Re: admin control over cancelling autovacuum when blocked by a lock

Поиск
Список
Период
Сортировка
От Simon Riggs
Тема Re: admin control over cancelling autovacuum when blocked by a lock
Дата
Msg-id CANP8+j+yBChxtUPriLttV=orwdF7RWtpP+zjF_fCB2yscuJz=g@mail.gmail.com
обсуждение исходный текст
Ответ на Re: admin control over cancelling autovacuum when blocked by a lock  (Tom Lane <tgl@sss.pgh.pa.us>)
Ответы Re: admin control over cancelling autovacuum when blocked by a lock  (John Lumby <johnlumby@hotmail.com>)
Список pgsql-admin
On Tue, 30 Apr 2019 at 02:44, Tom Lane <tgl@sss.pgh.pa.us> wrote:
John Lumby <johnlumby@hotmail.com> writes:
> On 04/29/2019 07:15 PM, Simon Riggs wrote:
>> Why would you want this?

> Because it greatly reduces rate of growth of certain indexes in some
> workloads
> (based on test in which I modified the code to change the cancel to a
> plain WARNING)

[ raised eyebrow... ]  Shouldn't do so, I think.  What exactly are you
doing that's causing the cancel?

> I *think* when an autovacuum worker is cancelled,  all or most of the work
> it did on its last table is undone  --  true?

...
 
If you're losing index cleanup
work, it suggests that you're repeatedly executing DDL that requires
exclusive lock on the table.

Exactly the point of my question.

But further than this, if we changed things as requested the DDL being executed would be forced to wait behind the autovacuum, frustrating the timely execution of the DDL.

Hence: Why would you want this?
 
Maybe you could avoid doing that?

Seems like the only way out of this dilemma. 

--
Simon Riggs                http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services

В списке pgsql-admin по дате отправления:

Предыдущее
От: Tom Lane
Дата:
Сообщение: Re: admin control over cancelling autovacuum when blocked by a lock
Следующее
От: John Lumby
Дата:
Сообщение: Re: admin control over cancelling autovacuum when blocked by a lock