Roles

Role management commands

Managing object privileges

Configuration

Space shortcuts

Child pages

SQL Standard based hive authorization (New in Hive 0.13)

Status of Hive Authorization before hive 0.13

SQL Standards based hive authorization

Restrictions on hive commands,statements

Privileges

Objects

Object ownership

Users and Roles

Role management commands

Managing object privileges

Configuration