3.8 Switching Sources During Failover

You can tell a replica to change to a new source using the CHANGE REPLICATION SOURCE TO statement (from MySQL 8.0.23) or CHANGE MASTER TO statement (before MySQL 8.0.23). The replica does not check whether the databases on the source are compatible with those on the replica; it simply begins reading and executing events from the specified coordinates in the new source's binary log. In a failover situation, all the servers in the group are typically executing the same events from the same binary log file, so changing the source of the events should not affect the structure or integrity of the database, provided that you exercise care in making the change.

Replicas should be run with binary logging enabled (the --log-bin option), which is the default. If you are not using GTIDs for replication, then the replicas should also be run with --log-slave-updates=OFF (logging replica updates is the default). In this way, the replica is ready to become a source without restarting the replica mysqld. Assume that you have the structure shown in Figure 3.4, “Redundancy Using Replication, Initial Structure”.

Figure 3.4 Redundancy Using Replication, Initial Structure

In this diagram, the MySQL Source holds the source database, the MySQL Replica hosts are replicas, and the Web Client machines are issuing database reads and writes. Web clients that issue only reads (and would normally be connected to the replicas) are not shown, as they do not need to switch to a new server in the event of failure. For a more detailed example of a read/write scale-out replication structure, see Section 3.5, “Using Replication for Scale-Out”.

Each MySQL replica (Replica 1, Replica 2, and Replica 3) is a replica running with binary logging enabled, and with --log-slave-updates=OFF. Because updates received by a replica from the source are not logged in the binary log when --log-slave-updates=OFF is specified, the binary log on each replica is empty initially. If for some reason MySQL Source becomes unavailable, you can pick one of the replicas to become the new source. For example, if you pick Replica 1, all Web Clients should be redirected to Replica 1, which writes the updates to its binary log. Replica 2 and Replica 3 should then replicate from Replica 1.

The reason for running the replica with --log-slave-updates=OFF is to prevent replicas from receiving updates twice in case you cause one of the replicas to become the new source. If Replica 1 has --log-slave-updates enabled, which is the default, it writes any updates that it receives from Source in its own binary log. This means that, when Replica 2 changes from Source to Replica 1 as its source, it may receive updates from Replica 1 that it has already received from Source.

Make sure that all replicas have processed any statements in their relay log. On each replica, issue STOP REPLICA IO_THREAD, then check the output of SHOW PROCESSLIST until you see Has read all relay log. When this is true for all replicas, they can be reconfigured to the new setup. On the replica Replica 1 being promoted to become the source, issue STOP REPLICA and RESET MASTER.

On the other replicas Replica 2 and Replica 3, use STOP REPLICA and CHANGE REPLICATION SOURCE TO SOURCE_HOST='Replica1' or CHANGE MASTER TO MASTER_HOST='Replica1' (where 'Replica1' represents the real host name of Replica 1). To use CHANGE REPLICATION SOURCE TO | CHANGE MASTER TO, add all information about how to connect to Replica 1 from Replica 2 or Replica 3 (user, password, port). When issuing the statement in this scenario, there is no need to specify the name of the Replica 1 binary log file or log position to read from, since the first binary log file and position 4, are the defaults. Finally, execute START REPLICA on Replica 2 and Replica 3.

Once the new replication setup is in place, you need to tell each Web Client to direct its statements to Replica 1. From that point on, all update statements sent by Web Client to Replica 1 are written to the binary log of Replica 1, which then contains every update statement sent to Replica 1 since Source stopped.

The resulting server structure is shown in Figure 3.5, “Redundancy Using Replication, After Source Failure”.

Figure 3.5 Redundancy Using Replication, After Source Failure

When Source becomes available again, you should make it a replica of Replica 1. To do this, issue on Source the same CHANGE REPLICATION SOURCE TO | CHANGE MASTER TO statement as that issued on Replica 2 and Replica 3 previously. Source then becomes a replica of Replica 1 and picks up the Web Client writes that it missed while it was offline.

To make Source a source again, use the preceding procedure as if Replica 1 was unavailable and Source was to be the new source. During this procedure, do not forget to run RESET MASTER on Replica 1 before making Replica 1, Replica 2, and Replica 3 replicas of Source. If you fail to do this, the replicas may pick up stale writes from the Web Client applications dating from before the point at which Source became unavailable.

You should be aware that there is no synchronization between replicas, even when they share the same source, and thus some replicas might be considerably ahead of others. This means that in some cases the procedure outlined in the previous example might not work as expected. In practice, however, relay logs on all replicas should be relatively close together.

One way to keep applications informed about the location of the source is to have a dynamic DNS entry for the source server. With bind you can use nsupdate to update the DNS dynamically.