Cluster.rescan() did not correctly handle missing recovery account users. If it encountered a user with an unexpected format, and the correct format was missing, it did not attempt to create the correct user.

Errors similar to the following were logged by the Cluster.status() command:

WARNING: Incorrect recovery account (mysql_innodb_cluster_3337079193) being used. Use Cluster.rescan() to repair.
      

As of this release, Cluster.rescan() creates the missing user. (Bug #35828910)

Under certain circumstances, dba.rebootClusterFromCompleteOutage() failed with malformed GTID errors relating to GROUP_CONCAT. dba.rebootClusterFromCompleteOutage() must query the complete GTID set of the channel and this query failed if the default GROUP_CONCAT_MAX_LEN value was too low.

As of this release, queries which do not require the GROUP_CONCAT function, do not use it and queries which require it, use a GROUP_CONCAT_MAX_LEN value of 1GB. (Bug #35356006)

As of this release, util.loadDump() and util.copyInstance() automatically exclude the mysql_audit and mysql_firewall schemas if the target is a MySQL HeatWave Service DB System. (Bug #35830920)

util.loadDump() ignored leading zeroes (0) in S3 bucket prefix names. (Bug #36041691)

It was not possible to migrate from a compatible database using the copyInstance() or dumpInstance() utilities. Errors similar to the following error were returned:

Util.copyInstance: Unknown system variable 'activate_all_roles_on_login' (MYSQLSH 1193)

(Bug #35963431)