Re: Explain [Analyze] produces parallel scan for select Into table statements.

Поиск
Список
Период
Сортировка
От Mithun Cy
Тема Re: Explain [Analyze] produces parallel scan for select Into table statements.
Дата
Msg-id CAD__OuiFZWwohrfceu6SEdZ3Y4M7UbLFsT_KhFLJc5DARDNVnQ@mail.gmail.com
обсуждение исходный текст
Ответ на Re: Explain [Analyze] produces parallel scan for select Into table statements.  (Amit Kapila <amit.kapila16@gmail.com>)
Ответы Re: Explain [Analyze] produces parallel scan for select Into table statements.  (Amit Kapila <amit.kapila16@gmail.com>)
Список pgsql-hackers


On Sat, Mar 12, 2016 at 12:28 PM, Amit Kapila <amit.kapila16@gmail.com> wrote
>I don't see how this test will fail with force_parallel_mode=regress and max_parallel_degree > 0 even without the patch proposed to fix the issue in >hand.  In short, I don't think this test would have caught the issue, so I don't see much advantage in adding such a test.  Even if we want to add such a >test case, I think as proposed this will substantially increase the timing for "Select Into" test which might not be an acceptable test case addition >especially for testing one corner case.


Without above patch the make installcheck fails for select_into.sql with below diff

when
force_parallel_mode = on
max_parallel_degree = 3

diff results/select_into.out expected/select_into.out

104,110c104,107

< QUERY PLAN

< ------------------------

< Gather

< Number of Workers: 1

< Single Copy: true

< -> Seq Scan on mt1

< (4 rows)

---

> QUERY PLAN

> -----------------

> Seq Scan on mt1

> (1 row)


Again with postgresql.conf non default settings. 

force_parallel_mode = on
max_parallel_degree = 3
parallel_tuple_cost = 0

[mithun@localhost regress]$ diff results/select_into.out expected/select_into.out

104,109c104,107

< QUERY PLAN

< --------------------------------

< Gather

< Number of Workers: 3

< -> Parallel Seq Scan on mt1

< (3 rows)

---

> QUERY PLAN

> -----------------

> Seq Scan on mt1

> (1 row)

To reduce the time of execution I can set the generate_series parameter to 500, which is fast in my machine and also fails with above diff but this time only one worker is assigned as per plan.

--
Thanks and Regards
Mithun C Y

В списке pgsql-hackers по дате отправления:

Предыдущее
От: Konstantin Knizhnik
Дата:
Сообщение: Re: eXtensible Transaction Manager API (v2)
Следующее
От: Michael Paquier
Дата:
Сообщение: Re: [PATCH] Use MemoryContextAlloc() in the MemoryContextAllocZero() and MemoryContextAllocZeroAligned()