Scenarios
After a migration task is started, if the task is interrupted due to an exception, you can manually retry the task without re-pulling data from the beginning.
Full migration phase: The full migration phase is based on a chunk-based export/import mechanism. Each chunk is marked. After a task restarts, Data Transfer Service (DTS) identifies the position of the interrupted chunk and resumes data synchronization from the position of the unsynchronized chunk.
Incremental phase: The incremental phase is based on a checkpoint mechanism. When parsing the binlog of the source database and writing data to the target database, DTS generally inserts a checkpoint message every 10 seconds to mark the data synchronization checkpoint. After the task restarts, DTS resumes data synchronization from the interrupted checkpoint.
The retry operation is supported in the following data migration scenarios:
During the structure export phase or the export phase of tables without primary keys, retrying a task requires restarting synchronization from the beginning.
During the export phase of tables with primary keys, if the interval between the original synchronization start time X and the retry time Y is less than or equal to 48 hours, resumable transfer is supported; if the interval exceeds 48 hours, synchronization restarts from the beginning. After a manual or automatic restart that triggers a full resynchronization, the start time X will be reset.
During the full import or incremental synchronization phase, retrying a task supports resumable transfer.
Applicable Scope
Data migration links between MySQL, TDSQL-C for MySQL, MariaDB, Percona, and TDSQL for MySQL support retrying tasks.
Operation Steps
Log in to the DTS console, select Data Migration in the left sidebar, select the specified migration task, and click Retry in the Operation column.