Overview
After a migration task is started, if it is interrupted due to an exception, you can manually retry it without the need to pull data all over again.
Full stage: The chunk-based import/export mechanism is implemented in this stage. Each chunk is assigned a unique identifier. With these identifiers, DTS can identify the unsynced chunk and resume the data sync from there when a disrupted task is restarted.
Incremental stage: The checkpoint-based mechanism is implemented in this stage. When DTS parses the source database binlogs before writing them to the target database, a checkpoint message is inserted to indicate the offset of the data being synced every 10 seconds. With these checkpoint messages, DTS can resume the data sync from the checkpointed offset when a disrupted task is restarted.
Below describes the scenarios where the task retry is supported for data migration:
|
Data migration links among MySQL, TDSQL-C for MySQL, MariaDB, and Percona. | Full export Full import Incremental sync | Description for full export: Retry is only supported for exporting unlocked tables with primary keys. Retry is only supported when a task is interrupted by table locking failures due to time-consuming SQL statements in the source instance. It doesn’t work in other cases. |
Migration links among MySQL, TDSQL-C for MySQL, MariaDB, Percona, and TDSQL for MySQL. | Incremental sync | - |
Application Scope
Data migration links among MySQL, TDSQL-C for MySQL, MariaDB, Percona, and TDSQL for MySQL.
Directions
Log in to the DTS console, select Data Migration on the left sidebar, select the target migration task, and click Retry in the Operation column.
Was this page helpful?