Postgres Pro
Downloads
Home   >   mailing lists

Thread: Turning off hot_standby_feedback

Turning off hot_standby_feedback

From
Simon Riggs
Date:
There is a bug in hot_standby_feedback that causes the xmin of the
walsender process to not be reset when hot_standby_feedback is turned
off after it had previously sent at least one feedback message.

Clearly, that is a bad thing and deserves backpatching to 9.1, unless
I hear otherwise real soon.

Attached patch sends an immediate feedback message if it is turned off
on the standby. No protocol changes, just changes to how message is
handled and when it is sent.

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

Re: Turning off hot_standby_feedback

From
Robert Haas
Date:
On Sun, Feb 3, 2013 at 5:25 PM, Simon Riggs <simon@2ndquadrant.com> wrote:
> There is a bug in hot_standby_feedback that causes the xmin of the
> walsender process to not be reset when hot_standby_feedback is turned
> off after it had previously sent at least one feedback message.
>
> Clearly, that is a bad thing and deserves backpatching to 9.1, unless
> I hear otherwise real soon.

+1.

-- 
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company

Re: Turning off hot_standby_feedback

From
Simon Riggs
Date:
On 4 February 2013 15:52, Robert Haas <robertmhaas@gmail.com> wrote:
> On Sun, Feb 3, 2013 at 5:25 PM, Simon Riggs <simon@2ndquadrant.com> wrote:
>> There is a bug in hot_standby_feedback that causes the xmin of the
>> walsender process to not be reset when hot_standby_feedback is turned
>> off after it had previously sent at least one feedback message.
>>
>> Clearly, that is a bad thing and deserves backpatching to 9.1, unless
>> I hear otherwise real soon.
>
> +1.

Patches for 9.1 and 9.2 different to the one for HEAD.

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