Filter Binlog Events

This document describes how to filter binlog events when you use DM to perform continuous incremental data replication. For the detailed replication instructions, refer to the following documents by scenarios:

• Migrate Small Datasets from MySQL to TiDB
• Migrate Large Datasets from MySQL to TiDB
• Migrate and Merge MySQL Shards of Small Datasets to TiDB
• Migrate and Merge MySQL Shards of Large Datasets to TiDB

Configuration

To use binlog event filter, add a filter to the task configuration file of DM, as shown below:

filters:
  rule-1:
    schema-pattern: "test_*"
    table-pattern: "t_*"
    events: ["truncate table", "drop table"]
    sql-pattern: ["^DROP\\s+PROCEDURE", "^CREATE\\s+PROCEDURE"]
    action: Ignore

• schema-pattern/table-pattern: Filters matching schemas or tables

• events: Filters binlog events. Supported events are listed in the table below:

  Event	Category	Description
  all		Includes all events
  all dml		Includes all DML events
  all ddl		Includes all DDL events
  none		Includes no event
  none ddl		Excludes all DDL events
  none dml		Excludes all DML events
  insert	DML	Insert DML event
  update	DML	Update DML event
  delete	DML	Delete DML event
  create database	DDL	Create database event
  drop database	DDL	Drop database event
  create table	DDL	Create table event
  create index	DDL	Create index event
  drop table	DDL	Drop table event
  truncate table	DDL	Truncate table event
  rename table	DDL	Rename table event
  drop index	DDL	Drop index event
  alter table	DDL	Alter table event

• sql-pattern: Filters specified DDL SQL statements. The matching rule supports using a regular expression.

• action: Do or Ignore

  • Do: the allow list. A binlog event is replicated if meeting either of the following two conditions:

    • The event matches the rule setting.
    • sql-pattern has been specified and the SQL statement of the event matches any of the sql-pattern options.

  • Ignore: the block list. A binlog event is filtered out if meeting either of the following two conditions:

    - The event matches the rule setting.
    - sql-pattern has been specified and the SQL statement of the event matches any of the sql-pattern options.
    

    If both Do and Ignore are configured, Ignore has higher priority over Do. That is, an event satisfying both Ignore and Do conditions will be filtered out.

Application scenarios

This section describes the application scenarios of binlog event filter.

Filter out all sharding deletion operations

To filter out all deletion operations, configure a filter-table-rule and a filter-schema-rule, as shown below:

filters:
  filter-table-rule:
    schema-pattern: "test_*"
    table-pattern: "t_*"
    events: ["truncate table", "drop table", "delete"]
    action: Ignore
  filter-schema-rule:
    schema-pattern: "test_*"
    events: ["drop database"]
    action: Ignore

Migrate only DML operations of sharded schemas and tables

To replicate only DML statements, configure two Binlog event filter rule, as shown below:

filters:
  do-table-rule:
    schema-pattern: "test_*"
    table-pattern: "t_*"
    events: ["create table", "all dml"]
    action: Do
  do-schema-rule:
    schema-pattern: "test_*"
    events: ["create database"]
    action: Do

Filter out SQL statements not supported by TiDB

To filter out SQL statements not supported by TiDB, configure a filter-procedure-rule, as shown below:

filters:
  filter-procedure-rule:
    schema-pattern: "*"
    sql-pattern: [".*\\s+DROP\\s+PROCEDURE", ".*\\s+CREATE\\s+PROCEDURE", "ALTER\\s+TABLE[\\s\\S]*ADD\\s+PARTITION", "ALTER\\s+TABLE[\\s\\S]*DROP\\s+PARTITION"]
    action: Ignore

See also

Filter Binlog Events Using SQL Expressions