To: "Mauro Bertoli" <bertolima@yahoo.it>, "Richard Huxton" <dev@archonet.com>
Cc: pgsql-sql@postgresql.org
Subject: Re: Postgres performance
References: <20050306144219.43338.qmail@web51408.mail.yahoo.com>
Message-ID: <opsm8dq2c7th1vuj@musicbox>
Date: Sun, 06 Mar 2005 20:29:16 +0100
From: PFC <lists@boutiquenumerique.com>
Organization: =?iso-8859-15?Q?La_Boutique_Num=E9rique?=
Content-Type: text/plain; format=flowed; delsp=yes; charset=iso-8859-15
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
In-Reply-To: <20050306144219.43338.qmail@web51408.mail.yahoo.com>
User-Agent: Opera M2/7.54u1 (Linux, build 892)


>> 	Really ?
>> 	In my experience this is a sure way to get
>> inconsistencies slowly
>> creeping into your database, and you also get a load
>> of funky concurrency
>> issues.
> Yes, you are rigth... my insert/update are very simple
> and without problems and so I think to use 'foreign
> key' coded to make faster/simpler the management and
> don't overloading the db (and use exception code
> management )... but I had a problem with pgSQL because
> server was very busy and the same query was replicate
> (because users refresh the page... :( ) so now I've
> foreign keys...

	It's a typical case, program being stopped between insertion of parent  
and child row. Although in this case FK's is not the solution,  
transactions are.

> What about it? (i think it's the same with mySQL...)
> I don't know how users can connect to... 1, 10,
> 1000... I must create a pool with 1000 connections? is
> this fine? if  connections aren't released I must
> reset manually, it is dangerous... do you think?

	Just set the max number of connections for postgres a bit higher than the  
max number of apache processes in apache.conf.