Re: BUG #18234: Nested Loop joint strategy is ignored for a tiny table joined with UNION ALL of two filtered parts

Hi!

To quickly address the valid point about non-self-contained bug reports, here is the export of the schema from the DB Fiddle:

/* Huge inheritance-partitioned table */
create table huge(id bigint, filter_out bool);
create table huge_partition1(id bigint, filter_out bool);
create table huge_partition2(id bigint, filter_out bool);
alter table huge_partition1 inherit huge;
alter table huge_partition2 inherit huge;

insert into huge_partition1(id, filter_out)
  select id, mod(id,7) = 0
  from generate_series(1,100000) id;

insert into huge_partition2(id, filter_out)
  select id, mod(id,7) = 0
  from generate_series(1,100000) id;

create index on huge_partition1(id);
create index on huge_partition2(id);
analyze huge_partition1;
analyze huge_partition2;

/* Medium inheritance-partitioned table (same structure, but 100x smaller) */
create table medium(id bigint, filter_out bool);
create table medium_partition1(id bigint, filter_out bool);
create table medium_partition2(id bigint, filter_out bool);
alter table medium_partition1 inherit medium;
alter table medium_partition2 inherit medium;

insert into medium_partition1(id, filter_out)
  select id, mod(id,7) = 0
  from generate_series(1,1000) id;

insert into medium_partition2(id, filter_out)
  select id, mod(id,7) = 0
  from generate_series(1,1000) id;

create index on medium_partition1(id);
create index on medium_partition2(id);
analyze medium_partition1;
analyze medium_partition2;

/* Tiny table of just 5 values */
create table tiny(id bigint);
insert into tiny(id) values (100),(200),(300),(400),(500);
analyze tiny;

/* Views that UNION ALL all non-filtered rows of HUGE and MEDIUM */
create view vw_broken as
  select id from huge where filter_out
  union all
  select id from medium where filter_out;

create view vw_not_broken as
  select id,filter_out from (
    select id,filter_out from huge
    union all
    select id,filter_out from medium
  ) q
  where filter_out;

/* This query does NOT use nested loops unexpectedly */
explain select * from tiny join vw_broken on tiny.id = vw_broken.id;

QUERY PLAN
---------------------------------------------------------------------------------------------
Hash Join  (cost=1.11..3798.30 rows=712 width=16)
  Hash Cond: (huge.id = tiny.id)
  ->  Append  (cost=0.00..3683.32 rows=28466 width=8)
        ->  Append  (cost=0.00..3222.91 rows=28181 width=8)
              ->  Seq Scan on huge huge_1  (cost=0.00..0.00 rows=1 width=8)
                    Filter: filter_out
              ->  Seq Scan on huge_partition1 huge_2  (cost=0.00..1541.00 rows=13987 width=8)
                    Filter: filter_out
              ->  Seq Scan on huge_partition2 huge_3  (cost=0.00..1541.00 rows=14193 width=8)
                    Filter: filter_out
        ->  Append  (cost=0.00..33.42 rows=285 width=8)
              ->  Seq Scan on medium medium_1  (cost=0.00..0.00 rows=1 width=8)
                    Filter: filter_out
              ->  Seq Scan on medium_partition1 medium_2  (cost=0.00..16.00 rows=142 width=8)
                    Filter: filter_out
              ->  Seq Scan on medium_partition2 medium_3  (cost=0.00..16.00 rows=142 width=8)
                    Filter: filter_out
  ->  Hash  (cost=1.05..1.05 rows=5 width=8)
        ->  Seq Scan on tiny  (cost=0.00..1.05 rows=5 width=8)


/* This query DOES use nested loops as expected */

explain select * from tiny join vw_not_broken on tiny.id = vw_not_broken.id;

QUERY PLAN
---------------------------------------------------------------------------------------------------------------------
Nested Loop  (cost=0.00..151.53 rows=712 width=17)
  ->  Seq Scan on tiny  (cost=0.00..1.05 rows=5 width=8)
  ->  Append  (cost=0.00..30.04 rows=6 width=9)
        ->  Seq Scan on huge  (cost=0.00..0.00 rows=1 width=9)
              Filter: (filter_out AND (tiny.id = id))
        ->  Index Scan using huge_partition1_id_idx on huge_partition1 huge_1  (cost=0.29..8.31 rows=1 width=9)
              Index Cond: (id = tiny.id)
              Filter: filter_out
        ->  Index Scan using huge_partition2_id_idx on huge_partition2 huge_2  (cost=0.29..8.31 rows=1 width=9)
              Index Cond: (id = tiny.id)
              Filter: filter_out
        ->  Seq Scan on medium  (cost=0.00..0.00 rows=1 width=9)
              Filter: (filter_out AND (tiny.id = id))
        ->  Index Scan using medium_partition1_id_idx on medium_partition1 medium_1  (cost=0.28..6.69 rows=1 width=9)
              Index Cond: (id = tiny.id)
              Filter: filter_out
        ->  Index Scan using medium_partition2_id_idx on medium_partition2 medium_2  (cost=0.28..6.69 rows=1 width=9)
              Index Cond: (id = tiny.id)
              Filter: filter_out

--

Best regards, Dmytro

On Thu, Dec 7, 2023 at 9:53 PM Tom Lane <tgl@sss.pgh.pa.us> wrote:

PG Bug reporting form <noreply@postgresql.org> writes:
> Summary of the issue: for a (5-row recordset) JOIN (massive partitioned
> recordset indexed by id) USING (id), the (Nested Loop over 5 values)
> strategy is completely ignored, and Hash Join or Merge Join is done instead,
> which does SeqScan over the "massive recordset".

> Reproduction in DB Fiddle:
> https://www.db-fiddle.com/f/sJUUWNgW7pqPWcJwihVoj5/1 (this demonstrates both
> the bad behaviour and a way to work around it)

We are generally not too happy with non-self-contained bug reports.
Once that DB Fiddle entry disappears, this bug report will be useless.
However ...

> 1)The massive recordset on the right side of the JOIN must come from the
> UNION ALL of two parts, both of which have a filter, like this view in my
> reproduction:

> create view vw_broken as
>   select id from huge where filter_out
>   union all
>   select id from medium where filter_out;

I suspect the WHERE clauses trigger the problem because the resulting
sub-selects can't be pulled up to become an "appendrel", per
is_safe_append_member:

     * It's only safe to pull up the child if its jointree contains exactly
     * one RTE, else the AppendRelInfo data structure breaks. The one base RTE
     * could be buried in several levels of FromExpr, however.  Also, if the
     * child's jointree is completely empty, we can pull up because
     * pull_up_simple_subquery will insert a single RTE_RESULT RTE instead.
     *
     * Also, the child can't have any WHERE quals because there's no place to
     * put them in an appendrel.  (This is a bit annoying...)

That means the sub-selects will be planned independently and there's
no chance to consider the nestloop-with-inner-indexscan plan you are
hoping for.

This is a longstanding wart, but improving matters would require some
fairly painstaking work.  The "appendrel" mechanism is core to both
traditional inheritance and partitioning; I don't recommend trying
to blow it up and start over.  I vaguely recall previous discussions
that identified some semantic issues with trying to just attach
WHERE clauses to appendrel members, but it was a long time ago and
the details escape me.

                        regards, tom lane