Re: Parallel leader process info in EXPLAIN - Mailing list pgsql-hackers

While working on some slides explaining EXPLAIN, I couldn't resist the
urge to add the missing $SUBJECT.  The attached 0001 patch gives the
following:

Gather  ... time=0.146..33.077 rows=1 loops=1)
  Workers Planned: 2
  Workers Launched: 2
  Buffers: shared hit=4425
  ->  Parallel Seq Scan on public.t ... time=19.421..30.092 rows=0 loops=3)
        Filter: (t.i = 42)
        Rows Removed by Filter: 333333
        Leader: actual time=0.013..32.025 rows=1 loops=1      <--- NEW
          Buffers: shared hit=1546                            <--- NEW
        Worker 0: actual time=29.069..29.069 rows=0 loops=1
          Buffers: shared hit=1126
        Worker 1: actual time=29.181..29.181 rows=0 loops=1
          Buffers: shared hit=1753

Without that, you have to deduce what work was done in the leader, but
I'd rather just show it.

The 0002 patch adjusts Sort for consistency with that scheme, so you get:

Sort  ... time=84.303..122.238 rows=333333 loops=3)
   Output: t1.i
   Sort Key: t1.i
   Leader:  Sort Method: external merge  Disk: 5864kB       <--- DIFFERENT
   Worker 0:  Sort Method: external merge  Disk: 3376kB
   Worker 1:  Sort Method: external merge  Disk: 4504kB
   Leader: actual time=119.624..165.949 rows=426914 loops=1
   Worker 0: actual time=61.239..90.984 rows=245612 loops=1
   Worker 1: actual time=72.046..109.782 rows=327474 loops=1

Without the "Leader" label, it's not really clear to the uninitiated
whether you're looking at combined, average or single process numbers.

Of course there are some more things that could be reported in a
similar way eventually, such as filter counters and hash join details.