hi,
we got about 440 tables currently in one set (with some 90 sequences)

and we are replicating over a WAN (across the world) and we do all 
schema changes with out-most care but we do modify the schema pretty 
heavily using EXECUTE_SCRIPT, but in a very organized manner (see my 
older comments). we got short release cycles around 3 months, and at 
that time we got about 200-300 DDL changes on average (new tables, 
dropping tables, column changes, renamings, etc)

But yes you need to do it with great care and extensive testing, and we 
do have a production outage of about 1h. but then again if it does fail 
it does take us a good 72h to fully replicate everything to all the 
nodes.. so not really an option.

normally we do full physical backup of all postgres data/ directories on 
all nodes before starting the update scripts, so if things do happen to 
go to crap due to some unforseen incident then we can just bring it back 
to the old state very quickly and try again.

cheers
Martin


Adam Olsen wrote:
> The issues described in section 7.2 here:
> http://slony.info/documentation/definingsets.html
>
> The most concerning one is the one regarding "EXECUTE SCRIPT", as, in
> the beginning our schema might change somewhat frequently.
>
> Adam
>
> On Tue, Aug 19, 2008 at 10:56 AM, Shahaf Abileah <shahaf at redfin.com> wrote:
>   
>> What deadlocking issues?
>>
>> We have about 100 tables in a single set and things are working out
>> pretty well.
>>
>> Note: not all of our 100 tables are related to one another via foreign
>> keys; we chose to keep everything in a single set purely for simplicity.
>>
>> --S
>>
>> -----Original Message-----
>> From: slony1-general-bounces at lists.slony.info
>> [mailto:slony1-general-bounces at lists.slony.info] On Behalf Of Adam Olsen
>> Sent: Tuesday, August 19, 2008 9:32 AM
>> To: slony1-general at lists.slony.info
>> Subject: [Slony1-general] Help with grouping sets
>>
>> Hello,
>>
>> First off, I'm sorry if this seems like a stupid question.
>>
>> I'm trying to group my tables into sets.  In the online documentation,
>> it says that "It will be vital to group tables together into a single
>> set if those tables are related via foreign key constraints".  I am
>> creating a Django application, and every table is related to some
>> other table via a foreign key.  Sometimes this relation is to the
>> "User" table, and sometimes this relation is to a table that has a
>> relation to the "User" table.  We don't have any tables that aren't
>> somehow associated to the "User" table.
>>
>> We have quite a few tables, so this means that, to my understanding,
>> all of our tables need to be in a single set.  From what I understand,
>> this can cause deadlocking issues, and I'd obviously rather avoid
>> that.
>>
>> What can I do to better group my sets?
>>
>> --
>> Adam Olsen
>> SendOutCards.com
>> http://www.vimtips.org
>> http://last.fm/user/synic
>> _______________________________________________
>> Slony1-general mailing list
>> Slony1-general at lists.slony.info
>> http://lists.slony.info/mailman/listinfo/slony1-general
>>
>>
>>
>>     
>
>
>
>