On Fri, 2011-02-18 at 09:41 -0600, Kevin Grittner wrote:
> Robert Haas <robertmhaas@gmail.com> wrote:
> > Simon Riggs <simon@2ndquadrant.com> wrote:
>
> >> Make a hard state change from catchup to streaming mode.
> >> More useful state change for monitoring purposes, plus a
> >> required change for synchronous replication patch.
> >
> > As far as I can see, this patch was not posted or discussed before
> > commit, and I'm not sure it's the behavior everyone wants. It has
> > the effect of preventing the system from ever going backwards from
> > "streaming" to "catchup". Is that what we want?
>
> We are looking at moving to streaming replication instead of WAL
> file shipping, but we often have WAN outages. These can last
> minutes, hours, or even a few days. What would be the impact of
> this patch on us during and after such outages?
None at all. The patch introduces no behavioural changes, only a useful,
but minor re-categorisation of what is already happening so that its
easier to monitor what happens following startup of a standby.
> I don't know how well such experience generalizes, but my personal
> experience with replication technology is that "hard state changes"
> tend to make things more "clunky" and introduce odd issues at the
> state transitions. Where different message types are intermingled
> without such hard state changes, I've seen more graceful behavior.
>
> Of course, take that with a grain of salt -- I haven't read the
> patch and am talking in generalities based on having written a
> couple serious replication tools in the past, and having used a few
> others.
I respect your experience.
-- Simon Riggs http://www.2ndQuadrant.com/books/PostgreSQL Development, 24x7 Support, Training and Services