On Mon, May 14, 2007 at 03:56:12PM +0400, Dmitry Koterov wrote:
> >> So, if anybody reads from sl_log_status, it blocks until anything is
> >written
> >> to sl_log_status. 

> >It sounds like a way to make everything a lot slower.
> 
> 
> Why? Where is the slowdown?

If you have many slaves trying to read, they're all blocked.  This is
effectively going to single-thread all the replicas.  It will be
slower.

> main problem - slave checking. The main problem for me is to reduce the
> number of polling queries without increasing the slave lag time.

And as people keep telling you, that just isn't possible _by design_. 
The polling is what keeps the slaves in sync.  I understand what you
are trying to do, but it wasn't a target use case for Slony-I (in
fact, we explicitly considered and rejected this case, AFAIR).  Not
all replication strategies are appropriate for every use case, and
this one is not appropriate for yours, I think.  Likely what you
really need is something like Postgres-R.  Or maybe statement
replication.

A

-- 
Andrew Sullivan  | ajs at crankycanuck.ca
This work was visionary and imaginative, and goes to show that visionary
and imaginative work need not end up well. 
		--Dennis Ritchie