Data node logging is improved with the following:

(Bug #38414245)

MySQL NDB Cluster now generates log entries for redo logging issues, including redo log buffer exhaustion, redo log space exhaustion, and file change problems, which can cause transactions to be aborted, queued, or delayed. A secondary overload control mechanism monitors the rate at which the redo log part's Redo buffer is flushed to disk and estimates the time required to complete writing all data in the part's Redo buffer. (Bug #37903091)

LogLevelBackup and LogLevelSchema Data node options were added in this release. (Bug #28004136)

If a data node restarted during a backup, the remaining data nodes could hang and the backup process stopped. The backup was marked as failed, and other nodes did not terminate their backup process, preventing further backups from being initiated. (Bug #38316252)

It was not possible to restore backups with BackupID values greater than 2147483647, with the ndb_restore tool. Errors were returned similar to the following:

        ndb_restore: [Warning] option 'backupid': signed value
        3000000000 adjusted to 2147483647. Failed to find backup
        2147483647.
      

(Bug #38260769)

When running CREATE or INSERT statements on a NDB Binlog server with sql_log_bin = 0, replication would fail due to a table not existing. (Bug #37953883)

Setting log level options for data nodes in MySQL NDB Cluster, such as LogLevelCheckpoint, LogLevelStatistic, LogLevelInfo, LogLevelCongestion, LogLevelError, and LogLevelConnection to a higher value does not provide additional log information beyond their default values. Errors were returned similar to the following:

        Ndb kernel thread 0 is stuck in: Job Handling elapsed=100,
        Watchdog: Warning overslept
      

(Bug #17847063)