16.4.3 Upgrading a Replication Setup

When you upgrade servers that participate in a replication setup, the procedure for upgrading depends on the current server versions and the version to which you are upgrading. This section provides information about how upgrading affects replication. For general information about upgrading MySQL, see Section 2.11, “Upgrading MySQL”

When you upgrade a source to 5.7 from an earlier MySQL release series, you should first ensure that all the replicas of this source are using the same 5.7.x release. If this is not the case, you should first upgrade the replicas. To upgrade each replica, shut it down, upgrade it to the appropriate 5.7.x version, restart it, and restart replication. Relay logs created by the replica after the upgrade are in 5.7 format.

Changes affecting operations in strict SQL mode (STRICT_TRANS_TABLES or STRICT_ALL_TABLES) may result in replication failure on an updated replica. For example, as of MySQL 5.7.2, the server restricts insertion of a DEFAULT value of 0 for temporal data types in strict mode. If you use statement-based logging (binlog_format=STATEMENT), if a replica is upgraded before the source, the nonupgraded source will execute statements without error that may fail on the replica and replication will stop. To deal with this, stop all new statements on the source and wait until the replicas catch up. Then upgrade the replicas. Alternatively, if you cannot stop new statements, temporarily change to row-based logging on the source (binlog_format=ROW) and wait until all replicas have processed all binary logs produced up to the point of this change. Then upgrade the replicas.

After the replicas have been upgraded, shut down the source, upgrade it to the same 5.7.x release as the replicas, and restart it. If you had temporarily changed the source to row-based logging, change it back to statement-based logging. The 5.7 source is able to read the old binary logs written prior to the upgrade and to send them to the 5.7 replicas. The replicas recognize the old format and handle it properly. Binary logs created by the source subsequent to the upgrade are in 5.7 format. These too are recognized by the 5.7 replicas.

In other words, when upgrading to MySQL 5.7, the replicas must be MySQL 5.7 before you can upgrade the source to 5.7. Note that downgrading from 5.7 to older versions does not work so simply: You must ensure that any 5.7 binary log or relay log has been fully processed, so that you can remove it before proceeding with the downgrade.

Some upgrades may require that you drop and re-create database objects when you move from one MySQL series to the next. For example, collation changes might require that table indexes be rebuilt. Such operations, if necessary, are detailed at Section 2.11.3, “Changes in MySQL 5.7”. It is safest to perform these operations separately on the replicas and the source, and to disable replication of these operations from the source to the replica. To achieve this, use the following procedure:

  1. Stop all the replicas and upgrade them. Restart them with the --skip-slave-start option so that they do not connect to the source. Perform any table repair or rebuilding operations needed to re-create database objects, such as use of REPAIR TABLE or ALTER TABLE, or dumping and reloading tables or triggers.

  2. Disable the binary log on the source. To do this without restarting the source, execute a SET sql_log_bin = OFF statement. Alternatively, stop the source and restart it without the --log-bin option. If you restart the source, you might also want to disallow client connections. For example, if all clients connect using TCP/IP, enable the skip_networking system variable when you restart the source.

  3. With the binary log disabled, perform any table repair or rebuilding operations needed to re-create database objects. The binary log must be disabled during this step to prevent these operations from being logged and sent to the replicas later.

  4. Re-enable the binary log on the source. If you set sql_log_bin to OFF earlier, execute a SET sql_log_bin = ON statement. If you restarted the source to disable the binary log, restart it with --log-bin, and without enabling the skip_networking system variable so that clients and replicas can connect.

  5. Restart the replicas, this time without the --skip-slave-start option.

If you are upgrading an existing replication setup from a version of MySQL that does not support global transaction identifiers to a version that does, you should not enable GTIDs on either the source or the replica before making sure that the setup meets all the requirements for GTID-based replication. For example server_uuid, which was added in MySQL 5.6, must exist for GTIDs to function correctly. See Section 16.1.3.4, “Setting Up Replication Using GTIDs”, which contains information about converting existing replication setups to use GTID-based replication.

When the server is running with global transaction identifiers (GTIDs) enabled (gtid_mode=ON), do not enable binary logging by mysql_upgrade.

It is not recommended to load a dump file when GTIDs are enabled on the server (gtid_mode=ON), if your dump file includes system tables. mysqldump issues DML instructions for the system tables which use the non-transactional MyISAM storage engine, and this combination is not permitted when GTIDs are enabled. Also be aware that loading a dump file from a server with GTIDs enabled, into another server with GTIDs enabled, causes different transaction identifiers to be generated.