Why does Postgresql use a Seq Scan on the first column of the PK?

Question

This is my table:

CREATE TABLE IF NOT EXISTS public.ob_samples_sc5555 (
    stamp timestamp(6) with time zone NOT NULL,
    oblvl smallint NOT NULL,
    olots integer NOT NULL,
    CONSTRAINT prk_ob_samples_sc5555 PRIMARY KEY (stamp, oblvl)
);

This is my query:

delete from ob_samples_sc5555
where stamp <= to_timestamp('2023-12-29 15:00:00', 'YYYY-MM-DD HH24:MI:SS');

Explain:

Finalize Aggregate (cost=10106164.53..10106164.54 rows=1 width=8)

->  Gather  (cost=10106163.59..10106164.50 rows=9 width=8)
      Workers Planned: 9
      ->  Partial Aggregate  (cost=10105163.59..10105163.60 rows=1 width=8)
            ->  Parallel Seq Scan on ob_samples_sc5555  (cost=0.00..9917330.49 rows=75133240 width=0)
                  Filter: (stamp <= to_timestamp('2023-12-29 15:00:00'::text, 'YYYY-MM-DD HH24:MI:SS'::text))

JIT: Functions: 6 Options: Inlining true, Optimization true, Expressions true, Deforming true

Why does PG13 use a seq scan instead of primary key, while the column stamp is the first column of the PK?

Why does Postgresql use a Seq Scan on the first column of the PK?

0 Answers0