Re: Synchronous replication patch v2

On Fri, 2008-11-14 at 19:23 +0200, Heikki Linnakangas wrote:
> Simon Riggs wrote:
> > On Fri, 2008-11-14 at 19:00 +0200, Heikki Linnakangas wrote:
> >> Why do we need a separate XLogsndRqst variable in shared memory? Don't 
> >> we always want to send the WAL up to the same point as we flush it?
> > 
> > If we're doing synch rep and we're committing.
> 
> You flush and send the WAL, up to the same point?

Yes, but you may make progress towards it in different size steps.

> > What happens when we're
> > doing async rep or running something like a large load. 
> 
> You don't flush, and you don't request the WAL to be sent? The 
> background writer and WAL sender can still wake up periodically, and 
> write and send the WAL as they find convenient.

With WAL writes we write and flush at the same time. With WAL sending
that doesn't sound such a good plan.

-- Simon Riggs           www.2ndQuadrant.comPostgreSQL Training, Services and Support