Dmitry Koterov wrote:
> Hello.
>
> Possibly this fact needs to be included in the documentation?
>
> Steps to reproduce:
>
> 1. Create table with name "b" on slave manually (emulate an error of 
> broken replication, see below).
> 2. Check that there is no table "b" on master.
> 3. Run via slonik's EXECUTE DDL within TRY block (in the single 
> transaction):
>
> CREATE TABLE a(id INTEGER);
> CREATE TABLE b(id INTEGER);
>
> 4. On master, this query successfully finishes. On slave, only table 
> "a" is created (and committed!), but "b" is not (it already exists).
> 5. Conclustion: on master, "a" and "b" are created within the single 
> transaction. On slave, "a" and "b" are created within different 
> transactions.
This behaviour seems quite surprising.

What version of Slony-I did you observe this in?  There was a bug of this sort that was resolved in 1.2.12.
-- 
(format nil "~S@~S" "cbbrowne" "ca.afilias.info")
<http://dba2.int.libertyrms.com/>
Christopher Browne
(416) 673-4124 (land)
"Bother,"  said Pooh,  "Eeyore, ready  two photon  torpedoes  and lock
phasers on the Heffalump, Piglet, meet me in transporter room three"