Cloudera Docs

Known issues in Cloudera Data Catalog (1.5.4-CHF3)

This section lists issues fixed in this release of the Cloudera Data Catalog service.

CDPDSS-3346: Profiler pods are stuck in init state
The CDP_ACCESS_KEY_ID and CDP_PRIVATE_KEY keys do not receive values when upgrading from Cloudera Data Catalog 1.5.2 to 1.5.4.
  1. Use the following command to restart the DSS App: kubectl rollout restart deployment/cdp-release-dssapp -n <namespace>
  2. Delete and relaunch the profilers after the upgrade.
CDPDSS-3353: Profiler Configuration page is giving error, if browser's preferred language is not UK English
The Profilers > Configs is may be incorrectly rendered if the web browser's preferred language is not set to en-gb.
CDPDSS-2913: Asset Details view does not provide details of the asset - exception
Entities without lineage information may display the get linage request failed Error from Atlas Not Found : Received 404.
CDPDSS-3057: Failed profiler job because of missing columns in log entries.
When the logs to be profiled by the Ranger Audit Profiler have missing columns, the profiling job fails. The ranger audit profiler pod log contains the following message:Ranger Audit Profiler failed: Failed to run Ranger Audit Profiler. Error: "['column1', 'column2', ...] not in index".

column1, column2 show the missing column entries in the logs.

Each Ranger audit profiler job has 3 sub-jobs (identified by the same job ID prefix), any of which could fail because of the above reason. It is possible that one or even two sub-jobs pass till the above scenario is encountered.

None
CDPDSS-3472: Profiler jobs are showing up twice when two data lakes are present
Cloudera Data Catalog does not support more than one data lakes (or base clusters). Connecting two data lakes will result in all profiler jobs showing as duplicated in the Profilers page.
None
CDPDSS-3381: The Ranger Audit Profiler is showing status as FINISHED/SUCCESS even though there is error with the profilers
The Ranger Audit Profiler shows the incorrect FINISHED/SUCCESS status when the NameNode is stopped.
None
CDPDSS-3348: Only Dataset owners should be allowed to delete their datasets
Datasets can be deleted by anyone who can access the Datasets page.
None
CDPDSS-3081: Datasets: Tenant Schema is getting passed as workload username while impersonating a user on ATLAS and Ranger API calls
While making API calls to Apache Atlas and Apache Ranger, there is a potential bug in code where a wrong value for workloadUserName is getting passed which causes the APIs to perform an auth check with a wrong parameter, causing unwanted issues.
None
CDPDSS-3080: Deleted Entities in Atlas cannot be removed from Datasets
Entities deleted in Apache Atlas cannot be removed from Cloudera Data Catalog datasets.
None
CDPDSS-3475: Not able to see the profiled data of 'Ranger Audit Profiler' if the Active NN (NameNode) switched over to standby
When the ActiveNN is set to Standby and the StandbyNN is set to Active, the profiled data of the Ranger Audit Profiler is not visible in Asset Details for any Hive asset.
Due to an exception in the pod logs, the job status was previously being marked as successful. As a result, the audit job assumed that no log files were available for profiling. To resolve this, the job status should be marked as failed when the exception occurs.
CDPDSS-3395: Incorrect results are being displayed after navigation, when "owners" filter is selected
In the Search page, when returning to the list of results after opening the Asset Details of one the results and repeating the search with the same filter, other entities belonging to owners different than the selected can appear.
CDPDSS-2688: Hive Column On-demand Profiler is failing but the Asset Details page is not updated with 'Failed State'
A failed Hive Column On-Demand Profiler job is incorrectly not updated to Failed state in Asset Details > Profilers.
CDPDSS-3509: Only power users should only be able to launch profiler and ODP profilers
When updating the roles of a user with the role poweruser to environment user, the user is still able to start on-demand profiling jobs. Starting on-demand profiling jobs should only be available for power users, not to environment users.