I have a very large query that also touches quite a bit of data. It
generates a large temporary file (actually, several, because the total
size is about 4.5GB). I'm wondering which part of the query's plan
causes these files to be created. The query has the following sequence
of plan nodes when run through EXPLAIN:
HashAggregate -> Result -> Append -> [Seq Scan | Bitmap Heap Scan, ...]
Where there are many "Seq Scan" or "Bitmap Heap Scan" nodes being
generated, one for each partition of the partitioned table.
So, why are these files being created?
Also, I suppose this data on-disk can be kept in memory instead by increasing
work_mem to a suitable value?
Thanks,
-Ryan Kelly