MySQL 8.0 Reference Manual Including MySQL NDB Cluster 8.0

18.4.3 Distributed Recovery

Whenever a new member joins a replication group, it connects to a suitable donor and fetches the data that it has missed up until the point it is declared online. This section describes the original process through which a member joining a group catches up with the remaining servers in the group, called distributed recovery. Distributed recovery uses a standard asynchronous replication channel (named group_replication_recovery) that is established between the donor and the server joining the group. The donor's binary logs are replicated to and applied by the joining member until the point that the view change happened when the server joining the group became part of the group.

From MySQL 8.0.17, Group Replication can use a remote cloning operation as an alternative to distributed recovery to provide transactions to a joining member. If group members are set up to support cloning, Group Replication automatically selects the best strategy for state transfer after you issue START GROUP_REPLICATION on the joining member, and manages the process for you. For information on cloning, see Section 18.4.4, “Cloning for Group Replication”.