Re: pgpool-II (max_pool and num_init_children)

From: Guillaume Lelarge <guillaume(at)lelarge(dot)info>
To: Geoffrey <lists(at)serioustechnology(dot)com>
Cc: pgsql-general(at)postgresql(dot)org
Subject: Re: pgpool-II (max_pool and num_init_children)
Date: 2010-07-02 14:08:06
Message-ID: 4C2DF2C6.2010900@lelarge.info
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-general

Le 02/07/2010 15:46, Geoffrey a écrit :
> Tatsuo Ishii wrote:
>>> I'm trying to get a handle on sane values for these two parameters.
>>> I assume that they should somehow correlate to my existing
>>> max_connections in my postgresql.conf file. Anyone using pgpool-II
>>> care to comment?
>>>
>>> I'm sure it's a balancing act between how many connections you want
>>> verses how many cached connections you have, but not sure how to
>>> properly assess this issue.
>>>
>>> (I've tried posting to the pgpool list, but it's apparently
>>> unavailable at this time)
>>
>> Here is an recommended rule:
>>
>> max_pool*num_init_children <= (max_connections -
>> superuser_reserved_connections)
>>
>> Because num_init_children defines the max concurrent connections
>> pgpool-II can accept, you might want to keep num_init_children as
>> large as possible. In this case you could lower max_pool to as low as
>> 1, which may cause performance degration because of low effect of
>> connection cache if you would have more than 1 user/database
>> combinations. However lower max_pool will not any error or any
>> problems except performance.
>
> We have a complicated set up. We have 13 databases with 12 postmasters.
> So that means 2 databases share a postmaster, the rest have their
> own. Consequently, separate pools for each postmaster. We will
> frequently have 20-30 or more users connecting to any one database,
> although some databases are much lower. This is further complicated by
> a web application that also connects to the databases.
>

You won't be able to handle more than one postmaster with one pgPool-II
in pooling mode. You'll need at least one pgPool-II for each postmaster.
And so, do one configuration for each pgPool-II.

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

In response to

Responses

Browse pgsql-general by date

  From Date Subject
Next Message Tom Lane 2010-07-02 14:12:42 Re: C-functions using SPI Missing Magic Block Error
Previous Message Scott Marlowe 2010-07-02 13:59:46 Re: Postgresql partitioning - single hot table or distributed