1. RELEASE INFO
|
Product |
AGILE DATA ENGINE |
|
Release |
25.2 (25.2.637) |
|
Release date |
21st of October 2025 |
|
Release status |
RELEASED |
|
Release type |
MAJOR.MINOR.UPDATE.PATCH |
2. CONTENT OF THE RELEASE
2.1 Highlights
-
User management version 2 together with Admin UI to manage users and roles are generally available now
-
Insights has a new Search for metadata and custom SQL (magnifier icon in header)
-
New Snowflake physical types DYNAMIC_TABLE and ICEBERG_TABLE in public preview
-
Snowflake and Google BigQuery search optimization support with selected columns only
-
New authentication types supported: Snowflake PAT token and Azure SQL Database Entra ID Service Principal
-
User experience improved with header area improvements including breadcrumb in Designer
-
Inherited load schedule(s) listed in load UI when no explicit schedule is selected in Designer
2.2 New features
-
The default user management is changed to the version 2, which will enable authentication to multi-tenant services like Admin UI and Insights in future. See Notifications below.
-
New Snowflake physical types supported: DYNAMIC_TABLE and ICEBERG_TABLE
-
Iceberg tables' base location is set to ‘<entity_schema>_<entity_name>’
-
Dynamic tables are by default refreshed in case their parent entities are being modified
-
The default refresh can be disabled with the environment variable system.deployment.entity_recreate.dependent_dynamic_table.refresh
-
DYNAMIC_TABLE will require STORAGE.CHANGE_TRACKING on all source entities.
-
-
New authentication type: Snowflake PAT token
-
New authentication type: Azure SQL Database Entra ID Service Principal
-
Search optimization support with certain attributes only in Snowflake, see SEARCH_INDEX_KEY and STORAGE.SEARCH_OPTIMIZATION_OPTIONS
-
Support for Google BigQuery’s SEARCH INDEX, see the new key type SEARCH_INDEX_KEY, STORAGE.SEARCH_OPTIMIZATION and STORAGE.SEARCH_OPTIMIZATION_OPTIONS.
-
Configurable deletion method for entities (rename as backup vs. drop) with environment variables:
-
Accepted values: DROP, RENAME_AS_BACKUP (default), RENAME_AS_BACKUP_BY_RECREATE (since no ALTER command for base location with Iceberg tables)
2.3. Improvements
-
User experience of Designer header area is improved and changed including new breadcrumb
-
Inherited load schedule(s) listed in load UI when no explicit schedule is selected in Designer
-
Database tags and keys are supported in entity import and export (CSV)
-
Continue with Authenticate login button is changed to Authenticate
-
Multi-line string parameters with escaping are supported in Freemarker Templates (FTL)
-
References' metadata is included in Freemarker Template scope (FTL)
-
Split VIEW RECREATE as DROP and CREATE in case the name is being reused by another entity
-
Multi-active satellite’s SQL pattern is changed from LEFT JOIN to aggregate based logic in Google BigQuery
2.4 Fixes
Designer
-
Move entity to another package: uncommitted changes should be shown in both packages
-
Unmapped entities detection from SQL not working when editing in the preview mode
-
Management of UUIDs and Structs in entity import from a source file
-
Entity search when creating outgoing load does not support schema
-
References do not work in Entity Import (CSV) when managed entities are involved
-
Trying to delete schedule which is upstream in precondition gives error
-
Code preview crashes in case logical datatype has no mapping for Google BigQuery
Deployment Management
-
Entity modification quota exceeds in Google BigQuery in case there are tens of changes for a single entity – an extra sleep time between commands is added to avoid the quota exceeding (for further details check the quota and limitations of Google BigQuery)
-
Google BigQuery database connection handling improved
-
Deployment successful after ENTITY_RECREATION_PREVENTED_BY_CHANGE_TRACKING, should be a warning
-
Unknown technical attribute type does not throw an error while deploying, no dependency checked
-
Deployment API improvement on multi-az failover
Templating Framework
-
General fixes to templating framework
-
Fixes to Config Load Templates' changes section in commit view
-
target_entity_attribute_list behaves different than documented in file loads
-
target_entity_attribute_list_with_transform_cast_parquet should have a colon instead of double colon when referencing an attribute
Other
-
Fixed: Affected rows not read correctly in Databricks loads
-
Improved session/query timeout settings for Azure SQL Database
-
Improved logging of network errors for Azure SQL Database
-
Third party software version updates
2.5 Insights
-
New: Tag and tag group filter to Release Notes and Issue Tracking
-
Fixed: Long load names overflows in KVP component
-
Fixed: Issue Tracking items ordered descending based on issue number (not the whole ticket id string)
3. NOTIFICATIONS
3.1 The default user management is migrated to v2 for non-federated cases
Customers who have still had the user management v1 in use, and have no federated Entra ID, will be migrated to the new user management v2 in this release. This will enable authentication to multi-tenant services like End-user Admin UI and Insights add-on (if subscribed). After the migration the login page is changed to ADE theme and the users are required to setup their password and MFA anew with the standard Password recovery functionality. Support portal can be used to request tenantAdmin roles in End-user Admin UI for system administrator users.
3.2 Multi-active satellite SQL has been changed with Google BigQuery
NOTE! That the multi-active satellite’s generated SQL logic has changed from LEFT JOIN to aggregate based logic with Google BigQuery. The results should be correct after the next time the load gets generated (ie. the package with such an entity is deployed). But it is good to double-check the generated logic after the upgrade with the code preview in Designer and compare with the old logic. This is a fix for reported problems with the original SQL logic.
3.3 Known issue that the code preview does not show CREATE <physical type> syntax in load UI
There is a known issue that CREATE <physical type> syntax cannot be viewed with code preview in load UI. This would be relevant for VIEWs, MATERIALIZED VIEWs and DYNAMIC TABLEs. A change for this is under development.