proxysql

Commit Graph

Author	SHA1	Message	Date
Rahim Kanji	720441af20	Refactored and Optimized Query Cache with PgSQL Support * Added Query Cache Support for PgSQL: * Introduced new variables for PgSQL Query Cache management: * pgsql-query_cache_size_MB * pgsql-query_cache_soft_ttl_pct * pgsql-query_cache_stores_empty_result * pgsql-query_cache_handle_warnings (not yet utilized) * Overwriting Transaction Status: * Sending current transaction state to the client when a result is returned from the query cache. * This currently does not differentiate between normal and error transaction states—both are treated as active transaction states. * Cache Lifetime Management: Replaced manual reference counting with std::shared_ptr to efficiently manage the lifecycle of cache entries, ensuring proper memory deallocation and reducing complexity. * Unified Purging Logic: Refactored the query cache purging logic to use a single thread for both MySQL and PgSQL cache entries. * Code Cleanup: Removed unnecessary methods and data members to streamline the codebase and improve maintainability. * Performance Optimization: Improved memory management, reducing memory footprint and enhancing performance.	1 year ago

Author

SHA1

Message

Date

Rahim Kanji

720441af20

Refactored and Optimized Query Cache with PgSQL Support

* Added Query Cache Support for PgSQL:

	* Introduced new variables for PgSQL Query Cache management:
		* pgsql-query_cache_size_MB
		* pgsql-query_cache_soft_ttl_pct
		* pgsql-query_cache_stores_empty_result
		* pgsql-query_cache_handle_warnings (not yet utilized)
	* Overwriting Transaction Status:
		* Sending current transaction state to the client when a result is
 returned from the query cache.
		* This currently does not differentiate between normal and error
transaction states—both are treated as active transaction states.

* Cache Lifetime Management:
	Replaced manual reference counting with std::shared_ptr to efficiently manage the lifecycle of cache entries,
ensuring proper memory deallocation and reducing complexity.

* Unified Purging Logic:
	Refactored the query cache purging logic to use a single thread for both
MySQL and PgSQL cache entries.

* Code Cleanup:
	Removed unnecessary methods and data members to streamline the codebase
and improve maintainability.

* Performance Optimization:
	Improved memory management, reducing memory footprint and enhancing
performance.

1 Commits (8ebd7e48ae58bf02fa40b4e1ae0bea801db41f52)