Re: Checking for stale reads on hot standby

Le 27/09/2010 02:20, Fujii Masao a écrit :
> On Mon, Sep 27, 2010 at 9:09 AM, Gurjeet Singh <singh.gurjeet@gmail.com> wrote:
>> See the nuggets hidden in section 25.2.5.2. "Monitoring" at
>> http://www.postgresql.org/docs/9.0/static/warm-standby.html#STREAMING-REPLICATION
>>
>> After an UPDATE, your application can cache the info from
>> 'pg_current_xlog_location()' result on the primary and then compare that
>> with the result of  'pg_last_xlog_receive_location()' on the standby to see
>> if it is seeing fresh enough data.
>
> Yep, but since recovery might fall behind WAL receiving,
> pg_last_xlog_replay_location should be called instead of
> pg_last_xlog_receive_location.
>

pgPool-II can do that automatically for you in load balancing mode, and
not use a standby node if it lags too much.


--
Guillaume
 http://www.postgresql.fr
 http://dalibo.com