[pgpool-general: 5668] How to extend load balancing in pgpool II

Dinesh Kumar vdineshmenon at gmail.com
Tue Aug 1 22:07:51 JST 2017


Hello,

I would like to know whether it is possible to configure load balancing
module in pgpool II so that it will be possible to redirect requests on the
basis of tables (in the sql query) as well. For e.g., configurations like
database_redirect_preference_list and app_name_redirect_preference_list
gives us the granularity to redirect on the basis of database / clients.
I'm looking for something like

table_name_redirect_preference_list =
'table1:primary;table2:1,table3:standby'.


I couldn't find any such configurations - Is it possible to write some
extensions (custom code to do the stuff mentioned above ) and attach it to
pgpool II or will it be required to branch and add that piece of code to
pgpool source and do a build / install ?

Reason : - I have set up a postgres cluster with SR (Streaming
Replication). Due to inherent design, things are asynchronous and there
will be a read miss if a recently inserted row in master was read from the
slave. I went through the configuration sr_check_period and
delay_threshold. What will happen to requests (insert row; read same row;)
which was done under 1 second (lowest possible value for sr_check_period).
I do not want to opt for sync replication due to performance reasons and
also due to the fact that majority of the queries are "reads".

I'm aware of tables with heavy writes - Hence such a configuration / module
(if available in runtime - even better) will help me to circumvent the
problem. Of course, I'm pretty much new to pgpool II - Please correct me
I've got it completely wrong.

Regards,
Dinesh
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.sraoss.jp/pipermail/pgpool-general/attachments/20170801/89d6af3c/attachment.html>


More information about the pgpool-general mailing list