Question

当我运行持久查询[1]以根据子查询的结果更新几何列时，出现以下错误“无效的内存分配请求大小1073741824”。

有人知道如何解决此错误吗？还是如何修改更新查询以生成几何的更好方法？

[1]

UPDATE hp 
SET the_geom = lines.the_geom 
FROM (
    SELECT hp_id, st_makeline(the_geom) the_geom
    FROM ( 
        SELECT hp_id, st_collect(the_geom) the_geom
        FROM (
            SELECT sbp.hp_id, point_sequence, sobr.the_geom
            FROM sbp 
            JOIN sobr ON sbp.bp_id=sobr.id
            ORDER BY hp_id, point_sequence
        ) t
        GROUP BY hp_id
    ) tt
    GROUP BY hp_id) lines 
WHERE lines.hp_id = hp.id AND hp.the_geom IS NULL;

这是查询计划：

Update on hp  (cost=133655367.27..139530992.38 rows=1 width=268)
  ->  Nested Loop  (cost=133655367.27..139530992.38 rows=1 width=268)
        ->  Subquery Scan on lines  (cost=133655366.69..139529271.88 rows=200 width=104)
              ->  GroupAggregate  (cost=133655366.69..139529269.88 rows=200 width=40)
                    Group Key: sbp.hp_id
                    ->  GroupAggregate  (cost=133655366.69..139510248.44 rows=1267929 width=40)
                          Group Key: sbp.hp_id
                          ->  Sort  (cost=133655366.69..134489514.21 rows=333659008 width=48)
                                Sort Key: sbp.hp_id, sbp.poradove_cislo_bodu
                                ->  Hash Join  (cost=7282892.98..76155670.90 rows=333659008 width=48)
                                      Hash Cond: (sbp.bp_id = sobr.id)
                                      ->  Seq Scan on sbp  (cost=0.00..8104984.08 rows=333659008 width=24)
                                      ->  Hash  (cost=4761304.88..4761304.88 rows=124139688 width=40)
                                            ->  Seq Scan on sobr  (cost=0.00..4761304.88 rows=124139688 width=40)
        ->  Index Scan using hp_id_key on hp  (cost=0.57..8.59 rows=1 width=172)
              Index Cond: (id = lines.hp_id)
              Filter: (the_geom IS NULL)

感谢所有建议！

Answer 1

我认为您正在采取不必要的步骤，我会尝试：

SET the_geom = lines.the_geom 
FROM (
    SELECT sbp.hp_id,
    st_makeline(sobr.the_geom ORDER BY hp_id, point_sequence) as the_geom
    FROM sbp 
    JOIN sobr 
    ON sbp.bp_id=sobr.id
    GROUP BY hp_id
    ) lines
WHERE lines.hp_id = hp.id AND hp.the_geom IS NULL;

但是如果没有样本数据，我不能百分百确定它会起作用

Postgres无效的内存分配请求大小1073741824

1 个答案: