Known issues and limitations for IBM Knowledge Catalog

Assets imported with the user admin instead of cpadmin

For Cloud Pak for Data clusters with Identity Management Service enabled, the default administrator is cpadmin. However, for import, the default administrative user admin is used. Therefore, the assets are imported with the admin user instead of cpadmin.

Applies to: 5.3.0

Workaround:

Before running the import, apply the following workaround:

1. Edit the config map by executing oc edit cm catalog-api-exim-cm

2. Manually update the environment variable admin_username in import-job.spec.template.spec.env from:

   - name: admin_username
   value: ${admin_username}
   

   to:

   - name: admin_username
   value: cpadmin
   

Search bar returning incorrect results

• Searching for assets when using the search bar returns unexpected results if only one or two characters are used.

  Applies to: 5.3.0
  

  Workaround: Type at least three characters in the search bar.

Assessment link in email notifications doesn't work

Applies to: 5.3.0

In email notifications for data quality SLA assessment, the link to the assessment does not work due to an extra dot in the URL.

Workaround: To access the assessment, copy the link to a new browser window, remove the extra dot between ibm and com, and press Enter.

When upgrading with the Identity Managements service enabled, the upgrade may fail

Applies to: 5.3.0

If you have the Identity Managements service enabled and you have IBM Knowledge Catalog Premium or IBM Knowledge Catalog Standard installed, or a service that is dependent on either of these two versions of IBM Knowledge Catalog, upgrading may fail.

Workaround Run the follwing steps to fix this issue:

1. Backup your data from the PersistentVolumeClaims (PVC). This is not mandatory but it is recommended in the event of any issues that may occur while fixing the upgrade issue. Run the following to backup your data:

   oc get po | grep semantic-automation
   oc cp semantic-automation-xx-yy:/exchange/configuration ./configuration/
   

2. Remove the PVC:

   oc delete pvc volumes-ikcfilestorage-pvc
   

3. Restart the pods:

   semantic-automation
   semantic-embedding
   semantic-text-generation
   pods
   oc get po| grep semant
   oc delete po <x y z>
   

4. Debug any failing job pods:

   oc get po| grep ikc-lib-volume
   oc debug ikc-lib-volume-instance-xxxxx
   

5. Recreate the PVC volume:

   /bin/sh /wkc/genkeys.sh
   exit
   

6. Check that the semantic pods are running again:

   oc get po| grep semant
   

During uninstalling IBM Knowledge Catalog, the glossary cr dependency is not deleted

Applies to: 5.3.0

When uninstalling IBM Knowledge Catalog, the glosssary cr dependency is not deleted. This means that if you try to re-install IBM Knowledge Catalog, it will not properly install. With the IKC operator gone during the initial uninstall, there is no way to clean up the glossary cr that remains.

Workaround Run the following command in the cluster to patch the glossary cr:

oc patch Glossary glossary-cr -p '{"metadata":{"finalizers": null}}' --type=merge

IBM Knowledge Catalog 5.3.0 installation might not complete

Applies to: 5.3.0

The installation of IBM Knowledge Catalog 5.3.0 might not complete because the Spark cluster instance could not be provisioned.

Workaround: Force reconciliation of the wkc custom resource:

1. Delete the wdp-profiling-iae-init job:

   oc delete job wdp-profiling-iae-init -n ${PROJECT_CPD_INST_OPERANDS}
   

2. Make sure that the job and the corresponding pod are deleted:

   oc get job -n ${PROJECT_CPD_INST_OPERANDS} | grep wdp-profiling-iae-init
   

   oc get pod -n ${PROJECT_CPD_INST_OPERANDS} | grep wdp-profiling-iae-init
   

   These commands should not return any output.

3. Initiate reconciliation of the wkc operator by deleting the wkc-operator pod:

   oc -n ${PROJECT_CPD_INST_OPERATORS} delete pod `oc get pods -n ${PROJECT_CPD_INST_OPERATORS} | grep ibm-cpd-wkc-operator | cut -d' ' -f1`
   

   Check whether the wkc-cr status changed to InProgress and monitor the newly created wkc-operator pod log.

IBM Knowledge Catalog portal-catalog pod out of sync after upgrade

Applies to: 5.3.0

When upgrading IBM Knowledge Catalog and changing editions, the portal-catalog pod may become out of sync, leading to missing functionality that should be enabled from the upgrade.

Workaround: To enable the missing functionality, restart the portal-catalog pod after upgrading IBM Knowledge Catalog.

After installing IBM Knowledge Catalog, some pods may go into the error state and related jobs will fail

Applies to: 5.3.0

After installing IBM Knowledge Catalog, during the post-install steps when the apply-cr commands are running, pods related to kg-resync-glossary and jobs related to these pods might fail.

Workaround: To fix this issue, run the following steps:

1. Check for pods that are in the failed status:

   oc get pod -n ${PROJECT_CPD_INST_OPERANDS} | grep kg-resync-glossary-
   

2. Check the corresponding job status for those pods:

   oc get job kg-resync-glossary -n ${PROJECT_CPD_INST_OPERANDS}
   

3. Delete the kg-resync-glossary job:

   oc delete job kg-resync-glossary -n ${PROJECT_CPD_INST_OPERANDS}
   

4. Reconcile the custom resource (CR) by restarting the wkc-operator pod:

   oc delete pod ibm-cpd-wkc-operator-xxxx-xxxx -n ${PROJECT_CPD_INST_OPERATORS}
   

5. Wait for the CR reconciliation to complete and check the pods. Then, the kg-resync-glossary-xxxx pod should be completed.

In IBM Knowledge Catalog Standard 5.3, semantic automation pods might be missing some labels

Applies to: 5.3.0

If custom models are enabled in IBM Knowledge Catalog Standard 5.3, some semantic automation pods might be missing the icpdsupport/app: and icpdsupport/module labels. As a result, these pods can't be tracked on the IBM Software Hub Monitoring page or might be missed by serviceability scripts that identify pods based on the labels.

Unauthorized users might have access to profiling results

Applies to: 5.3.0

Users who are collaborators with any role in a project or a catalog can view an asset profile even if they don't have access to that asset at the data source level or in Data Virtualization.

Workaround: Before you add users as collaborators to a project or a catalog, make sure they are authorized to access the assets in the container and thus to view the asset profiles.

Cannot run import operations on a container package exported from another Cloud Pak for Data cluster

Applies to: 5.3.0

When you're importing a container package exported from another Cloud Pak for Data cluster, permissions on the archive must be configured so that export operations are available on the target cluster and the files within the archive can be accessed.

Workaround: To extract the export archive and modify permissions, complete the following steps:

1. Create a temporary directory:

   mkdir temp_directory
   

2. Extract the archive:

   tar -xvf cpd-exports-<export_name>-<timestamp>-data.tar --directory temp_directory
   

3. Clients must run the following command on the target cluster:

   oc get ns $CLUSTER_CPD_NAMESPACE -o=jsonpath='{@.metadata.annotations.openshift\.io/sa\.scc\.supplemental-groups}'
   

   Example output: 1000700000/10000.

4. Apply the first part of the output of the previous step (ex. 1000700000) as the new ownership on all files within the archive. Example:

   cd temp_directory/
   chown -R 1000700000:1000700000 <export_name>
   

5. Archive the fixed files with the directory. Use the same export name and timestamp as the original exported tar:

   tar -cvf cpd-exports-<export_name>-<timestamp>-data.tar <export_name>/
   

6. Upload the archive.

Data protection rules don't apply to column names that contain spaces

Applies to: 5.3.0

If a column name contains trailing or leading spaces during import, the column cannot be masked using data protection rules.

Workaround: When you're importing columns, ensure that column names don't contain trailing or leading spaces.

Preview of data from file-based connections other than IBM Cloud Object Storage is not fully supported

Applies to: 5.3.0

Connected assets from file-based connections other than IBM Cloud Object Storage do not preview correctly. Data might appear in a table with missing and/or incorrect data. There is no workaround at this time.

Scroll bar is not visible when adding assets to a project on MacOS

When adding assets to a project, the scroll bar might not be available in the Selected assets table, showing a maximum of 5 assets.

Applies to: 5.3.0

Workaround: Change the MacOS settings:

1. Click the Apple symbol in the top-left corner of your Mac's menu bar, then click System Settings.
2. Scroll down and select Appearance.
3. Under the Show scroll bars option, click the radio button next to Always.

Unexpected assets filtering results in catalogs

Applies to: 5.3.0

In catalogs, when you are searching for an asset by using Find assets field, the search might return assets whose names don't match the name string that you typed in the search field and assets that contain a keyword in a property or a related item associated with the typed name string.

Can't use Git projects with identical data assets

Applies to: 5.3.0

Identical data assets don't work with Git projects.

Workaround: To publish assets from catalog to a Git project, check out a branch in the Git project first.

Mismatched column metadata information

Applies to: 5.3.0

If you add columns to an existing asset, you can see the new columns in the Assets tab, but not in the Overview or Profile tabs.

Workaround: Reprofile the asset to view the changes by running a metadata import with the Discover option.

For SQL assets, the SQL validation dropdown might not open automatically when you retest

Applies to: 5.3.0

If you create a query-based assed and you close the SQL validation dropdown after an initial test, the drop-down does not automatically open when you click the Test query button again.

Workaround: Open the dropdown by clicking the chevron next to the SQL validation label.

Inconsistent behaviour when publishing assets after metdata enrichment

Applies to: 5.3.0

If you run a metadata enrichment on a connected data asset in a project and publish the asset back to a governed catalog from the metdata enrichment screen, the status of the asset doesn't change from Draft to Referenced. As a result, any changes to shared properties made on a catalog-level aren't reflected on a project-level for such an asset.

Workaround: Publish the asset from the asset browser instead.

Publishing asset from project to catalog fails despite success message

Applies to: 5.3.0

When publishing an asset from a project to a catalog, the system displays a success notification, but the asset does not appear in the catalog as expected.

Workaround:

Check the network console for any errors.

Data quality score not available for columns in the catalog lineage view

You can't view data quality score for columns in the catalog lineage view for connected assets. The score is available in the catalog where the asset is published, but it doesn't show up in the lineage view of a different catalog that contains the same connected asset.

Applies to: 5.3.0

Error Couldn't fetch reference data values shows up on screen after publishing reference data

Applies to: 5.3.0

When new values are added to a reference data set, and the reference data set is published, the following error is displayed when you try to click on the values:

Couldn't fetch reference data values. WKCBG3064E: The reference_data_value for the reference_data which has parentVersionId: <ID> and code: <code> does not exist in the glossary. WKCBG0001I: Need more help?

When the reference data set is published, the currently displayed view changes to Draft-history as marked by the green label on the top. The Draft-history view does not allow to view the reference data values.

Workaround: To view the values, click Reload artifact so that you can view the published version.

Publishing large reference data sets fails with Db2 transaction log full

Applies to: 5.3.0

Publishing large reference data sets might fail with a Db2 error such as:

The transaction log for the database is full. SQLSTATE=57011

Workaround: Publish the set in smaller chunks, or increase Db2 transaction log size as described in the following steps.

1. Modify the transaction log settings with the following commands:

   db2 update db cfg for bgdb using LOGPRIMARY 5 --> default value, should not be changed
   db2 update db cfg for bgdb using LOGSECOND 251
   db2 update db cfg for bgdb using LOGFILSIZ 20480
   

2. Restart Db2.

You can calculate the required transaction log size as follows:

(LOGPRIMARY + LOGSECOND) * LOGFILSIZ

For publishing large sets, the following Db2 transaction log sizes are recommended:

• 5GB for 1M reference data values and 300K relationships
• 20GB for 1M reference data values and 1M relationships
• 80GB for 1M reference data values and 4M relationships

where the relationship count is the sum of the parent, term and value mapping relationships for reference data values in the set.

Imported data assets with assigned out-of-the-box data classes or terms have incorrect identifiers resulting in no enforcement of data protection rules

When you migrate data assets across Cloud Pak for Data instances and these assets have out-of-the-box data classes or terms assigned, the imported data assets indicate correct data class or term assignments but the assigned artifact ID is incorrect. As a result, any operations that reference the data class or term, such as data protection rules, can't be applied to the imported data assets.

Relationships between catalog assets and out-of-the-box governance artifacts cannot be migrated correctly.

Applies to: All versions of Cloud Pak for Data beginning with 4.0 and later.

Workaround: none

Can't cascade delete a Knowledge Accelerator category

Applies to: 5.3.0

If you run cascade delete of a Knowledge Accelerator category, the operation might fail due to deadlocks.

Workaround: In case of deadlocks, retry cascade delete of the same root category until it's deleted.

wkc-glossary-service pod restarting when updating or creating multiple business terms

When updating or creating large numbers of business terms, the wkc-glossary-service pod is restarting due to reaching the CPU limit.

Applies to: 5.3.0

Workaround: Increase the CPU limit for wkc-glossary-service as described in Manually scaling resources for services.

Column data is not redacted in accordance with the rule precedence

Applies to: 5.3.0

Some advanced partial masking options get ignored by the rule precedence mechanism when the rules resolution is executed, which results in incorrect data redaction.

Workaround: If there are multiple rules with partial masking specified, ensure that all of the partial masking columns are the same in all of these rules for the same data class.

Publishing asset from project to catalog fails despite success message

Applies to: 5.3.0

When publishing an asset from a project to a catalog, the system displays a success notification, but the asset does not appear in the catalog as expected.

Workaround:

Check the network console for any errors.

Workflow types not available

When custom request workflow are used and a user group is added to the list of start users (in Workflow Type > Overview > Access), it is not possible to retrieve the workflow types anymore. This leads to missing workflow types in workflow management UI and task inbox. The Start new request button is not available either.

Applies to: 5.3.0

Workaround: To solve this issue manually you can access the workflow type that contains the user group using the following URL:

https://<HOST>/gov/workflow/types/<WORKFLOW_TYPE_ID>

Go to Overview > Access and remove the user group from the list of users that can start a request.

If you don't know the workflow type ID, follow these steps to get all IDs:

1. Open a terminal and run the following command to get an authorization bearer token:

   curl -k -X POST https://<HOST>/icp4d-api/v1/authorize     -H 'cache-control: no-cache'     -H 'content-type: application/json'     -d '{"username":"<USERNAME>","password":"<PASSWORD>"}'
   

2. Run the following command to retrieve all workflow types:

   curl -X GET "https://<HOST>/v3/workflow_types" --insecure -H "accept: application/json" -H "Authorization: Bearer <TOKEN>"
   

Connection of wkc-workflow-service to its Postgres database does not use SSL/TLS encryption

Applies to: 5.3.0
Fixed in: IBM Knowledge Catalog 5.3.0 and IBM watsonx.data intelligence 2.3.0 patch

Starting in version 5.2.0, the workflow service is using a Postgres Database as its internal database, which resides on the same Openshift cluster. When connecting to the database, the wkc-workflow-service is not using SSL/TLS encryption. Therefore, this intra-cluster connection is unencrypted.

Assets are not imported from the IBM Cognos Analytics source when the content language is set to Japanese

Applies to: 5.3.0

If you want to import metadata from the Cognos Analytics connection, where the user's content language is set to Japanese, no assets are imported. The issue occurs when you create a metadata import with the Get BI report lineage goal.

Workaround: In Cognos Analytics, change the user's content language from Japanese to English. Find the user for which you want to change the language, and change this setting in the Personal tab. Run the metadata import again.

Dummy assets get created for any file assets that come from Amazon S3 to show the complete business data lineage if Get ETL job lineage is performed

Applies to: 5.3.0

If you perform Get ETL job lineage import involving Amazon S3 connection, dummy assets get created for any file assets that come from Amazon S3 connection to show the complete business data lineage. If you perform metadata import for the same Amazon S3 connection, a duplicate asset will get created for the dummy asset created from Get ETL job lineage import and a valid asset discovered during the metadata import.

SocketTimeoutException during metadata import

Applies to: 5.3.0

During metadata import, when records from a CSV file that contains more than 30,000 rows are read, SocketTimeoutException is returned. This indicates a network issue where the connection between the client and server was unexpectedly closed.

Workaround:

1. Log in to the OpenShift console.

2. Go to Workloads > Pods > metadata-discovery-pod.

3. Go to the Environment section.

4. Search for the manta_wf_export_download environment variable and set it to true.

   Example:

   manta_wf_export_download=true
   

   By setting the variable, you're bypassing the socket timeout issue and downloading the CSV file to the local system. As a result, the CSV file can be read locally rather than over the network. After the CSV file is read, the locally downloaded file is deleted from the local system.

Business lineage shows Name not available when you're using Get BI report lineage option to import Tableau data assets

Applies to: 5.3.0

When you're using Get BI report lineage option to import Tableau data assets, business lineage for some of these assets shows Name not available instead of the actual data asset names due to extra object types that are generated by MANTA Automated Data Lineage service.

Workaround: Use technical data lineage to show the names of all data assets.

Reference assets are not marked as Removed from scope or Removed from source

Applies to: 5.3.0

If you remove an asset that was initially imported into the catalog with the associated primary and foreign keys from the metadata import scope or remove the asset from the data source, the asset itself is marked accordingly as Removed from scope or Removed from source in the catalog on reimport, but the associated PK and FK constraint assets aren't. However, when you delete the data asset from the catalog, the associated constraint assets are also deleted.

Assets removed from the metadata import scope are marked incorrectly

Applies to: 5.3.0
Fixed in: IBM Knowledge Catalog 5.3.0 and IBM watsonx.data intelligence 2.3.0 patch

If you remove an asset from the metadata import scope and rerun the import, the data asset in the catalog has the tag Removed from source instead of Removed from scope.

Running primary key or relations analysis doesn't update the enrichment and review statuses

Applies to: 5.3.0

The enrichment status is set or updated when you run a metadata enrichment with the configured enrichment options (Profile data, Analyze quality, Assign terms). However, the enrichment status is not updated when you run a primary key analysis or a relationship analysis. In addition, the review status does not change from Reviewed to Reanalyzed after review if new keys or relationships were identified.

Issues with the Microsoft Excel add-in

Applies to: 5.3.0

The following issues are known for the Review metadata add-in for Microsoft Excel:

• When you open the drop-down list to assign a business term or a data class, the entry Distinctive name is displayed as the first entry. If you select this entry, it shows up in the column but does not have any effect.

• Updating or overwriting existing data in a spreadsheet is currently not supported. You must use an empty template file whenever you retrieve data.

• If another user works on the metadata enrichment results while you are editing the spreadsheet, the other user's changes can get lost when you upload the changes that you made in the spreadsheet.

• Only assigned data classes and business terms are copied from the spreadsheet columns Assigned / suggested data classes and Assigned / suggested business terms to the corresponding entry columns. If multiple business terms are assigned, each one is copied to a separate column.

Republishing doesn't update primary key information in catalog

Applies to: 5.3.0

If you remove primary key information from a data asset that initially was published with the primary key information to a catalog with the duplicate-asset handling method Overwrite original assets in the metadata enrichment results and then republish the asset to that catalog, the primary key information on the catalog asset remains intact.

Workaround: Delete the existing catalog asset before you republish the data asset from the metadata enrichment results.

Masked data might be profiled when the data source is IBM watsonx.data

Applies to: 5.3.0

If a user who is not the owner of a protected data asset in IBM watsonx.data adds such asset to a project and runs metadata enrichment on it, the masked data is sent for profiling. As a result, even the asset owner will see the profile with masked data.

Workaround: None.

In small system configurations, large metadata enrichment jobs can fail

Applies to: 5.3.0

In a system that is configured for smaller workloads, metadata enrichments that contain a lot of assets can fail with out-of-memory errors.

Workaround: To resolve the issue, you have these options:

• Increase the CPU and memory values of the profiling pod.
• Add one more replica.

Update these parameters in the wkc-cr custom resource:

• For the number of replicas: the wdp_profiling_min_replicas and wdp_profiling_max_replicas values
• For the CPU and memory values: the requests and limits entries for the wdp_profiling_resources parameter

Use the oc patch command to update the values. See the following example:

oc patch wkc wkc-cr -n ${PROJECT_CPD_INSTANCE} --type=merge -p '{"spec":{"wdp_profiling_min_replicas":"4","wdp_profiling_max_replicas":"4","wdp_profiling_resources":{"requests":{"cpu": "300m", "memory": "600Mi"}, "limits":{"cpu": "4000m", "memory": "8192Mi"}}}}'

Data quality output configuration in metadata enrichment can't be removed

Applies to: 5.3.0

After you saved a metadata enrichment that you configured with the Run data quality analysis objective and a custom data quality output table, you can no longer remove the output table configuration from the metadata enrichment. When you edit the metadata enrichment, you can click Customize in the Run data quality analysis tile and then switch off the output table setting, but that change doesn't persist.

Workaround: To remove the configured data quality output database from the metadata enrichment, complete these steps:

1. Generate a token as described in Generating a bearer token.

2. Set the HOST variable:

   export HOST=<host_name>
   

   where <host_name> is the hostname of your deployment.

3. Set the PROJECT_ID variable:

   export PROJECT_ID=<project_id>
   

   where <project_id> is the ID of the project that holds the metadata enrichment. You can find this ID in the browser URL between the project_id= and &context strings.

4. Set the MDE_ID variable:

   export MDE_ID=<mde_id>
   

   where <mde_id> is the ID of the metadata enrichment. You can find this ID in the browser URL between the /display/ and ?project_id= strings.

5. Set the TOKEN variable:

   export TOKEN=<access_token_value>
   

   where <access_token_value> is the token that you created in the first step.

6. Run the following cURL command:

   curl -X PATCH -H "Content-Type: application/merge-patch+json" -H "Authorization: Bearer $TOKEN" https://$HOST/metadata_enrichment/v3/metadata_enrichment_assets/$MDE_ID?project_id=$PROJECT_ID" -d '{"objective":{"data_quality":{"structured":{"dq_exceptions_database":null,"dq_exception_records_database":null}}}}'
   

   You might need to run the command with the -k option if the certificate verification fails.

Rules bound to columns of the data type NUMERIC in data assets from Oracle data sources might not work

Applies to: 5.3.0

Testing or running a data quality rule that is bound to a NUMERIC column in a data asset from an Oracle data source fails if the data source is connected through a Generic JDBC connection.

Workaround: Use the native connector.

Runs of migrated data quality rules complete with warnings

Applies to: 5.3.0

When you run a data quality rule that was migrated from from InfoSphere Information Server, you might see a message like the following one:

7/2/2025 11:22:30 WARNING IIS-DSEE-TFIP-00072 <Modify_1> When checking operator: When binding output schema variable "outRec": When binding output interface field "col1" to field "col1": Implicit conversion from source type "int32" to result type "string[variable_max=10]": Converting number to string.

Such warnings are displayed when you run the DataStage job for a data quality rule that was created in InfoSphere Information Server as a Rule Stage with an output link of the type Violation details and then migrated to Cloud Pak for Data.

Workaround: You can ignore such warnings or set up a DataStage message handler to suppress such messages or to redurce their severity.

Data quality rules with duplicated variable names from multiple data quality definitions might not work

Applies to: 5.3.0

When you create a data quality rule from multiple data quality definition that reference the same variable name, the variables are internally renamed for differentiation by adding a suffix. If the renaming results in a name collision with other variables from the definitions, the data quality rule will not work.

Can't drill-down into data quality output on DataStax Enterprise

Applies to: 5.3.0

When you try to access details of data quality issues that are stored in a table on DataStax Enterprise, an error similar to the following one is displayed:

 Loading failed values failed.
 There was a problem loading the values that
 failed profiling.
 Internal Server Error Failed to get Fight info: CDICO2005E: Table could not be found:
 SQL syntax error: [IBM][Cassandra JDBC Driver)[Cassandra syntax error or access rule
 violation: base table or view not found: dgexceptions. If the table exists, then the user
 may not be authorized to see it.

This error occurs because the name of the data quality output table is created in a way that makes it case-sensitive in the underlying Cassandra database so that an uppercase name is generated. However, the SELECT statement for the lookup is constructed in a way that Cassandra search looks for a lowercase table name.

In deployments with IBM Knowledge Catalog Premium, access to SLA features might be blocked

Applies to: 5.3.0

In a deployment with IBM Knowledge Catalog Premium, users might not have access to any data quality SLA features and erroneously see the error message The data quality SLA feature is not enabled via license..

Users cannot create or update data quality SLA rules, generate SLA assessments, or access information about SLA assessments on the data quality page for an asset.

Workaround: To fix the issue, a cluster or instance administrator should restart the wkc-data-rules service.

Data quality scores might not be available for data quality rules created in earlier versions

Applies to: 5.3.0

In some cases, no data quality scores are available for a rule that was created in an earlier product version. This issue is caused by missing asset relationships. Such relationships are usually set up when rules from an earlier release are migrated during the upgrade to the new product version.

Workaround: Complete the following steps:

1. Open the affected data quality rule.
2. In the Related items, select Add related items > Add related columns.
3. Select the Validates data quality of relationship and click Next.
4. Select a data asset and then the columns for which you want the rule to report a score report and click Add.

Can't create rules with plain-text queries for data quality output tables

Applies to: 5.3.0

For output tables that result from running a data quality rule, you cannot create a new SQL-based data quality rule by using plain text queries.

Socket timeout errors can occur for Text2SQL API calls

Applies to: 5.3.0

Socket timeout errors can occur for Text2SQL API calls if the load balancer timeout setting for watsonx.ai is too low.

Workaround: Check the load balancer configuration and make sure that the timeout is set to 600s (10m). For more information, see Changing load balancer timeout settings.

Can't work with a remote watsonx.ai instance if the system is configured to run models on CPU

Applies to: 5.3.0

If you installed IBM Knowledge Catalog or watsonx.data intelligence with the option enableModelsOn:cpu, the models used for the generative AI capabilities run locally on CPU. You cannot work with foundation models on a remote watsonx.ai instance.

For more information about setting up a connection with a remote watsonx.ai instance, see Configuring the setup for enabled gen AI capabilities in the Cloud Pak for Data documentation or Configuring the setup for enabled gen AI capabilities in the watsonx.data intelligence documentation.

Workaround: To be able to work with remote models, change your system configuration:

1. Log in to Red Hat OpenShift Container Platform as a user with sufficient permissions to complete the task.

   oc login <your_openshift_cluster_url>
   

2. Set the context to the project where IBM Knowledge Catalog or watsonx.data intelligence is deployed:

   oc project ${PROJECT_CPD_INST_OPERANDS}
   

3. Patch the semanticautomation-cr custom resource to set the installation parameter enableModelsOn to remote.

    oc patch sal semanticautomation-cr --type=merge -p '{"spec":{"enableModelsOn": "remote"}}'
   

Project collaborators with the Viewer role can use the toggle to enable or disable the Data intelligence settings

Applies to: 5.3.0

Project collaborators with the Viewer role can switch the toggle setting for enabling or disabling the Data intelligence settings even though they have only read access to the settings. As a result, changing the toggle setting returns an error similar to the following one:

Disabling failed
Disabling generative AI capabilities failed. Try again.

The project settings page for data intelligence shows the main IBM Cloud Pak for Data toolbar

Applies to: 5.3.0

In the IBM Cloud Pak for Data and Data Fabric experiences, the project settings page for data intelligence shows an additional IBM Cloud Pak for Data toolbar. This additional toolbar does not impact the functionality of the settings UI. You can still change the settings as required.

Known issues with the Text-to-SQL capability

Applies to: 5.3.0
Fixed in: IBM Knowledge Catalog 5.3.0 and IBM watsonx.data intelligence 2.3.0 patch

You might encounter the following known issues for the Text-to-SQL feature:

• Even if the Text-to-SQL capability is not enabled for IBM Knowledge Catalog or watsonx.data intelligence in IBM Software Hub, the UI for SQL assets and SQL-based data quality rules shows a notification asking to enable natural language queries.

• If a project contains similar data assets from various connections, incorrect SQL statements are created when you use plain text queries to create SQL assets or SQL-based data quality rules.

• Incorrect SQL statements might be created when you use plain text queries to create SQL assets from PostgreSQL connection.

• SQL assets (assets of the type query) might be processed when you use plain text queries to create new SQL assets or SQL-based data quality rules.

• Even if the Text-to-SQL capability is enabled for IBM Knowledge Catalog or watsonx.data intelligence in IBM Software Hub, the create wizards for SQL assets and SQL-based data quality rules do not show the Tech Preview AI tag.

• If IBM Knowledge Catalog or watsonx.data intelligence are installed with the installation option enableModels=CPU, the Text-to-SQL capability should not be enabled.

Metadata import jobs for getting lineage might take very long to complete

Applies to: 5.3.0

If multiple lineage scans are requested at the same time, the corresponding metadata import jobs for getting lineage might take very long to complete. This is due to the fact that MANTA Automated Data Lineage workflows can't run in parallel but are executed sequentially.

Chrome security warning for Cloud Pak for Data deployments where MANTA Automated Data Lineage for IBM Cloud Pak for Data is enabled

Applies to: 5.3.0

When you try to access a Cloud Pak for Data cluster that has MANTA Automated Data Lineage for IBM Cloud Pak for Data enabled from the Chrome web browser, the message Your connection is not private is displayed and you can't proceed. This is due to MANTA Automated Data Lineage for IBM Cloud Pak for Data requiring an SSL certificate to be applied and occurs only if a self-signed certificate is used.

Workaround: To bypass the warning for the remainder of the browser session, type thisisunsafe anywhere on the window. Note that this code changes every now and then. The mentioned code is valid as of the date of general availability of Cloud Pak for Data 4.6.0. You can search the web for the updated code if necessary.

Columns are displayed as numbers for a DataStage job lineage in the catalog

Applies to: 5.3.0

The columns for a lineage that was imported from a DataStage job are not displayed correctly in the catalog. Instead of column names, column numbers are displayed. The issue occurs when the source or target of a lineage is a CSV file.

MANTA Automated Data Lineage will not function properly on IBM Knowledge Catalog Standard

Applies to: 5.3.0

If you install MANTA Automated Data Lineage when you have IBM Knowledge Catalog Standard installed as the prerequisite, MANTA will not function properly.

If you want to install MANTA, you will need to have IBM Knowledge Catalog Premium installed.

Can't Get lineage with a DB2 connection in FIPS environments

Applies to: 5.3.0

If you try to import metadata for data assets with the Get lineage scenario with a DB2 connection in a FIPS environment, the metadata import fails and the following error message is displayed.

Error creating the metadata import Metadata import creation failed due to connection validation errors. Not all connections in the metadata import passed validation. Check the log for the complete validation results.

Data export is incomplete due to table deduction in Query Service

Applies to: 5.3.0

When you reference a table from another system in the Query Service using the synonym of that table, then the resulting object is deduced and exported data is incomplete.

Business data lineage is incomplete for the metadata imports with Get ETL job lineage or Get BI report lineage goals

Applies to: 5.3.0

In some cases, when you display business lineage between databases and ETL jobs or BI reports, some assets are missing, for example, a starting database. The data was imported by using the Get ETL job lineage or Get BI report lineage import option. Technical data lineage correctly shows all assets.

Workaround: Sometimes MANTA Automated Data Lineage cannot map the connection information from an ETL job or a BI report to the existing connections in IBM Knowledge Catalog. Follow these steps to solve the issue:

1. Open the MANTA Automated Data Lineage Admin UI:

   https://<CPD-HOSTNAME>/manta-admin-gui/
   

2. Go to Log Viewer and from the Source filter select Workflow Execution.

3. From the Workflow Execution filter, select the name of the lineage workflow that is associated with the incomplete business lineage.

4. Look for the dictionary_manta_mapping_errors issue category and expand it.

5. In each entry, expand the error and click View Log Details.

6. In each error details, look for the value of connectionString. For example, in the following error message, the value of the connectionString parameter is DQ DB2 PX.

   2023/11/14 18:40:12.186 PM [CLI] WARN - <provider-name> [Context: [DS Job 2_PARAMETER_SET] flow in project [ede1ab09-4cc9-4a3f-87fa-8ba1ea2dc0d8_lineage]]
   DICTIONARY_MANTA_MAPPING_ERRORS - NO_MAPPING_FOR_CONNECTION
   User message: Connection in use could not be automatically mapped to one of the database connections configured in MANTA.
   Technical message: There is no mapping for the connection Connection [type=DB2, connectionString=DQ DB2 PX, serverName=dataquack.ddns.net, databaseName=cpd, schemaName=null, userName=db2inst1].
   Solution: Identify the particular database technology DB2 leading to "DQ DB2 PX" and configure it as a new connection or configure the manual mapping for that database technology in MANTA Admin UI.
   Lineage impact: SINGLE_INPUT
   

7. Depending on the connection that you used for the metadata import, go to Configuration > CLI > connection server > connection server Alias Mapping, for example DB2 > DB2 Alias Mapping.

8. Select the connection used in workflow and click Full override.

9. In the Connection ID field, add the value of the connectionString parameter that you found in the error details, for example DQ DB2 PX.

10. Rerun the metadata import job in IBM Knowledge Catalog.

Hops for components and columns of components work only inside the data integration flow area of an expanded job node

Applies to: 5.3.0

When working with the lineage graph, hops for components and columns of data integration components work only inside the data integration flow area of an expanded job node and don't connect columns of nodes outside of the flow area.

Viewing large business lineage graph results in a service error

Applies to: 5.3.0

When you click the Lineage tab for a large business lineage graph, after around two minutes the following error message is displayed:

Lineage service error. An error occurred in the lineage service. Try to reload the lineage graph, and if the error persists, contact your administrator.

When you click the Lineage tab again, the lineage generation process starts, but after a while the following error is displayed:

Error 404 – Not Found

Workaround: To fix the issue, modify some of the configuration settings:

1. Increase the timeout value in nginx pod:
   1. Open the configuration YAML file by running the command oc edit zenextension wkc-base-routes -n ${PROJECT_CPD_INSTANCE}.
   2. In the section location /data/catalogs, change the value of the proxy_read_timeout variable to 10m, and save the changes.
   3. Restart the nginx pod by running the command oc rollout restart deploy/ibm-nginx -n ${PROJECT_CPD_INST_OPERANDS}
2. Allocate more memory to the service wkc-data-lineage-service, 8 GB or more, and enable horizontal scaling for the service:
   1. Update the lineage deployment with the environment variable oc edit deploy –n ${PROJECT_CPD_INSTANCE} wkc-data-lineage-service.
   2. In the containers section, change the default graph memory threshold by adding the variable lineage_memory_threshold_limit:

      - name: lineage_memory_threshold_limit
      value: "0.5"
      

   3. In the resources/limits section, increase the default memory allocation to 8192Mi:

      resources:
      limits:
         cpu: "2"
         ephemeral-storage: 1Gi
         memory: 8192Mi
      requests:
         cpu: 250m
         ephemeral-storage: 50Mi
         memory: 512Mi
      

   4. Change these settings to horizontally scale the lineage service to 2–3 instances:

      spec:
      progressDeadlineSeconds: 600
      replicas: 3
      HistoryLimit: 10
      

After you modify the configuration, try to open the Lineage tab again. If the error still occurs, change the value of the lineage_memory_threshold_limit parameter to 0.4.

404 Error occurs when opening relationship explorer

Applies to: 5.3.0

Clicking the Explore relationships button from a catalog or project asset results in a 404 error.

Workaround: Check if your Cloud Pak for Data experience includes the IBM Knowledge Catalog any edition service. The relationship explorer is an IBM Knowledge Catalog any edition feature and can’t be accessed without it.

Primary and foreign keys are not visible for Google BigQuery assets

Applies to: 5.3.0

When selecting a Google BigQuery asset on the relationship explorer canvas, the Item metadata panel is not showing primary and foreign keys.

Some relationships are not displayed in relationship explorer after the upgrade

Applies to: 5.3.0

After upgrading from 5.1.x or 5.2.x to 5.3.0, relationships created before the upgrade are not displayed in relationship explorer. Relationships created after the upgrade display correctly.

Workaround: After the upgrade, in the cluster where Knowledge Graph is deployed, delete the sync job. Then, run the reconciliation of the Knowledge Graph custom resource.

When uninstalling Manta Data Lineage, re-installing IBM Knowledge Catalog runs into issues

Applies to: 5.3.0

You can install Manta Data Lineage with IBM Knowledge Catalog. If you uninstall Manta Data Lineage, and then try to re-install the wkc-cr for IBM Knowledge Catalog, you may run into issues. The wkc-post-install-init pod may fail to restart.

Workaround: To fix this issue, restart the ibm-nginx pods, then restart the wkc-operator pod. This will put the wkc-operator in the completed state.

Duplicate actions fail if dynamic IP address are used

Applies to: 5.3.0

Duplicate actions work only for connections with static IP addresses. If the connection is using a hostname with a dynamic IP address, duplicate actions might fail during connection creation.

Long names of the asset owners get truncated when hovering over their avatars

Applies to: 5.3.0

When you are hovering over the avatar to show the long name of the asset owner in the side panel, the name gets truncated if it is longer than 40 characters or contains a space or a special character. If the name is longer than 40 characters, it will display correctly as long as it contains a space or '-' within the first 40 characters.

Can't add individual group members as asset members

Applies to: 5.3.0

You can't add individual group members as asset members. You can add individual group members as catalog collaborators and then as asset members.

Catalog asset search doesn't support special characters

Applies to: 5.3.0

If search keywords contain any of the following special characters, the search filter doesn't return the most accurate results.

Search keywords:

. + - && || ! ( ) { } [ ] ^ " ~ * ? : \

Workaround: To obtain the most accurate results, search only for the keyword after the special character. For example, instead of AUTO_DV1.SF_CUSTOMER, search for SF_CUSTOMER.

Missing default catalog and predefined data classes

Applies to: 5.3.0

The automatic creation of the default catalog after installation of the IBM Knowledge Catalog service can fail. If it does, the predefined data classes are not automatically loaded and published as governance artifacts.

Workaround: Ask someone with the Administrator role to follow the instructions for creating the default catalog manually.

Special or double-byte characters in the data asset name are truncated on download

Applies to: 5.3.0

When you download a data asset with a name that contains special or double-byte characters from a catalog, these characters might be truncated from the name. For example, a data asset named special chars!&@$()テニス.csv will be downloaded as specialchars!().csv.

The following character sets are supported:

• Alphanumeric characters: 0-9, a-z, A-Z
• Special characters: ! - _ . * ' ( )

Catalog UI does not update when changes are made to the asset metadata

Applies to: 5.3.0

If the Catalog UI is open in a browser while an update is made to the asset metadata, the Catalog UI page will not automatically update to reflect this change. Outdated information will continue to be displayed, causing external processes to produce incorrect information.

Workaround: After the asset metadata is updated, refresh the Catalog UI page at the browser level.

A blank page might be rendered when you search for terms while manually assigning terms to a catalog asset

Applies to: 5.3.0

When you search for a term to assign to a catalog asset and change that term while the search is running, it can happen that a blank page is shown instead of any search results.

Workaround: Rerun the search.

Project assets that are added while you create segmented data assets might not be available for selection

Applies to: 5.3.0

If assets are added to the project while you are viewing the list of data assets to pick the column for segmentation, these new assets are listed, but you cannot select them.

Workaround: Cancel the creation process and start anew.

An extra path to manage catalogs in the navigation menu

Applies to: 5.3.0

If you have the Manage catalogs user permission, an extra Administration > Catalogs path to manage catalogs shows up in the navigation menu.

Migrated connections aren't listed during importing from the asset browser

Applies to: 5.3.0

After you migrate connections and you try to import assets from any of the migrated connections, the asset browser is empty. No connections are listed even though the connection is available in the project.

Row filter data protection rules are not available for previews of SQL query asset type

Applies to: 5.3.0

Row filter data protection rules do not work for previews of SQL query asset type. If any of the row filter rule applies to the SQL query asset in a governed catalog, the asset viewer will see an error.

Workaround: Choose one of the following methods:

• As a Data Steward, disable the row filter rule, which is applied on the SQL query asset.
• Edit the SQL query with where predicates are required at the time of creation of the asset in the project and then publish it to the catalog.

Pagination selection dropdown no longer available on the Catalogs page

Applies to: 5.3.0

Pagination selection dropdown that let you navigate to a specfic page is no longer available on the Catalogs page. You must use the left (previous page) or right (next page) arrows to get to the required page.

Can't add empty or zero-byte files to catalogs

Applies to: 5.3.0

You can't add empty or zero-byte files when you're adding local files to a catalog (on the catalog asset page, when you click Add to catalog and select Local files).

Workaround: Use the API to add empty or zero-byte files to catalogs as described in the Data and AI Common Core API documentation.

Migrating data source definitions from the Platform assets catalog will fail

Applies to: 5.3.0

Data source definitions can't be migrated and attempts to migrate data source definitions will cause the migration to fail.

Workaround: There is currently no workaround for this issue.
You can migrate all other content from the Platform assets catalog without issues.

Associated connection isn't added when you add a vector index from catalog to project

Applies to: 5.3.0

When you add a vector index (In memory, Elasticsearch, watsonx.data Milvus) to a project from a catalog, the associated connection isn't automatically added with it.

Cannot use CSV to move data class between Cloud Pak for Data instances

Applies to: 5.3.0

If you try to export data classes with matching method Match to reference data to CSV, and then import it into another Cloud Pak for Data instance, the import fails.

Workaround: For moving governance artifact data from one instance to another, especially data classes of this matching method, use the ZIP format export and import. For more information about the import methods, see Import methods for governance artifacts in the Cloud Pak for Data documentation.

Metadata import jobs might be stuck due to issues related to RabbitMQ

Applies to: 5.3.0

If the metadata-discovery pod starts before the rabbitmq pods are up after a cluster reboot, metadata import jobs can get stuck while attempting to get the job run logs.

Workaround: To fix the issue, complete the following steps:

1. Log in to the OpenShift console by using admin credentials.
2. Go to Workloads > Pods.
3. Search for rabbitmq.
4. Delete the rabbitmq-0, rabbitmq-1, and rabbitmq-2 pods. Wait for the pods to be back up and running.
5. Search for discovery.
6. Delete the metadata-discovery pod. Wait for the pod to be back up and running.
7. Rerun the metadata import job.

Data assets might not be imported when running an ETL job lineage import for DataStage flows

Applies to: 5.3.0

When you create and run a metadata import with the goal Get ETL job lineage where the scope is determined by the Select all DataStage flows and their dependencies in the project option, data assets from the connections associated with the DataStage flows are not imported.

Workaround: Explicitly select all DataStage flows and connections when you set the scope instead of using the Select all DataStage flows and their dependencies in the project option.

When a job for importing lineage metadata hangs, it cannot be stopped

Applies to: 5.3.0

When you run a lineage metadata import and the job stops responding, the job can't be stopped.

Only files with .sql extension can be provided as manual input for metadata import from the Oracle and PostgreSQL sources

Applies to: 5.3.0

When you import metadata from the Oracle and PostgreSQL sources, only .sql files can be used as manual input. Other formats like files with .pck extension can't be used. This limitation is applicable when you install IBM Manta Data Lineage.

Assets and lineage might not be imported if many connections use the same data source

Applies to: 5.3.0

If more than one connection points to the same data source, for example to the same Db2 database, importing lineage might not be successful. Assets and lineage metadata might not be imported in such case. When you create a connection to use with metadata import, make sure that only one connection points to a selected data source.

When you import a project from a .zip file, the metadata import asset is not imported

Applies to: 5.3.0

When you import a project from a file, metadata import assets might not be imported. The issue occurs when a metadata import asset was imported to a catalog, not to a project, in the source system from which the project was exported. This catalog does not exist on the target system and the metadata import asset can't be accessed.

Workaround: After you import the project from a file, duplicate metadata import assets and add them to a catalog that exists on the target system. For details, see Duplicating a metadata import asset.

Lineage metadata cannot be imported from the Informatica PowerCenter connection

Applies to: 5.3.0

When you import lineage metadata from the Informatica PowerCenter connection, the metadata job run fails with the following message:

400 [Failed to create discovery asset. path=/GLOBAL_DESEN/DM_PES_PESSOA/WKF_BCB_PES_PESSOA_JURIDICA_DIARIA_2020/s_M_PEJ_TOTAL_03_CARREGA_ST3_2020/SQ_FF_ACFJ671_CNAE_SECUND�RIA details=ASTSV3030E: The field 'name' should contain valid unicode characters.]",
"more_info" : null

Workaround: Ensure that the encoding value is the same in the workflow file in Informatica PowerCenter and in the connection that was created in Automatic Data Lineage. If the values are different, use the one from the Informatica PowerCenter workflow file.
To solve the issue, complete these steps:

1. Open Automatic Data Lineage:

   https://<CPD-HOSTNAME>/manta-admin-gui/
   

2. Go to Connections > Data Integration Tools > IFPC and select the connection for which the metadata import failed.

3. In the Inputs section, change the value of the Workflow encoding parameter to match the value from the Informatica PowerCenter workflow file.

4. Save the connection.

5. In IBM Knowledge Catalog, reimport assets for the metadata import that failed.

Profiling in catalogs, projects, and metadata enrichment might fail for Teradata connections

Applies to: 5.3.0

If a Generic JDBC connection for Teradata exists with a driver version before 17.20.00.15, profiling in catalogs and projects, and metadata enrichment of data assets from a Teradata connection fails with an error message similar to the following one:

2023-02-15T22:51:02.744Z - cfc74cfa-db47-48e1-89f5-e64865a88304 [P] ("CUSTOMERS") - com.ibm.connect.api.SCAPIException: CDICO0100E: Connection failed: SQL error: [Teradata JDBC Driver] [TeraJDBC 16.20.00.06] [Error 1536] [SQLState HY000] Invalid connection parameter name SSLMODE (error code: DATA_IO_ERROR)

Workaround: For this workaround, users must be enabled to upload or remove JDBC drivers. For more information, see Enable users to upload, delete, or view JDBC drivers.

Complete these steps:

1. Go to Data > Connectivity > JDBC drivers and delete the existing JAR file for Teradata (terajdbc4.jar).
2. Edit the Generic JDBC connection, remove the selected JAR files, and add SSLMODE=ALLOW to the JDBC URL.

For assets from SAP OData sources, the metadata enrichment results do not show the table type

Applies to: 5.3.0

In general, metadata enrichment results show for each enriched data asset whether the asset is a table or a view. This information cannot be retrieved for data assets from SAP OData data sources and is thus not shown in the enrichment results.

Rules run on columns of type timestamp with timezone fail

Applies to: 5.3.0

The data type timestamp with timezone is not supported. You can't apply data quality rules to columns with that data type.

MANTA Automated Data Lineage for IBM Cloud Pak for Data

Not all stages are displayed in technical data lineage graph for the imported DataStage ETL flow

Applies to: 5.3.0

When you import a DataStage ETL flow and view it in the technical data lineage graph, only three stages are displayed, even when four stages were imported.

Workaround: By default, three connected elements are displayed in the graph. To display more elements, click the expand icon on the last or the first displayed element on the graph.

Lineage is not reimported after the connection is reimported in Manta Admin UI

Applies to: 5.3.0

After you delete and restore connections in Manta Admin UI, reimporting the lineage metadata for these connections in Cloud Pak for Data fails. The issue occurs when the following steps are taken:

1. You create projects, connections to data sources and catalogs in Cloud Pak for Data.
2. You run metadata imports with the goal Get lineage. The imports are successful and the connections are listed in Manta Admin UI.
3. You export all connections from Manta Admin UI by using the Export all connections option.
4. You remove all connections from Manta Admin UI.
5. You import the same connections in Manta Admin UI again by using the Import connection option.
6. You rerun the metadata imports with the goal Get lineage again for the same connections but the imports are not successful and errors are displayed.

Workaround: The issue is caused by the presence of the truststore properties in the exported .json connection files. For example, for the Db2 data source, the properties are db2.extractor.truststore.path and db2.extractor.truststore.password. You can apply one of the following workarounds:

• Before you import the connection back in Manta Admin UI, manually remove these properties from the connection .json file.
• Export connections one by one, not by using the Export all connections option. When you export the connections individually, the truststore properties are not saved in the .json file.

An unnecessary edge appears when expanding data integration assets

Applies to: 5.3.0

After expanding a data integration asset and clicking Show next or Show all, the transformer nodes will have an unnecessary edge that points to themselves.