Known Issues in Ranger
Learn about the known issues in Ranger, the impact or changes to the functionality, and the workaround.
- CDPD-66092: Upgrade from CDP-7.1.8 to 7.1.9 fails, Ranger shows no policies at all, due to Java patches not being applied during upgrade. This causes rolling restart of services to fail.
-
Skip applying any of the following java patches NOT applicable to the underlying environment. In other words, do not apply patches for a service definition that does not appear in ranger database.
Table 1. Java Patches and Related Service Definitions Java patch Servic-definition PatchForHiveServiceDefUpdate_J10027.java Hive PatchForTagServiceDefUpdate_J10028.java Tag PatchForHBaseServiceDefUpdate_J10035.java HBase PatchForHdfsAddChainedPluginProvider_J10038.java HDFS PatchForHdfsRemoveChainedPluginProvider_J10039.java HDFS PatchForOzoneServiceDefUpdate_J10041.java Ozone PatchForOzoneServiceDefConfigUpdate_J10051.java Ozone PatchForOzoneServiceDefUpdate_J10057.java Ozone - CDPD-60489: Jackson-dataformat-yaml 2.12.7 and Snakeyaml 2.0 are not compatible.
- You must not use Jackson-dataformat-yaml through the platform for YAML parsing.
- CDPD-56803: When there is no existing policy for user and a revoke request comes from hbase, then will get this error.
-
hbase:001:0> revoke 'hrt_11' ERROR: org.apache.hadoop.hbase.coprocessor.CoprocessorException: HTTP 400 Error: processSecureRevokeRequest processing failed at org.apache.ranger.authorization.hbase.RangerAuthorizationCoprocessor.preRevoke(RangerAuthorizationCoprocessor.java:1309) at org.apache.ranger.authorization.hbase.RangerAuthorizationCoprocessor.preRevoke(RangerAuthorizationCoprocessor.java:1128) at org.apache.hadoop.hbase.master.MasterCoprocessorHost$162.call(MasterCoprocessorHost.java:1857) at org.apache.hadoop.hbase.master.MasterCoprocessorHost$162.call(MasterCoprocessorHost.java:1854) at org.apache.hadoop.hbase.coprocessor.CoprocessorHost$ObserverOperationWithoutResult.callObserver(CoprocessorHost.java:558) at org.apache.hadoop.hbase.coprocessor.CoprocessorHost.execOperation(CoprocessorHost.java:631) at org.apache.hadoop.hbase.master.MasterCoprocessorHost.preRevoke(MasterCoprocessorHost.java:1854) at org.apache.hadoop.hbase.master.MasterRpcServices.revoke(MasterRpcServices.java:2740) at org.apache.hadoop.hbase.shaded.protobuf.generated.MasterProtos$MasterService$2.callBlockingMethod(MasterProtos.java) at org.apache.hadoop.hbase.ipc.RpcServer.call(RpcServer.java:387) at org.apache.hadoop.hbase.ipc.CallRunner.run(CallRunner.java:139) at org.apache.hadoop.hbase.ipc.RpcExecutor$Handler.run(RpcExecutor.java:369) at org.apache.hadoop.hbase.ipc.RpcExecutor$Handler.run(RpcExecutor.java:349)
- CDPD-56741: Improvement in log message when jwtauth not used
- The following exception is printed at startup only and it is not cluttering logs:
- CDPD-56738: Ranger RMS showing FileNotFoundException: /usr/share/java/oraclepki.jar in Oracle 19 setup
- This is a warning log printed in catalina.out file when Ranger RMS server is initialised. Following exception is observed only in Oracle 19 setup. FileNotFoundException: /usr/share/java/oraclepki.jar
- CDPD-55107: Not able to search using muliple user filter in access audit tab
- If you were using multiple user search filters in Audit > Access Tab on Ranger Admin UI, after upgrading to CDP-7.1.9 that would not be supported. You can continue to search user with single search filter.
- CDPD-48975: Ranger KMS KTS to KMS DB migration : keys with the same name but different case are not migrated
- KMS keys are not case sensitive
- CDPD-58704: hadoop roll key / key delete command shows operation failed error when one KMS host is down, even when operation succeeds
- In case of rollover/delete, client sends one more (last after delete request) request to KMS instances to clean their cache and that too to all registered kms instances. if one KMS instance is stopped ( not deleted), client gets runtime exception.
- CDPD-41582: Atlas Resource Lookup : Classification for "entity-type" lists only classification for the following payload:
-
{"resourceName": "classification", "userInput": "", "resources": {"classification": []}}]
expectation is to return all the classifications . But the response has only "classification"Happens similarly for entity-label , entity-business-metadata.
- CDPD-42598: Kafka policy creation allowed with incorrect permissions.
-
When creating a Kafka policy from the UI, the permissions "Idempotent write"and "Cluster action" are not displayed as they are not applicable for the "topic" resource, but when creating a policy for the "topic" resource with the permissions "Idempotent write" and "Cluster Action", the policy is created successfully when the expected behaviour is that the policy creation must fail as the permission is not applicable for the Kafka topic resource
- CDPD-40734: User allowed to insert data into a hive table when there is a deny policy on a table column.
-
A user is allowed to enter data into a table even if there is a deny policy present on one of the table columns.
Test scenario details:Policy setup :- policy 1 :- all access policy for hrt_qa, hive and impala users resources - database - * , table - *, column - * users : hrt_qa, hive, impala access - all access allowed policy 2 :- policy on test_1.table_1 for hrt_5 users : hrt_5 resources : database - test_1, table - table_1, column - * access :- all access allowed policy 3 :- deny policy on test_1.table_1.c0 for hrt_5 users : hrt_5 resources : database - test_1, table - table_1, column - c0 access - all access denied data setup :- database - test_1 table - table_1(c0 int, c1 int)
The user is able to insert data into the table.
- CDPD-58860: After upgrading CDP-7.1.8 to CDP-7.1.9, cdp-proxy token missing from knox-ranger policy.
-
As part of OPSAPS-67480 in 719 default ranger policy is added from cdp-proxy-token topology, so that after a new installation of CDP-7.1.9, the knox-ranger policy includes cdp-proxy-token. However, upgrades do not add cdp-proxy-token to cm_knox policies automatically.
- CDPD-68806: The Revoke operation for users belonging to a group or role permission does not function as expected.
-
List command is listing all the tables even when the user permission is revoked. And also the command does not add any deny policy to Ranger for that specific user.
For example:
user_1 is the part of group_1
user_2 is the part of role_1
Create a Ranger policy for group_1 and role_1 to access all the resources in HBase.
Try to revoke the permission for user_1 and user_2 using HBase shell global revoke command:
revoke 'user_1'
revoke 'user_2'
But there is no deny policy added for that user. And access is still granted to user_1 and user_2.
- CDPD-64939: Impala behavior differences across 717 SP3 and 719 SP1 releases.
- Upgrading from Cloudera Runtime 7.1.7 Service Pack 3 to 7.1.9 Service Pack 1, for a GRANT statement involving multiple columns in a table, Impala service creates one Ranger policy for each column.
- CDPD-68739: The revoke command does not work when using the HBase shell
-
While using the HBase shell, running the revoke command does not cancel the user permission. Users are able to perform actions even after running the revoke command.
For example:
When a user is granted permission to a particular table using the HBase grant command.
hbase:003:0> grant 'user_1', 'RWCA', 'some_abc_table'
And then execute the global revoke command for that user:
hbase:005:0> revoke 'user_1'
Permission for that user 'user_1' for the table 'some_abc_table' not got revoked.
- CDPD-67238: Multiple Columns Revoke not generating policies with correct number of columns
-
As an example, when "revoke select(col1, col2,col3) on table demo.test from role Role3;" is done, the generated policy does not revoke the columns. Currently the revoke statement is only revoking if there is only one column.
- CDPD-69412: revokeAccess() behaves differently from secureRevokeAccess() after RANGER-4638
- In a non-Kerberized environment, you must use within revokeAccess(), "!isPolicyResourceSameAsRevokedResource" as the condition as it is used in secureRevokeAccess().