Under normal circumstances, there should not be any issues when re-enabling referential integrity constraints on any parent/child table on the Target database for tables in replication so long as the constraint is enforced on Source.
The problem occurs due to following circumstances (and there could be many more, which are not explored here that can lead to similar errors).
The SharePlex processes (Export on Source or Post on Target) are stopped whereby not all data from Source has reached the Target.
The Source is active whereby the SharePlex queues are in a state of flux.
So it is very difficult to time the enabling of constraints that would ensure all parent and child rows have posted successfully to allow the enabling without any error.
One can attempt to re-enable them multiple times as needed (for a table that fails to enable it).
If the Source is quiet and the queues are fully drained, the constraints should re-enable fine, unless there are data integrity issues on Source, a condition rarely observed.
