Migration Task Precheck
Before using DM to migrate data from upstream to downstream, a precheck helps detect errors in the upstream database configurations and ensures that the migration goes smoothly. This document introduces the DM precheck feature, including its usage scenario, check items, and arguments.
To run a data migration task smoothly, DM triggers a precheck automatically at the start of the task and returns the check results. DM starts the migration only after the precheck is passed.
To trigger a precheck manually, run the
tiup dmctl check-task ./task.yaml
Descriptions of check items
After a precheck is triggered for a task, DM checks the corresponding items according to your migration mode configuration.
This section lists all the precheck items.
If a mandatory check item does not pass, DM returns an error after the check and does not proceed with the migration task. In this case, modify the configurations according to the error message and retry the task after meeting the precheck requirements.
If a non-mandatory check item does not pass, DM returns a warning after the check. DM automatically starts a migration task if the check result contains only warnings but no errors.
Common check items
Regardless of the migration mode you choose, the precheck always includes the following common check items:
MySQL version > 5.5
MariaDB version >= 10.1.2
Compatibility of the upstream MySQL table schema
- Check whether the upstream tables have foreign keys, which are not supported by TiDB. A warning is returned if a foreign key is found in the precheck.
- (Mandatory) Check whether there are compatibility differences in character sets. For more information, see TiDB Supported Character Sets.
- (Mandatory) Check whether the upstream tables have primary key constraints or unique key constraints (introduced from v1.0.7)
Check items for full data migration
For the full data migration mode (
task-mode: full), in addition to the common check items, the precheck also includes the following check items:
(Mandatory) dump permission of the upstream database
- SELECT permission on INFORMATION_SCHEMA and dump tables
- RELOAD permission if
- LOCK TABLES permission on the dump tables if
(Mandatory) Consistency of upstream MySQL multi-instance sharding tables
In the pessimistic mode, check whether the table schemas of all sharded tables are consistent in the following items:
- Number of columns
- Column name
- Column order
- Column type
- Primary key
- Unique index
In the optimistic mode, check whether the schemas of all sharded tables meet the optimistic compatibility.
If a migration task was started successfully by the
start-taskcommand, the precheck of this task skips the consistency check.
Auto-increment primary key in sharded tables
- If sharded tables have auto-increment primary keys, the precheck returns a warning. If there are conflicts in auto-increment primary keys, see Handle conflicts of auto-increment primary key for solutions.
Check items for incremental data migration
For the incremental data migration mode (
task-mode: incremental), in addition to the common check items, the precheck also includes the following check items:
(Mandatory) Upstream database REPLICATION permission
- REPLICATION CLIENT permission
- REPLICATION SLAVE permission
(Mandatory) Database primary-secondary configuration
- The database ID
server_idof the upstream database must be specified (GTID is recommended for non-AWS Aurora environments).
- The database ID
(Mandatory) MySQL binlog configuration
- Check whether binlog is enabled (required by DM).
- Check whether
binlog_format=ROWis configured (DM only supports the migration of binlog in the ROW format).
- Check whether
binlog_row_image=FULLis configured (DM only supports
binlog_ignore_dbis configured, check whether the database tables to be migrated meet the conditions of
(Mandatory) Check if the upstream database is in an Online-DDL process (in which the
ghosttable is created but the
renamephase is not executed yet). If the upstream is in the online-DDL process, the precheck returns an error. In this case, wait until the DDL to complete and retry.
Check items for full and incremental data migration
For the full and incremental data migration mode (
task-mode: all), in addition to the common check items, the precheck also includes the full data migration check items and the incremental data migration check items.
Ignorable check items
Prechecks can find potential risks in your environments. It is not recommended to ignore check items. If your data migration task has special needs, you can use the
ignore-checking-items configuration item to skip some check items.
|Checks the dump privilege of the user in the upstream MySQL instance.|
|Checks the replication privilege of the user in the upstream MySQL instance.|
|Checks the version of the upstream database.|
|Checks whether server_id is configured in the upstream database.|
|Checks whether binlog is enabled in the upstream database.|
|Checks the compatibility of the table schemas in the upstream MySQL tables.|
|Checks the consistency of the table schemas in the upstream MySQL multi-instance shards.|
|Checks whether the auto-increment primary key conflicts in the upstream MySQL multi-instance shards.|
Configure precheck arguments
The migration task precheck supports processing in parallel. Even if the number of rows in sharded tables reaches a million level, the precheck can be completed in minutes.
To specify the number of threads for the precheck, you can configure the
threads argument of the
mydumpers field in the migration task configuration file.
mydumpers: # Configuration arguments of the dump processing unit global: # Configuration name threads: 4 # The number of threads that access the upstream when the dump processing unit performs the precheck and exports data from the upstream database (4 by default) chunk-filesize: 64 # The size of the files generated by the dump processing unit (64 MB by default) extra-args: "--consistency none" # Other arguments of the dump processing unit. You do not need to manually configure table-list in `extra-args`, because it is automatically generated by DM.