Re: BUG #11518: full_page_writes is off after executing pg_start_backup()

From: Michael Paquier <michael(dot)paquier(at)gmail(dot)com>
To: sudalait2(at)gmail(dot)com
Cc: pgsql-bugs(at)postgresql(dot)org
Subject: Re: BUG #11518: full_page_writes is off after executing pg_start_backup()
Date: 2014-09-30 05:15:30
Message-ID: CAB7nPqRCgGAr6qLg0DaYjONSVyfGXfhswBCjx2KQM2wENi_2Tg@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-bugs

On Tue, Sep 30, 2014 at 1:18 AM, <sudalait2(at)gmail(dot)com> wrote:

> The documentation says, full_page_writes is forcibly ON between
> pg_start_backup() and pg_stop_backup().
> I started the server with full_page_writes=off. And i executed
> pg_start_up(), the full_page_writes is not changed. I executed "show
> full_page_writes" it shows OFF.
> But it should be On.
>
There are two things controlling full_page_writes in the backend:
- The GUC parameter, which is the output result of what you can see with
pg_settings or SHOW
- A flag in the structure called XLogCtl, defined at the top of xlog.c in
shared memory that is used to control the state of XLOGs, and defined
depending on the value of the GUC full_page_writes or the state of system
like a server in backup state. It is actually this shared memory flag that
is updated when pg_start_backup is used. Only the flag in shared memory is
enforced to true when pg_start_backup is kicked, meaning that the value
update is actually not visible for the user, but for the system it is in
shared memory.
This is not a bug.
--
Michael

In response to

Browse pgsql-bugs by date

  From Date Subject
Next Message Michael Paquier 2014-09-30 05:34:39 Re: BUG #11520: Cannot uncheck to set false
Previous Message David Gould 2014-09-30 00:44:11 Re: BUG #11264: Auto vacuum wraparound job blocking everything