are foreign keys a big source of multixact IDs so not recommended on tables with a lot of data and a lot of churn?
I am curious to hear other answers or if anything has changed. I experienced this problem a couple of times on PG 11. In each situation my setup looked something like this:
create table small(id int primary key, data text);
create table large(id bigint primary key, small_id int references small(id));
Table large is hundreds of GB or more and accepting heavy writes in batches of 1K records, and every batch contains exactly one reference to each row in small (every batch references the same 1K rows in small).
The only way I was able to get around problems with multixact member ID wraparound was dropping the foreign key constraint.