On Fri, Oct 9, 2020 at 6:26 PM Amit Kapila <amit.kapil...@gmail.com> wrote: > > On Fri, Oct 9, 2020 at 3:51 PM Greg Nancarrow <gregn4...@gmail.com> wrote: > > > > On Fri, Oct 9, 2020 at 8:09 PM Amit Kapila <amit.kapil...@gmail.com> wrote: > > > > OK, I will update the comments for this. > > Basically, up to now, the "force_parallel_mode" has only ever operated > > on a SELECT. > > But since we are now allowing CMD_INSERT to be assessed for parallel > > mode too, we need to prevent the force_parallel_mode logic from > > sticking a Gather node over the top of arbitrary INSERTs and causing > > them to be run in parallel. Not all INSERTs are suitable for parallel > > operation, and also there are further considerations for > > parallel-safety for INSERTs compared to SELECT. INSERTs can also > > trigger UPDATEs. > > > > Sure but in that case 'top_plan->parallel_safe' should be false and it > should stick Gather node atop Insert node. >
/should/should not. -- With Regards, Amit Kapila.
