Heads up when doing a network import with datapump ( avoid imaginary rows )

Heads up when doing a network import…or how to avoid imaginary rows coming in….

Symptoms:

You are doing a network import of a schema which includes one or more partitioned tables and at the end of your import ….datapump can’t create one or more unique constraints on some of the tables …when the same table(s) in the source have valid unique constraints….

The other symptom you may see…is that your sequence values end up being inconsistent between your source and your target…even though your entire ‘copy’ operation was done in one operation

Cause:

This is because by default..a datapump import is not expected to be consistent…if you want to maintain consistency…you have to use the flashback_time ( or flashback_scn ) option in your import….