Re: [HACKERS] Parallel bitmap heap scan

Поиск
Список
Период
Сортировка
От Dilip Kumar
Тема Re: [HACKERS] Parallel bitmap heap scan
Дата
Msg-id CAFiTN-s9O-7tZ2zgm089Gh5UsPhL=HFBrNgnLVh=wtBKC9j-vQ@mail.gmail.com
обсуждение исходный текст
Ответ на Re: [HACKERS] Parallel bitmap heap scan  (tushar <tushar.ahuja@enterprisedb.com>)
Ответы Re: [HACKERS] Parallel bitmap heap scan  (tushar <tushar.ahuja@enterprisedb.com>)
Список pgsql-hackers
On Tue, Jan 10, 2017 at 2:19 PM, tushar <tushar.ahuja@enterprisedb.com> wrote:
> We found a regression , earlier the testcase was working fine (against the
> older patches of Parallel bitmap heap scan)  but now getting a server crash
> against v8 patches.
>
> Testcase - (one of the table of TPC-H )
>
> postgres=#explain analyze verbose
> SELECT SUM(l_extendedprice) FROM lineitem
> WHERE (l_shipdate >= '1995-01-01'::date)
> AND (l_shipdate <='1996-03-31'::date);

While fixing some of the review comments in v7 and v8, I had allocated
new memory for pagetable, and missed to initialize it.

Thanks for reporting, I have fixed it in v9. After fix query is
running fine for me. Please try attached patch and confirm from your
side.

postgres=# explain analyze verbose
SELECT SUM(l_extendedprice) FROM lineitem
WHERE (l_shipdate >= '1995-01-01'::date)
AND (l_shipdate <='1996-03-31'::date);

        QUERY PLAN


--------------------------------------------------------------------------------------------------------------------------------------------------------------
-----------
 Finalize Aggregate  (cost=798002.46..798002.47 rows=1 width=32)
(actual time=15501.245..15501.245 rows=1 loops=1)
   Output: sum(l_extendedprice)
   ->  Gather  (cost=798002.24..798002.45 rows=2 width=32) (actual
time=15494.358..15498.919 rows=3 loops=1)
         Output: (PARTIAL sum(l_extendedprice))
         Workers Planned: 2
         Workers Launched: 2
         ->  Partial Aggregate  (cost=797002.24..797002.25 rows=1
width=32) (actual time=15492.937..15492.937 rows=1 loops=3)
               Output: PARTIAL sum(l_extendedprice)
               Worker 0: actual time=15491.218..15491.219 rows=1 loops=1
               Worker 1: actual time=15493.514..15493.514 rows=1 loops=1
               ->  Parallel Bitmap Heap Scan on public.lineitem
(cost=147461.75..791014.31 rows=2395170 width=8) (actual
time=8553.301..15061.333 rows=189294
7 loops=3)
                     Output: l_extendedprice
                     Recheck Cond: ((lineitem.l_shipdate >=
'1995-01-01'::date) AND (lineitem.l_shipdate <= '1996-03-31'::date))
                     Rows Removed by Index Recheck: 6451177
                     Heap Blocks: exact=27963 lossy=164938
                     Worker 0: actual time=8548.957..15054.511
rows=1887239 loops=1
                     Worker 1: actual time=8554.817..15050.317
rows=1902477 loops=1
                     ->  Bitmap Index Scan on idx_lineitem_shipdate
(cost=0.00..146024.65 rows=5748409 width=0) (actual
time=8533.701..8533.701 rows=5678841
loops=1)
                           Index Cond: ((lineitem.l_shipdate >=
'1995-01-01'::date) AND (lineitem.l_shipdate <= '1996-03-31'::date))
 Planning time: 2.742 ms
 Execution time: 15509.696 ms
(21 rows)



-- 
Regards,
Dilip Kumar
EnterpriseDB: http://www.enterprisedb.com

-- 
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers

Вложения

В списке pgsql-hackers по дате отправления:

Предыдущее
От: Kyotaro HORIGUCHI
Дата:
Сообщение: Re: [HACKERS] Radix tree for character conversion
Следующее
От: Ashutosh Bapat
Дата:
Сообщение: Re: [HACKERS] postgres_fdw bug in 9.6