> For the complete documentation index, see [llms.txt](https://docs.collibra.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.collibra.com/data-quality-self-hosted/release-notes/data-quality-observability-self-hosted-release-archive/release-2026-03.md). # Release 2026.03 * Release date of Data Quality & Observability (self-hosted) 2026.03: April 3, 2026 * Release notes publication date: March 9, 2026 ## Announcement {% hint style="warning" %} In the DGC 2026.07 release, changes to the DQ-DGC integration API caused an incompatibility with Classic DQ 2026.03 and earlier when DQ-DGC auto-integration is enabled. This issue is resolved in DGC 2026.07.4. If you are using this configuration, upgrade to DGC 2026.07.4 to restore compatibility. {% endhint %} ## New and improved ### Platform {% hint style="danger" %} Starting in Data Quality & Observability (self-hosted) 2026.03, all Data Quality & Observability (self-hosted) repo keys used to pull images for self-hosted Kubernetes installations will expire on April 30, 2026. You must update your workflows before this date to avoid disruptions.As of Data Quality & Observability (self-hosted) 2026.02, a self-service option to download temporary repo keys from the "Docker Images" section of the [Downloads page](https://productresources.collibra.com/downloads/data-quality-observability-classic-2026-02/) (login required) is available. Temporary repo keys are valid for 5 days from the download date. The process of requesting long-lived keys from Collibra Support will no longer be supported.To ensure deployment stability, we recommend using the temporary keys to download images, uploading them to your internal Artifactory or registry, and generating internal credentials for production deployments. There is no impact to self-hosted Spark Standalone installations. For instructions on pulling images from the Collibra registry, go to [Install on self-hosted Kubernetes](/data-quality-self-hosted/installation/installing-data-quality-observability-self-hosted-on-self-hosted-kubernetes/install-on-self-hosted-kubernetes.md) . {% endhint %} {% hint style="warning" %} You must configure the [ALLOWED\_LOCAL\_PATHS](/data-quality-self-hosted/installation/installing-data-quality-observability-self-hosted-on-self-hosted-spark-standalone/additional-spark-standalone-configuration-options.md) variable in either "owl-env.sh" or "ConfigMap" to specify all local file base paths where files, such as drivers, are stored. Ensure you set this variable for both "dq-web" and "dq-agent" configurations if the configurations are not shared.For example: If DQ Classic is installed to /home/collibradq/owl and JDBC drivers are located at /home/collibradq/owl/drivers, then /home/dq/config/owl-env.sh should include the following: `export ALLOWED_LOCAL_PATHS=/opt/collibradq/owl`If drivers are stored in a different directory structure, such as /data/storage/owl/drivers, then you must append that to the configuration. {% endhint %} {% hint style="warning" %} If you receive an error when loading the Explorer page for datasets, verify and update the `qualified=` parameter in the URL. Ensure the value is in the correct schema.table format. {% endhint %} * The Amazon Athena Simba JDBC driver is now updated to version 2.2.2. * You can now pass an encrypted SAML keystore password to the SAML setup. * The ENABLE\_STRICT\_TAGGING flag on the Configuration Settings > Application Config page now defaults to "false." This ensures that the behavior of metatags remains unchanged for existing customers. * Connection URLs now include all connection parameters when rewritten with an authentication token. This ensures complete and accurate URL generation. ### Jobs * The struct column type for the SDO\_GEOMETRY SQL data type is now correctly identified as "unsupported" when creating or editing datasets. ### Findings * Source export now consolidates multiple files for SOURCE\_VALUE, SOURCE\_COUNT, and SOURCE\_SCHEMA into a single SOURCE Excel file, matching the behavior in legacy. ### Rules * The “Custom Template Rules” rule query value from Rule Builder now persists on the Findings > Rules page and the Rule Definitions page. ### Dataset Manager * You can now choose metatags from a pre-selected drop-down list when ENABLE\_STRICT\_TAGGING is set to "false." ENABLE\_STRICT\_TAGGING is set to "true," you can add your own metatags and also select from the pre-selected list. This replaces the previous t1, t2, t3, and t4 metatag fields with pre-selected and user-approved values. ### APIs * Time zone settings are now properly preserved during export and import processes when using APIs. ### Collibra Platform integration * Dimensions are now correctly assigned to layers, even if the dimension name appears multiple times in the DQ\_dimension table. Duplicate dimension names are now supported, ensuring proper layer assignment when the DQ\_dimension table is correctly configured. * When you delete a rule in Data Quality & Observability (self-hosted), it is now marked as "Deleted" in Collibra Platform instead of "Suppressed." Additionally, deleted rules are no longer included in the export object or JSON. ## Fixes ### Platform * Pendo no longer causes a "pendo.clearSession()" error when logging in with a block script set to false. Additionally, Pendo does not activate upon login when it is blocked internally. ### Jobs * White spaces in dataset source queries are now handled correctly. Schema and table names now show properly in the metadata bar and the Dataset Manager page. * You no longer receive an error message on the Explorer page when local file management configuration is disabled. ### Findings * When you re-run a job after invalidating fuzzy match duplicates, saving, and retraining on the Findings page, the job no longer contains the same invalidated fuzzy match duplicates. ### Rules * Secondary rules now support functions in the `DATE_FORMAT(DATE('${rd}'), 'yyyyMMdd')` query. * Alerts are no longer blocked when an adaptive rule with an associated alert is configured to never exist. Data Quality & Observability Classic now handles the absence of adaptive rule behavior gracefully, ensuring that job re-runs do not trigger alerts or send emails. Additionally, the email processing queue isolates failures, preventing errors in one alert payload from blocking downstream alerts. ### APIs * Time zone settings are now properly preserved during export and import processes when using APIs. ### Collibra Platform integration * Dataset integration now succeeds even when metatags contain spaces. Spaces in metatag values are automatically replaced with hyphens during saving to prevent integration errors. * The toaster message for the auto map feature now states "Tables with the provided schema \ are not found" when no tables or views exist, instead of just stating "The schema was not found." ## Patch release ### 2026.03.1 * Data Quality & Observability (self-hosted) now starts successfully when the SAML tenant manager is enabled. Previously, the web application failed to initialize after upgrading to version 2026.03 if SAML\_TENANT\_MANAGER\_ENABLED was set to true.