Ed L. pgsql
Wed Oct 27 16:29:17 PDT 2004
On Wednesday October 27 2004 9:26, Ed L. wrote:
> On Wednesday October 27 2004 4:18, Andrew Sullivan wrote:
> > On Tue, Oct 26, 2004 at 11:21:57PM -0600, Ed L. wrote:
> > > If I'm reading you correctly, I think you're saying that sequences
> > > will be not syncronized between master and slave until failover.  Is
> > > that correct?
> >
> > No.  What Jan said is that they'll be more accurately synchronised at
> > switchover.  At failover, you'll possibly have a bigger gap than you
> > expected.
>
> Are sequences syncronized through slon *AT ALL* during replication
> without a failover?  If so, what slonik commands set that up?

And I'm talking specifically about syncronization meaning "select last_value 
from sequence_name", because that's the one that matters in the vast 
majority of cases where a sequence object is being used as a serial 
auto-incrementing value.  If nextval gives a larger number on the slave 
than it would have on the master, that won't cause a problem for this 
primary case, and will only be a problem for folks who had false 
assumptions about how sequences work with regard to gaps.

Ed



More information about the Slony1-general mailing list