A few of the minor problems I had using TopLink JPA v2 build 03 with PostgreSQL:
# TopLink doesn't use sequences for id generation by default on Postgres. This means we have to hard code sequence names which reduces portability.
# We cannot specify a schema to use for entities because 1) TopLink will throw an Exception when run on a database that doesn't support schemas (not portable); and 2) The sequence generator is broken for postgres when using a non-default schema (currval() call fails).
# TopLink ignores the @Transient annotation and xml metadata.
# Fields specified as @Lob are created as VARCHAR(255) instead of TEXT.
# TopLink creates TIMESTAMP fields instead of TIMESTAMP WITH TIMEZONE.
# TopLink writes foreign-key restrictions to the schema generation script but these don't actually appear in the database when using create-tables.
we're using PostgresSQL in Production too. Recently we had the same problem you described in your posting. If we use the sequence generator, there is a exception that the
currval() call fails. Perhaps you could explain me in detail how I can avoid this trouble.