DM 2.0 RC Release Notes

Release date: August 21, 2020

DM version: 2.0.0-rc

Improvements

• Support high availability for data migration tasks
• Add an optimistic mode for sharding DDL statements
• Add the handle-error command to handle errors during DDL incremental replication
• Add a workaround field in the error returned by query-status to suggest the error handling method
• Improve the monitoring dashboards and alert rules
• Replace Mydumper with Dumpling as the full export unit
• Support the GTID mode when performing incremental replication to the downstream
• Support TLS connections between upstream and downstream databases, and between DM components
• Support the incremental replication scenarios where the table of the downstream has more columns than that of the upstream
• Add a --remove-meta option to the start-task command to clean up metadata related to data migration tasks
• Support dropping columns with single-column indices
• Support automatically cleaning up temporary files after a successful full import
• Support checking whether the table to be migrated has a primary key or a unique key before starting a migration task
• Support connectivity check between dmctl and DM-master while starting dmctl
• Support connectivity check for downstream TiDB during the execution of start-task/check-task
• Support replacing task names with task configuration files for some commands such as pause-task
• Support logs in json format for DM-master and DM-worker components
• Remove the call stack information and redundant fields in the error message returned by query-status
• Improve the binlog position information of the upstream database returned by query-status
• Improve the processing of auto resume when an error is encountered during the full export

Bug fixes

• Fix the issue of goroutine leak after executing stop-task
• Fix the issue that the task might not be paused after executing pause-task
• Fix the issue that the checkpoint might not be saved correctly in the initial stage of incremental replication
• Fix the issue that the BIT data type is incorrectly handled during incremental replication

Detailed bug fixes and changes

• Support high availability for data migration tasks #473
• Add an optimistic mode for sharding DDL statements #568
• Add the handle-error command to handle errors during DDL incremental replication #850
• Add a workaround field in the error returned by query-status to suggest the error handling method #753
• Improve the monitoring dashboards and alert rules #853
• Replace Mydumper with Dumpling as the full export unit #540
• Support the GTID mode when performing incremental replication to the downstream #521
• Support TLS connections between upstream and downstream databases, and between DM components #569
• Support the incremental replication scenarios where the table of the downstream has more columns than that of the upstream #379
• Add a --remove-meta option to the start-task command to clean up metadata related to data migration tasks #651
• Support dropping columns with single-column indices #801
• Support automatically cleaning up temporary files after a successful full import #770
• Support checking whether the table to be migrated has a primary key or a unique key before starting a migration task #870
• Support connectivity check between dmctl and DM-master while starting dmctl #786
• Support connectivity check for downstream TiDB during the execution of start-task/check-task #769
• Support replacing task names with task configuration files for some commands such as pause-task #854
• Support logs in json format for DM-master and DM-worker components #808
• Remove the call stack information in the error message returned by query-status #746
• Remove the redundant fields in the error message returned by query-status #771
• Improve the binlog position information of the upstream database returned by query-status #830
• Improve the processing of auto resume when an error is encountered during the full export #872
• Fix the issue of goroutine leak after executing stop-task #731
• Fix the issue that the task might not be paused after executing pause-task #644
• Fix the issue that the checkpoint might not be saved correctly in the initial stage of incremental replication #758
• Fix the issue that the BIT data type is incorrectly handled during incremental replication #876