InnoDB Cluster uses Group Replication and therefore your server instances must meet the same requirements. See Group Replication Requirements. AdminAPI provides the dba.checkInstanceConfiguration() method to verify that an instance meets the Group Replication requirements, and the dba.configureInstance() method to configure an instance to meet the requirements.

Group Replication members can contain tables using a storage engine other than InnoDB, for example MyISAM. Such tables cannot be written to by Group Replication, and therefore when using InnoDB Cluster. To be able to write to such tables with InnoDB Cluster, convert all such tables to InnoDB before using the instance in an InnoDB Cluster.

The Performance Schema must be enabled on any instance which you want to use with InnoDB Cluster.

The provisioning scripts that MySQL Shell uses to configure servers for use in InnoDB Cluster require access to Python. On Windows MySQL Shell includes Python and no user configuration is required. On Unix Python must be found as part of the shell environment. To check that your system has Python configured correctly issue:

$ /usr/bin/env python

If a Python interpreter starts, no further action is required. If the previous command fails, create a soft link between /usr/bin/python and your chosen Python binary. For more information, see Supported Languages.

From version 8.0.17, instances must use a unique server_id within an InnoDB Cluster. When you use the Cluster.addInstance(instance) operation, if the server_id of instance is already used by an instance in the cluster then the operation fails with an error.

From version 8.0.23, instances should be configured to use the parallel replication applier. See Configuring the Parallel Replication Applier.

During the process of configuring an instance for InnoDB Cluster, the majority of the system variables required for using an instance are configured. But AdminAPI does not configure the transaction_isolation system variable, which means that it defaults to REPEATABLE READ. This does not impact a single-primary cluster, but if you are using a multi-primary cluster then unless you rely on REPEATABLE READ semantics in your applications, we recommend using the READ COMMITTED isolation level. See Group Replication Limitations.