v.24.11Improvement

Improved Read Performance for Asynchronous Metrics in Heavy Load Scenarios

Previously reading from system.asynchronous_metrics would wait for concurrent update to finish. This can take long time if system is under heavy load. With this change the previously collected values can always be read. #71798 (Alexander Gololobov).
Reading from system.asynchronous_metrics no longer waits for concurrent updates to complete, allowing immediate access to the most recently collected metric values.

Why it matters

This feature addresses the issue where reading from system.asynchronous_metrics could block and wait for concurrent updates to finish, which could cause long delays under heavy system load. By enabling immediate reads of the last collected metrics, it improves query responsiveness and reduces wait times for users monitoring asynchronous metrics.

How to use it

Users can directly query the system.asynchronous_metrics table as before. The change is automatic and does not require any configuration; queries will now return the last collected values immediately without blocking for ongoing updates.