Zhang Mingli
www.hashdata.xyz
On Oct 10, 2024 at 22:41 +0800, Tomas Vondra <tomas@vondra.me>, wrote:
You can try setting parallel_setup_cost and parallel_tuple_cost to 0,
and reducing min_parallel_{table,index}_scan_size. That should force a
parallel plan, and you'll see if parallelism is helpful.
In addition to this, it seems you’r eager a parallel-aware HashJoin.
Check if enable_parallel_hash = on, an explain(verbose) will show related GUCs.