Tomas Vondra <tomas.von...@2ndquadrant.com> writes: > On Mon, Apr 06, 2020 at 04:14:38PM -0400, Tom Lane wrote: >> Did you ever use force_parallel_mode = regress?
> Ah, not sure - probably not in this round of tests and there were some > changes in the explain code. Thanks for the hint. Locally, things pass without force_parallel_mode, but turning it on produces failures that look similar to rhinoceros's (didn't examine other BF members). At a guess, looks like missing or incorrect logic for propagating some state back from parallel workers. regards, tom lane