MySQL Cluster Manager 8.0 Release Notes  /  Changes in MySQL Cluster Manager 8.0.32 (2023-01-17, General Availability)

Changes in MySQL Cluster Manager 8.0.32 (2023-01-17, General Availability)

Functionality Added or Changed

  • Agent: A new mcmd option, --core-file, has been added. If its value is 1 (default), when mcmd quits unexpectedly, it writes a core file to the location where it was started or to another core dump location as specified by the operating system. (Bug #34881109)

  • Agent: A set command can now be used to perform a reset on a configuration parameter, which allows set and reset operations to be run by a single set command to avoid multiple rolling restarts. See Performing a reset with a set command for details. (Bug #23590684)

  • Agent: A new --restart option has been added to the set command for overriding the default restart type of the parameters MaxNoOfExecutionThreads and ThreadConfig for data nodes. See the option's description for details. Please note that the option should only be used under guidance by support personnel from Oracle.

Bugs Fixed

  • Agent: mcmd allowed itself to be started by the root user of the operating system, which should not happen. With this fix, mcmd throws an error in the situation. (Bug #34839125)

  • Agent: If the mcmd configuration file had multiple sections whose names started with mcmd (which made the configuration invalid), mcmd proceeded with starting, but then quit unexpectedly. With this fix, a proper error is thrown in the situation. (Bug #34759709)

  • Agent: When running the list backups command, mcmd printed its whole binary message into the log file if the log severity level was set at debug. (Bug #34657441)

  • Agent: When a cluster is running with EncryptedFileSystem= 1, the mcmd log did not show the exact command by which the data node was started, but had the ndbd or ndbmtd command options shuffled. With this fix, the command is shown with its options properly sorted, and the command is also printed to the file mcm_data/clusters/<clustername>/pid/cmd.<nodeid> for easy access. (Bug #34643817)

  • Agent: When a connection using the classic protocol to mysqld was closed before handshake was finished, mcmd quit unexpectedly. With this fix, a proper error is thrown in the situation. (Bug #34385879)

  • Client: A set filesystem-password-file command run on the mcm client was not kept in the client's history. It is now made available in the history for easy access. (Bug #34709187)