Details
-
Task
-
Status: Closed (View Workflow)
-
Major
-
Resolution: Won't Do
-
None
Description
multi thread for bulk insert when load data is very useful, so we implement it.
1. handler->ha_get_bg_result: means wait for the background thread in handler API, maybe we can use it in commit/rollback interface.
2. add parameter of "sql_type" in spider_bg_conn_search , to distinguish select and the other type.
3. spider_set_conn_bg_param_for_dml: check whether it can insert parallel mode. Maybe it should decided by a new spider parameter
4. spider_db_bulk_insert: main logic for parallel insert_lock_tables_list
5. spider_bg_conn_action: reuse the most logic. Take careful of dbton_handler->set_sql_for_exec
6. add spider_bgs_mode=1 spider_bg_dml=1 to control whether start multi thread