【Tips】DolphinScheduler Task Data Cleanup and Backup Strategy to Ensure Page Smoothness

Problem Description

As Apache DolphinScheduler runs for an extended period, the number of tasks increases. The task data is stored in the database's t_ds_task_instance and t_ds_process_instance tables. The continuous data growth in these two tables leads to system page lag.

Solution

To address the above issue, the measure taken is to regularly clean up data in the t_ds_process_instance and t_ds_task_instance tables that is more than one month old.

Data Backup

Before cleaning up the data, first back up the original table data to ensure data safety.

use dolphinscheduler;
-- Create backup tables t_ds_process_instance_backup20241120 and t_ds_task_instance_backup20241120
CREATE TABLE t_ds_process_instance_backup20241120 LIKE t_ds_process_instance;
CREATE TABLE t_ds_task_instance_backup20241120 LIKE t_ds_task_instance;
-- Backup the original table data into the corresponding backup tables
INSERT INTO t_ds_process_instance_backup20241120
SELECT * FROM t_ds_process_instance;
INSERT INTO t_ds_task_instance_backup20241120
SELECT * FROM t_ds_task_instance;

Check Backup Status

To ensure the backup operation is successful, you can check the data row count of the backup tables and the original tables.

-- Check the data row count of the backup tables
SELECT COUNT(*) FROM t_ds_process_instance_backup20241120;
SELECT COUNT(*) FROM t_ds_task_instance_backup20241120;
-- Check the data row count of the original tables
SELECT COUNT(*) FROM t_ds_process_instance;
SELECT COUNT(*) FROM t_ds_task_instance;

Data Cleanup

After the backup is completed and confirmed to be error-free, perform the cleanup operation to delete data before October 19, 2024, 23:59:59.

-- Delete data before October 19, 2024, 23:59:59 in the t_ds_task_instance table
DELETE FROM t_ds_task_instance
WHERE submit_time < '2024-10-19 23:59:59';
-- Delete data before October 19, 2024, 23:59:59 in the t_ds_process_instance table
DELETE FROM t_ds_process_instance
WHERE end_time < '2024-10-19 23:59:59';