______________________

> From: Jan Wieck <JanWieck at Yahoo.com>
>To: Vick Khera <vivek at khera.org> 
>Cc: "slony1-general at lists.slony.info" <slony1-general at lists.slony.info> 
>Sent: Friday, 6 July 2012, 3:41
>Subject: Re: [Slony1-general] slonik_execute_script not working
> 
>On 7/4/2012 3:39 PM, Vick Khera wrote:
>> On Mon, Jul 2, 2012 at 3:29 PM, Wolf Schwurack <wolf at uen.org
>> <mailto:wolf at uen.org>> wrote:
>>
>>     You can not run DDL statement using slony, it states that in slony docs
>>
>>
>> Where? That's the only safe way to perform DDL on a replicated table.
>
>In 1.x yes. Since 2.0 DDL can be done directly with psql or by other 
>means. If doing so, one needs to be careful which modifications are done 
>to which nodes in which order so that inserts/updates aren't coming into 
>a node with columns or things like that. But Slony by itself won't stand 
>in the way any more.
>

Hi Jan,

I'm assuming most rules regarding replicated tables that applied in 1.x still apply in 2.0 though, i.e. slony still needs to know about extra colums added outside of execure script, primary keys etc.

Disregarding things such as trigger activity etc, isn't Vick's statement is still correct if the DDL changes the structure or primary keys of  the tables?  Slony needs to know about those changes.


We're on 2.0 and I've got a console that reads the slony configuration and allows our devs to execute a restricted set of DDL under their own credentials via a 2 phase commit, the rest is done via execute script.