Cluster data inconsistency may be caused if following conditions hold:
an event fires, and event execution processes rows acquiring InnoDB GAP locks
event execution has advanced to replication phase, but has not committed yet
replication applies a write set, which writes over the event's GAPs, and as a result
the event will be aborted
Because the event has already replicated, other cluster nodes, will apply the event. Only the originating node will rollback the transaction, and the data state in originating node will differ by the changes done by the event execution. When the event fires next time, and if it processes the same data set again, it may result in inconsistency aborting of all other nodes in the cluster.
Cluster data inconsistency may be caused if following conditions hold:
Because the event has already replicated, other cluster nodes, will apply the event. Only the originating node will rollback the transaction, and the data state in originating node will differ by the changes done by the event execution. When the event fires next time, and if it processes the same data set again, it may result in inconsistency aborting of all other nodes in the cluster.