On Wed, Aug 8, 2018 at 11:33 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> Oh ... never mind that last. The parent Append will run its children
> sequentially, so that the Gathers execute one at a time, and at no
> point will more than 2 workers be active.
Yep.
> Nonetheless, it's a damfool query plan, because we'll be going through
> parallel worker startup/shutdown overhead 4 times for no benefit.
> We really should put in some kind of logic to force those sibling
> Gathers to be aggregated into one, or else disallow them altogether.
Disallowing them could be a big performance regression. Combining
them is reasonable, but it's not clear to me how you'd fit that into
the planner structure. There's no convenient RelOptInfo to associate
with the aggregated-Gather. It can't use the parent RelOptInfo unless
all of the children have a partial path, and in that case this plan
never would have been generated in the first place.
--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company