On Thursday, May 24, 2012 06:35:06 PM Heikki Linnakangas wrote:
> On 24.05.2012 18:16, Robert Haas wrote:
> > On Thu, May 24, 2012 at 10:13 AM, Joachim Wieland<joe@mcknight.de> wrote:
> >> I wouldn't have assumed any corruption was possible given that I did
> >> clean shutdowns on both sides...
> >
> > The thing that's worrying me is that there's not really any such thing
> > as a "clean" shutdown on a standby. When you shut down the master, it
> > checkpoints. When you shut down the standby, it can't checkpoint, so
> > I think it's still going to enter recovery at startup. It'd be
> > interesting to know where that recovery began and ended as compared
> > with the minimum recovery point just before the shutdown.
>
> Perhaps we should introduce the concept of a clean standby shutdown. We
> can't write a checkpoint record, but we could write the same information
> somewhere else. Like in the control file. At startup, we'd see that we
> did a clean shutdown at WAL point X/X, and start up without having to
> read through all the WAL from the last master checkpoint.
The control file currently is not a very good match because of the current
requirement of staying below 512 bytes. If we would include the list of
running xacts that wouldn't be enough.
I wondered before if there is more to do to fix that then to do the atomic
write();fsync();rename();fsync(); dance. I don't see a problem with the cost
of that personally...
Andres