Re: Error 1112 on Innodb to ndb replication
Hi Vasilis,
Good that you have only 1 replication channel in use.
Have you checked the number of replication threads setting?
The configuration parameter is slave_parallel_workers.
Regarding the memory usage - I am surprised that the MySQLD continues to operate after failing to allocate memory - usually that is a fatal problem.
You do not state how much memory and swap you have available, but I guess that if MySQLD cannot allocate the memory it needs, according to its configuration, then it may not operate correctly, depending on which components are affected. There is not really enough log supplied for me to know at what stage this allocation failure occurred, and therefore what it might affect.
When used mostly as 'clients' to MySQL Cluster, MySQLDs generally do not need a memory-heavy configuration, so probably the configuration can be tuned to reduce the memory foorprint so that it can fit (reducing InnoDB buffer pools etc).
Finally, note that although MySQL 8.0 is Generally Available (GA), we have not yet declared that MySQL Cluster bundled in 8.0 is GA, so you are not using a fully supported version there. MySQL Cluster 7.6 (based on MySQL Server 5.7) is our most recent, generally available version. As you are using a pre-GA version, there is always the chance that you find issues.
Hopefully something of the above is useful,
Frazer