Notice
- Before you upgrade from a previous version to ODC V4.4.1, you must read the Considerations section in this topic.
- Also, if you want to use the features of this version, you must be aware of the ODC V4.4.1 limitations.
Version description
ODC V4.4.1 focuses on enhancing job management capabilities, improving ticket collaboration efficiency, and optimizing the intelligent SQL development experience.
A new job module is introduced, integrating data archiving, cleaning, SQL, and partitioning plans. It supports multi-dimensional filtering and search, and independently manages the lifecycle of jobs and approval flows, improving operational clarity and execution efficiency.
The ticket engine is comprehensively upgraded, unifying the state machines and execution methods for all ticket types. New states such as pre-check and queued are added, optimizing the user operation path. It supports persistent filtering conditions and enhances task classification, improving collaboration efficiency.
NL2SQL is introduced, deeply integrated with Tongyi Qianwen, DeepSeek, Doubao, and mainstream large models compatible with OpenAI API. It allows users to generate, correct, rewrite, and optimize SQL statements in the SQL window using natural language, making data queries more intelligent and efficient.
For more information about the version, please refer to the feature updates.
Considerations
Upgrade notes
- Please ensure that your ODC version is V4.3.0 or later. If not, upgrade to V4.3.0-V4.4.0 first.
- Upgrading ODC V4.3.x to V4.4.1 involves data migration and changes. Back up the metadata database before the upgrade.
Upgrade window
- This upgrade involves the migration of historical data of scheduled task tickets. If your historical data is large, the migration will take longer. It is known that migrating 1 million tickets takes 20 minutes. Therefore, schedule the upgrade during a business idle window.
- To minimize the upgrade time, follow the Pre-upgrade operations section in Preparations for upgrade to migrate historical data first, then proceed with the upgrade.
Ticket tasks
- If there are pending tickets before the upgrade, they may be canceled.
Scheduled tasks
This upgrade migrates data archiving, data cleaning, SQL plans, and partitioning plans to the job module.
If there are pending approval flows during the upgrade, the approval flows may be terminated after the upgrade.
If there are open windows for data archiving, data cleaning, partitioning plans, or SQL plans in the SQL console before the upgrade, error messages may appear after the upgrade. Reopening the corresponding windows will no longer show errors.
Partitioning plans
- After migration, the new job numbers will differ from the original ticket numbers.
- Partitioning plans that have not been approved (pending approval, expired, or rejected) will not be displayed in the job list.
- Partitioning plans with empty partitioning strategies will not be displayed in the job list.
Execution records
- After migration, the historical execution record numbers in new jobs will differ from the original execution records.
- If there are pending execution records for partitioning plans or SQL plans, the execution record status may not match the actual situation after the upgrade. You can correct the data after the upgrade. For details, see the Correct data section in Upgrade verification.
- Historical execution records for partitioning plans and SQL plans will still be displayed in the database change list. If you do not want to display them, you can hide the data after the upgrade. For details, see the Hide data section in Upgrade verification.
Logical database changes
- After the upgrade, historical data of logical database change tickets will no longer be viewable. You can retrieve historical change data from the metadata database. For details, see the Check the history of logical database changes section in Upgrade verification.
ODC V4.4.1-BP1
Current version: V4.4.1-BP1
Previous version: V4.4.1
Release date: January 20, 2026
Supported upgrade path: You must first upgrade ODC V2.0.0 and later to V4.3.0–V4.4.0 before upgrading to this version.
Feature updates
This iteration fixes 8 defects, including the following key defects:
Fixed the issue where clicking the three dots on the right side does not allow switching the preview SQL details when setting partitioning strategies for a batch of partitions.
Fixed the issue where the generated partition name is incorrect when the partitioning plan name and partition value format are inconsistent.
Fixed the issue where the size of the primary shard is incorrectly applied to the secondary shard in data cleanup and data archiving tasks.
Fixed the issue where data cleanup and data archiving tasks cannot be rescheduled if a subtask is in a failed state.
Fixed the issue where the pre-check fails when the data archiving filter condition is a subquery.
Fixed the issue where partitions cannot be added to a list-partitioned table.
Fixed the issue where the client fails to upgrade to V4.4.1.
Fixed the issue where the frontend crashes when viewing the multi-database change template if the database has been deleted.
ODC V4.4.1
Current version: V4.4.1
Previous version: V4.4.1
Release date: December 10, 2025
Supported upgrade path: You must first upgrade ODC V2.0.0 and later to V4.3.0–V4.4.0 before you can upgrade it to this version.
Feature updates
Job module
The job module is newly introduced, providing job and execution perspectives. In the job perspective, you can view the definitions of all jobs. In the execution perspective, you can view the execution records of all tasks initiated by the jobs.
Data archiving, data cleanup, SQL plans, and partition migration plans are migrated to the job module. The state machines of these four types of jobs and job dispatching tasks are unified, and the operations that users can perform in each state are clarified.
The job module provides comprehensive and unified search and filtering capabilities.
- In the job perspective, you can filter jobs by job type, job status, approval status, project, and creation time.
- In the job perspective, you can search for jobs by job name, job ID, creator, database, data source, cluster, and tenant.
- In the execution perspective, you can filter tasks by job type, job status, project, and creation time.
- In the execution perspective, you can search for tasks by task ID, job name, job ID, creator, database, data source, cluster, and tenant.
The job perspective provides a filter for pending approvals, allowing you to quickly find jobs that require your approval.
The job module supports persistent filtering and sorting conditions.
The approval process and lifecycle of jobs are independent.
The execution methods of all job types are unified (immediate execution, scheduled execution, and periodic execution).
Ticket engine upgrade
The ticket engine is comprehensively upgraded. The state machines and execution methods of all ticket types are unified. The states of "pre-checking", "queuing", and "execution success (including alerts)" are added, while the states of "rolling back", "rolling back failed", "rolled back", and "completed" are deprecated. The operations that users can perform in each state are clarified.
The filtering and search capabilities of tickets are upgraded.
- You can filter tickets by ticket type, ticket status, project, and creation time.
- You can search for tickets by ticket ID, ticket description, creator, database, data source, cluster, and tenant.
- The ticket list provides filters for "Pending Approval" and "Pending Execution", allowing you to quickly find tickets related to you.
- The filtering and sorting conditions of tickets are supported.
The execution methods of all ticket types are unified (manual execution, immediate execution, and scheduled execution).
NL2SQL
- NL2SQL is introduced, supporting the integration of external large models. Currently, it supports the integration of Qwen, DeepSeek, Doubao, and large models compatible with OpenAI API.
- The SQL window provides natural language to SQL capabilities, including SQL generation, error correction, rewriting, and optimization.
Workspace
- A new module for statistics on pending tasks (pending approval tickets, pending execution tickets, and pending approval jobs) is added.
- A new ticket statistics module is added, covering 10 types of tickets and 5 states. You can click the statistics to jump to the corresponding ticket list.
- The job statistics module is upgraded. You can click the statistics to jump to the corresponding job list.
- The task overview supports filtering tickets and jobs by project and time range.
- You can customize the display modules of the workspace.
Data archiving/cleanup
Pre-check is supported during job configuration. The pre-check checks the following items:
- Whether the archiving link is within the supported scope.
- Whether the filter conditions are correct and reasonable.
- Whether the field types are within the supported scope.
You can configure to ignore the execution timeout of job dispatching tasks.
The parameter names and descriptions of jobs are optimized to reduce the understanding cost.
The default values of data archiving throttling are adjusted. The default row throttling is 1,000 rows/s, and the default data size throttling is 10 MB/s.
Dependency check
A new dependency check capability is added. The following checks are supported:
- When a user is deleted, check whether there are unexpired jobs created by the user in the projects that the user participates in, and whether there are unexpired tickets, jobs, and job dispatching tasks in the user's personal space.
- When the project of a user is changed, check whether there are unexpired jobs created by the user in the project.
- When a data source is deleted, check whether there are unexpired tickets, jobs, and job dispatching tasks in the data source.
- When the project of a database is changed, check whether there are unexpired tickets, jobs, and job dispatching tasks in the database.
- When a project is archived, check whether there are unexpired tickets, jobs, and job dispatching tasks in the project.
SQL development
- The auto-completion capability of the SQL window is upgraded to support DML and DDL statements.
- The global search function is upgraded to allow you to search for databases, projects, and data sources without selecting a category first.
- Search entry points are provided for object categories and specific object names in the resource tree, allowing you to quickly initiate a global search.
OceanBase adaptation
- Java/Python UDFs can be managed in a GUI.
Security specifications
- To adapt to the restructuring of the ticket module and the addition of the job module, the task type in the risk identification rules is split into task type and job type.
Configuration center
- You can configure the minimum scheduling interval of jobs at the space level. The minimum value can be set to 1 minute.
Performance improvements
- The startup time of the desktop client is improved from 100 seconds to 20 seconds.
- The smoothness of scrolling large data tables is improved.