MySQL Cluster Manager 1.4 Release Notes

5 Changes in MySQL Cluster Manager 1.4.6 (2018-05-09)

Functionality Added or Changed

  • Agent: MySQL Cluster Manager now supports taking backups in parallel fashion on individual data nodes using multiple local data managers (LDMs) and restoring such backups. Parallel backups are performed with the backup cluster command as long as the data nodes satisfy the requirements specified in Taking an NDB Backup with Parallel Data Nodes, and the backups are restored with the restore cluster as usual. (Bug #27676066)

  • Agent: The list backups command has been extended with a new --agent option for listing agent backups created with the backup agents command. See the option description for details. (Bug #27168819, Bug #27850905)

Bugs Fixed

  • Agent: The collect logs command failed to include logs that had been rotated with the rotate log command. (Bug #27918231)

  • Agent: A list backups command failed with an Error: 1159 (timeout reading communication packets) from one of the mcmd agents when it had a huge number of backups to list. With this fix, mcmd has been optimized to enumerate a larger number of backups. (Bug #27868499)

  • Agent: The mcmd agent quit unexpectedly if the cluster’s global configuration file (config.ini usually) contained a [system] section. It was because in MySQL Cluster Manager, the handling for the [system] section was missing, which has now been added by this fix. The addition also allows mcmd to set the name attribute in the [system] section, which is required for MySQL Enterprise Monitor 4.0 to monitor the cluster. (Bug #27638138, Bug #27519205)

  • Agent: The collect logs command hung if a host was referenced earlier with the localhost IP address 127.0.0.1 in the create site or create cluster command. (Bug #27551932)

  • Agent: The create cluster command was successful even if the format of an IP address submitted with it as a hostname was inconsistent with the format used earlier with the add hosts command (for example, a host was added to a site as 127.0.0.1, and then added to a cluster as 127.0.01). A subsequent attempt to start the cluster failed. This fix added consistency check for the host names, so that inconsistent IP addresses are now rejected. (Bug #27551776)

  • Agent: When a maintenance restart occurred for a cluster node while mcmd was shutting down, the node remained in the starting stage indefinitely. With this fix, a node restart will not be attempted when mcmd is shutting down. (Bug #27513481)

  • Agent: An Internal error occurred when mcmd tried to set the MySQL Server system variable audit_log_read_buffer_size with a set command. It was because of the wrong data type used internally for the variable, which has now been fixed. (Bug #27413635)

  • Agent: mcmd quit unexpectedly sometimes when running a restore cluster command. It was due to a race condition in which mcmd started executing the restore before the required ndbapi node was ready for the task. This fix added checks to eliminate the race condition. (Bug #27342271)