Hi Robert,
AFAIU the problem is that when the pof tuple migrates to pot, the
partition-level FK constraint on polf sees the update as a delete
(because at that level that's what it is) and thus cascades to delete
the tuple on polf. There is nothing that automagically instantiates a
new tuple in polt.
To fix this problem, we would need a mechanism to insert an equivalent
referencing tuple in the other partition. I'm not sure how would we
achieve such a thing.
--
Álvaro Herrera https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services