Re: Parallel INSERT (INTO ... SELECT ...)

On Mon, Oct 5, 2020 at 10:36 PM Dilip Kumar <dilipbalaut@gmail.com> wrote:
>
> > > I have one question which is common to both this patch and parallel
> > > inserts in CTAS[1], do we need to skip creating tuple
> > > queues(ExecParallelSetupTupleQueues) as we don't have any tuples
> > > that's being shared from workers to leader?
> > >
> >
> > As far as this patch is concerned we might need to return tuples when
> > there is a Returning clause. I think for the cases where we don't need
> > to return tuples we might want to skip creating these queues if it is
> > feasible without too many changes.
>

Hi Dilip,

You're right. I've included that in my latest version of the patch (so
Gather should only start tuple queues in the case of parallel SELECT
or parallel INSERT with a RETURNING clause).
Other functionality updated includes:
- Added more necessary exclusions for Parallel INSERT INTO ... SELECT
... (but allowing underlying query to still be parallel):
  - non-parallel-safe triggers
  - non-parallel-safe default and check expressions
  - foreign tables
  - temporary tables
- Added support for before/after statement-level INSERT triggers
(can't allow parallel workers to execute these)
- Adjusted cost of Gather node, for when RETURNING clause is not specified
I have not found issues with partition tables (yet) or toast column values.

Also, I have attached a separate patch (requested by Andres Freund)
that just allows the underlying SELECT part of "INSERT INTO ... SELECT
..." to be parallel.

Regards,
Greg Nancarrow
Fujitsu Australia