Re: Declarative partitioning and automatically generated row-IDs using BIGSERIAL

You should definitely check. Only 10% fill on each block is a bit crazy and has a ton of wasted space built into your table. If it is set low like that, I'd check the fillfactor on the indexes for that table as well.

select
       t.reloptions
from pg_class t
  join pg_namespace n on n.oid = t.relnamespace
where t.relname = 'clt_rec'
  and n.nspname = 'public';

 

Sequences are incremented and return the new value whenever they are called. If all partitions inherit the same sequence (behavior determined by which options you use when you create partitions LIKE parent), then they will all use the same series of values that cannot provide duplicates (until wrap around if allowed). If you want to ensure you don't get duplicates, then either do a proper IDENTITY which is generated as always, or ensure ID is a primary key/unique index. Since the partition key must include the ID column and partitions cannot overlap in the partition key space, each index on each partition would be part of a set.