On 6/26/2007 3:00 PM, Brad Nicholson wrote:
> On Tue, 2007-06-26 at 14:53 -0400, Jerry Sievers wrote:
>> Hello;
>> 
>> Existing cluster with 1 set; 1 provider and 3 subscribers.  Adding
>> several new tables and sequences.
>> 
>> Slony 1.1.5
>> 
>> -- begin;
>> storeset(2, 'foo');
>> setaddtable(...);
>> ...
>> setaddsequence(...);
>> ...
>> subscribeset(...);
>> subscribeset(...);
>> subscribeset(...);
>> mergeset(1,2);   <-- is this OK?
>> commit;
>> 
>> Please advise.  I'm curious and assume the merge will happen at the
>> right time when all subscribers reach the subscribed state?
>> 
>> Or do I need to verify completed subscriptions first before issuing
>> the merge command? 
> 
> 
> I don't know what checks are in place against this, but if it was me,
> I'd wait until the subs were done before merging.

It is possible ... with 1.2.10, which has finally a working SYNC and 
WAIT FOR EVENT. The merge set must NOT be issued before all the 
subscribers are completely done with their subscriptions. No cascaded 
subscribe set commands should be issued before their data providers are 
done with their subscriptions either. Note that waiting for a subscribe 
set is a sequence of waiting for the subscribe itself being confirmed by 
the set origin, then issuing a slonik SYNC and waiting on the origin for 
that to be confirmed by the new subscriber.


Jan

-- 
#======================================================================#
# It's easier to get forgiveness for being wrong than for being right. #
# Let's break this rule - forgive me.                                  #
#================================================== JanWieck at Yahoo.com #