you are right. What it happens is that the batch importing process drops the schema and recreates it. I would like some solution that is compatible with that.
I am sure partitioned tables will have an impact but on the other hand, it will solve the problem I have now without touching the legacy code.
On Thu, 18 Apr 2019 at 10:56, Alvaro Herrera <alvherre@2ndquadrant.com> wrote: > > Note that unless you regularly query for only-manually-inserted or > only-automatically-inserted data, this will be useless and will make > queries more expensive, with no upside.
Going by "Since those rows are inserted by hand, they will be lost when the table will be reimported.", I just imagined there must be some sort of TRUNCATE + INSERT/COPY operation going on, in which case there would be a good upside of not having to perform a DELETE of all automatic records instead of the TRUNCATE.