[Pgpool-general] PgPool install, Part Deux

[Marc G. Fournier]

Okay, doing some more thinking of whwat you state here ... what about 
stuff like:

CREATE TABLE test ( id serial, data text );

IF there are two connections to the databases (through pgpoool) and both 
issue:

INSERT INTO test ( data ) VALUES ( 'conn1' );
INSERT INTO test ( data ) VALUES ( 'conn2' );
INSERT INTO test ( data ) VALUES ( 'conn3' );

at the same time ... is it possible that backend 1 will have:

1, conn1
2, conn2
3, conn3

while backend 2 has:

1, conn2
2, conn1
3, conn3

Or does pgpool pass them back sequentially so that both sides will have it 
in the same order?

Basically, with your example, it is easy to see how now() could be a 
problem, as if there is a slight delay on backend2 over backend1, the 
dates could / would drift some ... but what about stuff like serial keys?

Tatsuo?


  On Wed, 8 Jul 2009, Marc G. Fournier wrote:

> On Wed, 8 Jul 2009, J. Carlos Muro wrote:
>
>> The meaning of "ready for production environments" somehow (or "some much") 
>> depends on your requirements.. I don't want to go into deep discussion with 
>> it as we will find no end. But, we can make the question in another 
>> direction: is you production environment ready for pgpool? In my company we 
>> aren't.. yet. We are migrating all databases from a very expensive db 
>> system to PG*. I have found that we have tables with fields "default 
>> now()", and in the applications there are inserts sort of "insert into 
>> table1 values (now())". I have instructed our developers to study the 
>> possibility of generating dates from the application level. Later, we will 
>> see how Hibernate behaves with pgpool.. Well, we are adapting our 
>> environment to work with pgpool. You know, if it finally works, we will 
>> save lots of money. :)
>
> Now *that* does help ... and I suspect that this may be part of what my 
> problem was last time ... sounds like the current PgPool-II (without 
> disparaging it) is more what you build around vs install pre-existing 
> software onto ... or, at least, not software you don't have access to the 
> source code for :)
>
> Have to play with this sum ...
>
> Thanks ... again, the above really puts it into perspective ...
>
> ----
> Marc G. Fournier           Hub.Org Networking Services (http://www.hub.org)
> Email . scrappy at hub.org                              MSN . scrappy at hub.org
> Yahoo . yscrappy               Skype: hub.org        ICQ . 7615664
>

----
Marc G. Fournier           Hub.Org Networking Services (http://www.hub.org)
Email . scrappy at hub.org                              MSN . scrappy at hub.org
Yahoo . yscrappy               Skype: hub.org        ICQ . 7615664