MariaDB replication can be used to replication between MariaDB Galera Cluster and MariaDB Server. This article will discuss how to do that.
Before we set up replication, we need to ensure that the cluster is configured properly. This involves the following steps:
log_slave_updates=ON on all nodes in the cluster. See Configuring MariaDB Galera Cluster: Writing Replicated Write Sets to the Binary Log and Using MariaDB Replication with MariaDB Galera Cluster: Configuring a Cluster Node as a Replication Master for more information on why this is important. This is also needed to enable wsrep GTID mode. server_id to the same value on all nodes in the cluster. See Using MariaDB Replication with MariaDB Galera Cluster: Setting server_id on Cluster Nodes for more information on what this means. If you want to use GTID replication, then you also need to configure some things to enable wsrep GTID mode. For example:
wsrep_gtid_mode=ON needs to be set on all nodes in the cluster. wsrep_gtid_domain_id needs to be set to the same value on all nodes in the cluster, so that each cluster node uses the same domain when assigning GTIDs for Galera Cluster's write sets. log_slave_updates needs to be enabled on all nodes in the cluster. See MDEV-9855 about that. And as an extra safety measure:
gtid_domain_id should be set to a different value on all nodes in a given cluster, and each of these values should be different than the configured wsrep_gtid_domain_id value. This is to prevent a node from using the same domain used for Galera Cluster's write sets when assigning GTIDs for non-Galera transactions, such as DDL executed with wsrep_sst_method=RSU set or DML executed with wsrep_on=OFF set. Before we set up replication, we also need to ensure that the MariaDB Server slave is configured properly. This involves the following steps:
server_id to a different value than the one that the cluster nodes are using. gtid_domain_id to a value that is different than the wsrep_gtid_domain_id and gtid_domain_id values that the cluster nodes are using. log_bin and log_slave_updates=ON if you want the slave to log the transactions that it replicates. Our process to set up replication is going to be similar to the process described at Setting up a Replication Slave with Mariabackup, but it will be modified a bit to work in this context.
The very first step is to start the nodes in the first cluster. The first node will have to be bootstrapped. The other nodes can be started normally.
Once the nodes are started, you need to pick a specific node that will act as the replication master for the MariaDB Server.
The first step is to simply take and prepare a fresh full backup of the node that you have chosen to be the replication master. For example:
$ mariabackup --backup \ --target-dir=/var/mariadb/backup/ \ --user=mariabackup --password=mypassword
And then you would prepare the backup as you normally would. For example:
$ mariabackup --prepare \ --target-dir=/var/mariadb/backup/
Once the backup is done and prepared, you can copy it to the MariaDB Server that will be acting as slave. For example:
$ rsync -avrP /var/mariadb/backup dc2-dbserver1:/var/mariadb/backup
At this point, you can restore the backup to the datadir, as you normally would. For example:
$ mariabackup --copy-back \ --target-dir=/var/mariadb/backup/
And adjusting file permissions, if necessary:
$ chown -R mysql:mysql /var/lib/mysql/
Now that the backup has been restored to the MariaDB Server slave, you can start the MariaDB Server process.
Before the MariaDB Server slave can begin replicating from the cluster's master, you need to create a user account on the master that the slave can use to connect, and you need to grant the user account the REPLICATION SLAVE privilege. For example:
CREATE USER 'repl'@'dc2-dbserver1' IDENTIFIED BY 'password'; GRANT REPLICATION SLAVE ON *.* TO 'repl'@'dc2-dbserver1';
At this point, you need to get the replication coordinates of the master from the original backup.
The coordinates will be in the xtrabackup_binlog_info file.
Mariabackup dumps replication coordinates in two forms: GTID strings and binary log file and position coordinates, like the ones you would normally see from SHOW MASTER STATUS output. In this case, it is probably better to use the GTID coordinates.
For example:
mariadb-bin.000096 568 0-1-2
Regardless of the coordinates you use, you will have to set up the master connection using CHANGE MASTER TO and then start the replication threads with START SLAVE.
If you want to use GTIDs, then you will have to first set gtid_slave_pos to the GTID coordinates that we pulled from the xtrabackup_binlog_info file, and we would set MASTER_USE_GTID=slave_pos in the CHANGE MASTER TO command. For example:
SET GLOBAL gtid_slave_pos = "0-1-2"; CHANGE MASTER TO MASTER_HOST="c1dbserver1", MASTER_PORT=3310, MASTER_USER="repl", MASTER_PASSWORD="password", MASTER_USE_GTID=slave_pos; START SLAVE;
If you want to use the binary log file and position coordinates, then you would set MASTER_LOG_FILE and MASTER_LOG_POS in the CHANGE MASTER TO command to the file and position coordinates that we pulled the xtrabackup_binlog_info file. For example:
CHANGE MASTER TO MASTER_HOST="c1dbserver1", MASTER_PORT=3310, MASTER_USER="repl", MASTER_PASSWORD="password", MASTER_LOG_FILE='mariadb-bin.000096', MASTER_LOG_POS=568, START SLAVE;
You should be done setting up the slave now, so you should check its status with SHOW SLAVE STATUS. For example:
SHOW SLAVE STATUS\G
Now that the MariaDB Server is up, ensure that it does not start accepting writes yet if you want to set up circular replication between the cluster and the MariaDB Server.
You can also set up circular replication between the cluster and MariaDB Server, which means that the MariaDB Server replicates from the cluster, and the cluster also replicates from the MariaDB Server.
Before circular replication can begin, you also need to create a user account on the MariaDB Server, since it will be acting as replication master to the cluster's slave, and you need to grant the user account the REPLICATION SLAVE privilege. For example:
CREATE USER 'repl'@'c1dbserver1' IDENTIFIED BY 'password'; GRANT REPLICATION SLAVE ON *.* TO 'repl'@'c1dbserver1';
How this is done would depend on whether you want to use the GTID coordinates or the binary log file and position coordinates.
Regardless, you need to ensure that the second cluster is not accepting any writes other than those that it replicates from the cluster at this stage.
To get the GTID coordinates on the MariaDB Server you can check gtid_current_pos by executing:
SHOW GLOBAL VARIABLES LIKE 'gtid_current_pos';
Then on the node acting as slave in the cluster, you can set up replication by setting gtid_slave_pos to the GTID that was returned and then executing CHANGE MASTER TO:
SET GLOBAL gtid_slave_pos = "0-1-2"; CHANGE MASTER TO MASTER_HOST="c2dbserver1", MASTER_PORT=3310, MASTER_USER="repl", MASTER_PASSWORD="password", MASTER_USE_GTID=slave_pos; START SLAVE;
To get the binary log file and position coordinates on the MariaDB Server, you can execute SHOW MASTER STATUS:
SHOW MASTER STATUS
Then on the node acting as slave in the cluster, you would set master_log_file and master_log_pos in the CHANGE MASTER TO command. For example:
CHANGE MASTER TO MASTER_HOST="c2dbserver1", MASTER_PORT=3310, MASTER_USER="repl", MASTER_PASSWORD="password", MASTER_LOG_FILE='mariadb-bin.000096', MASTER_LOG_POS=568; START SLAVE;
You should be done setting up the circular replication on the node in the first cluster now, so you should check its status with SHOW SLAVE STATUS. For example:
SHOW SLAVE STATUS\G
© 2019 MariaDB
Licensed under the Creative Commons Attribution 3.0 Unported License and the GNU Free Documentation License.
https://mariadb.com/kb/en/using-mariadb-replication-with-mariadb-galera-cluster-configuring-mariadb-r/