On Wed, 26 Jul 2006 10:49:25 +0200
Csaba Nagy <nagy at ecircle-ag.com> wrote:

> Gaviun,
> 
> > We can run this script again and again, and after a few seconds get a deadlock on a different table each time. The really annoying thing here is the deadlocked table isn't even one we want to ALTER. Is there anything we can do here except for 'try again at 4am when things are quiet'?
> 
> >From what I understood, slony will lock all tables in the set and remove
> the slony triggers before executing the script. It cannot possibly know
> which tables will be modified by the script, 

OK I thought there might have been a 'dry run' in a BEGIN/ROLLBACK block which watched which tables were being affected. Is this kind of thing planned for a future release?

> and it must lock the tables
> so nobody touches them between dropping the slony trigger, executing the
> script and recreating the slony triggers. So it locks all the tables in
> the set, in the order you defined them in the set. 

That I don't understand, because I specified the script to run on set 2, but I am only seeing deadlocks on tables which are only in set 1! :(

> If you have
> concurrently running transactions which lock tables in other order, you
> can easily get deadlocks...
> 
> So very likely the answer is do it at 4AM.

<slump>

Cheers,
Gavin.