Known issues

Learn about the known issues in Data Catalog, the impact or changes to the functionality, and the workaround.

Known Issues

CDPDSS-2707: Custom Sensitivity Profiler (CSP) regex logic fails intermittently for complex regex

CSP regex logic fails intermittently for complex regex. Due to this intermittently the Cap tags do not get attached to Hive columns on execution of sensitivity profilers.

None

CDPDSS-2706: Data Steward user role is not reflecting

Data Steward role is not reflecting in Control Plane.

None

CDPDSS-2843: While upgrading to Data Services 1.5.3, an unnamed module @0x4e38d975 exception error is thrown

You must restart your Data Catalog Pod.

CDPDSS-3051: Profiler Jobs are not in the sorted order

Profiler jobs are incorrectly ordered when the list cannot fit into a single page under Profiler > Jobs. The latest job is displayed in the last page.

CDPDSS-3047: The profiled data is not shown for Ranger Audit Profiler

The profiled data is not shown for Ranger Audit Profiler if the active NameNode is switched over to standby.

Switch the NameNode in standby to active state. For more information on restarting nodes, see Starting, Stopping, and Restarting Role Instances or Administrative commands.

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-3077:

Datasets containing deleted assets fail to render the assets list with the following error: pages.dataset.labels.err-undefined-msg(Error Code - undefined)

None

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-2830: Data Catalog running on JDK-17 in PvC causes ClassNotFoundException and further failures

Cloudera Data Catalog running on JDK-17 in PvC causes ClassNotFoundException and further failures which leads to the service not coming up with a Bad Gateways error and being unusable. The following is logged in the Data Catalog pod:

----
            INFO from c.g.i.Guice in [main] - - An exception was caught and reported. Message:
          java.lang.reflect.InaccessibleObjectException: Unable to make protected final
          java.lang.Class
          java.lang.ClassLoader.defineClass(java.lang.String,byte[],int,int,java.security.ProtectionDomain)
          throws java.lang.ClassFormatError accessible: module java.base does not "opens java.lang"
          to unnamed module @7a799159 - java.lang.IllegalStateException: Unable to load cache item
          ----

Use the environment variable in the deployment in the container specification:

----
            name: JDK_JAVA_OPTIONS
            value: --add-opens java.base/java.lang=ALL-UNNAMED --illegal-access=warn
            ----

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-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-3050: Username is not visible in Cloudera Data Catalog for the Local Administrator user

The username for local administrator users can be hidden instead of showing in the bottom left corner of the navigation sidebar.

CDPDSS-3049: Expanded graph is displayed on every page of the Schema tab

In the Asset Details page under the Schema tab, the bar graph of profiled assets might stay visible for a selected asset even when moving to the next page in the list.

CDPDSS-3048: Filtering based on Job status Finished is not working

The profiler job filter Finished is working correctly now, displaying the relevant jobs.

CDPDSS-3042: Edit classification is failing with guid undefined error

Trying to edit the classifications in the Asset Details screen can result in the "List properties for entity request failed Error from Atlas. Not Found : Received 404. Body {\"errorCode\":\"ATLAS-404-00-005\",\"errorMessage\":\"Given instance guid undefined is invalid/not found\"}" message.

CDPDSS-3037: Asset Details page does not highlight link for Datasets

The Asset Details now correctly highlights links to Datasets.

CDPDSS-3524: Profiler pod launch is failing with init error

During the profiler cluster setup, when all the services pods are being deployed, Cloudera Data Catalog attempts to complete the prerequisites to set up a service and request to create a machine user. If the UMS service has not come up properly at that time, the request fails, and does not create the machine user dp_profiler_user required for profiling.

1. Go to Management Console > > User Management > Users.
2. Search the machine user dp_profiler_user. Continue with the following steps, if the machine user is missing.
3. Use the following command to restart Cloudera Data Catalog: kubectl rollout restart deployment/cdp-release-dssapp -n <***NAMESPACE***>.
4. Delete and relaunch the profilers.

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.

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-3381: The Ranger Audit Profiler is showing status as FINISHED/SUCCESS even though there is error with the profilers

The Ranger Audit Profiler can shows an incorrect status when the NameNode is stopped, and it is can get stuck in FINISHED.

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.