Why does DM report the error `ERROR 1236 (HY000): The slave is connecting using CHANGE MASTER TO MASTER_AUTO_POSITION = 1, but the master has purged binary logs containing GTIDs that the slave requires.` if it restarts during an incremental task?

FAQ
1

This error indicates that the upstream binlog position recorded in the metadata file output by the dump unit has been purged during the full migration.

If this issue occurs, you need to pause the task, delete all migrated data in the downstream database, and start a new task with the --remove-meta option.

You can avoid this issue in advance by configuring in the following ways:

  1. Increase the value of expire_logs_days in the upstream MySQL database to avoid wrongly purging needed binlog files before the full migration task completes. If the data volume is large, it is recommended to use dumpling and TiDB-Lightning at the same time to speed up the task.
  2. Enable the relay log feature for this task so that DM can read data from relay logs even though the binlog position is purged.