Managing active-active clusters - Amazon Relational Database Service (original) (raw)

The following stored procedures set up and manage RDS for MySQL active-active clusters. For more information, see Configuring active-active clusters for RDS for MySQL.

These stored procedures are only available with RDS for MySQL DB instances running the following versions:

Topics

mysql.rds_group_replication_advance_gtid

Creates placeholder GTIDs on the current DB instance.

Syntax

CALL mysql.rds_group_replication_advance_gtid(
begin_id
, end_id
, server_uuid
);

Parameters

begin_id

The start transaction ID to be created.

end_id

The end transaction ID to be created.

begin_id

The group_replication_group_name for the transaction to be created. The group_replication_group_name is specified as a UUID in the DB parameter group associated with the DB instance.

Usage notes

In an active-active cluster, for a DB instance to join a group, all GTID transactions executed on the new DB instance must exist on the other members in the cluster. In unusual cases, a new DB instance might have more transactions when transactions are executed before joining the instance to group. In this case, you can't remove any existing transactions, but you can use this procedure to create the corresponding placeholder GTIDs on the othe DB instances in the group. Before doing so, verify that the transactions don't affect the replicated data.

When you call this procedure, GTID transactions of server_uuid:begin_id-end_id are created with empty content. To avoid replication issues, don't use this procedure under any other conditions.

Important

Avoid calling this procedure when the active-active cluster is functioning normally. Don't call this procedure unless you understand the possible consequences of the transactions you are creating. Calling this procedure might result in inconsistent data.

Example

The following example creates placeholder GTIDs on current DB instance.:

CALL mysql.rds_group_replication_advance_gtid(5, 6, '11111111-2222-3333-4444-555555555555');

mysql.rds_group_replication_create_user

Creates the replication user rdsgrprepladmin for group replication on the DB instance.

Syntax

CALL mysql.rds_group_replication_create_user(
replication_user_password
);

Parameters

replication_user_password

The password of the replication user rdsgrprepladmin.

Usage notes

Example

The following example creates the replication user rdsgrprepladmin for group replication on the DB instance:

CALL mysql.rds_group_replication_create_user('password');

mysql.rds_group_replication_set_recovery_channel

Sets the group_replication_recovery channel for an active-active cluster. The procedure uses the reserved rdsgrprepladmin user to configure the channel.

Syntax

CALL mysql.rds_group_replication_set_recovery_channel(
replication_user_password);

Parameters

replication_user_password

The password of the replication user rdsgrprepladmin.

Usage notes

The password of the replication user rdsgrprepladmin must be the same on all of the DB instances in an active-active cluster. A call to the mysql.rds_group_replication_create_user specifies the password.

Example

The following example sets the group_replication_recovery channel for an active-active cluster:

CALL mysql.rds_group_replication_set_recovery_channel('password');

mysql.rds_group_replication_start

Starts group replication on the current DB instance.

Syntax

CALL mysql.rds_group_replication_start(
bootstrap
);

Parameters

bootstrap

A value that specifies whether to initialize a new group or join an existing group. 1 initializes a new group with the current DB instance. 0 joins the current DB instance to an existing group by connecting to the endpoints defined in group_replication_group_seeds parameter in the of DB parameter group associated with the DB instance.

Example

The following example initializes a new group with the current DB instance:

CALL mysql.rds_group_replication_start(1);

mysql.rds_group_replication_stop

Stops group replication on the current DB instance.

Syntax

CALL mysql.rds_group_replication_stop();

Usage notes

When you stop replication on a DB instance, it doesn't affect any other DB instance in the active-active cluster.