This topic covers the feature updates for Realtime Compute for Apache Flink released on April 20, 2026.
For the specific upgrade schedule, check the latest announcements on the right side of the Realtime Compute for Apache Flink console. If you cannot access the new features, your account has not yet been upgraded. If you need an expedited upgrade, submit a ticket.
Overview
On April 20, 2026, Realtime Compute for Apache Flink launched a new version focused on AIOps, end-to-end observability, and fine-grained permission control. Key enhancements include AI-assisted diagnostics and tuning, job status push notifications, expanded permission capabilities, user experience optimizations, and enhanced OpenAPI capabilities.
Platform
New features
Flink AI assistant: This feature introduces a new agent for job diagnostics and tuning. You can initiate a diagnosis with a single click from the Flink AI assistant entry point or the job O&M page. The system automatically generates a diagnostic report with root cause analysis and optimization suggestions. It also supports one-click execution of tuning operations, creating a complete workflow from problem detection and root cause analysis to final tuning.
Message delivery: This feature allows you to push critical Flink job status changes to external systems. This capability covers the entire job lifecycle, helping you build more robust real-time monitoring and alerting systems.
Fine-grained permissions: You can now grant users a dedicated permission to unlock SQL/YAML job drafts. A user can unlock a job draft if they have either the "Develop SQL/YAML jobs (create, edit)" permission or the "Unlock SQL/YAML job drafts" permission. This enhancement supports stricter permission control by isolating development and production environments.
Enhancements
Workspace cloning:
The required permission for cloning a workspace is now Editor instead of Owner, making the feature more accessible.
You can automatically create a snapshot of a running streaming job before cloning to ensure data consistency.
Improved error messages for cross-architecture migrations and authorization exceptions accelerate troubleshooting.
Autotuning:
This release adds autotuning support for data ingestion jobs (CDC YAML).
The autotuning feature is now more tightly integrated with the job health score and dynamic update capabilities, improving the accuracy and timeliness of tuning decisions.
Member management: When you add a RAM role as a project member, the member list now directly displays the RAM role name. This improves the readability and manageability of member information.
Data lineage: The entry point for this feature is now in a submenu in the O&M center to simplify navigation.
OpenAPI
This release includes several optimizations for OpenAPI interfaces related to job drafts:
You can use the OpenAPI to create, update, query, deploy, and delete data ingestion job drafts (CDC YAML). You can also query a list of data ingestion job drafts.
The renaming mechanism for job drafts has been improved.