On Wed, Oct 04, 2006 at 12:57:35PM -0400, Tom Lane wrote:
> Ray Stell <stellr(at)cns(dot)vt(dot)edu> writes:
> > pg_ctl seems to want to be told where "data" is:
> > pg_ctl stop [-W] [-s] [-D datadir]
>
> > I've been scattering data across spindles and mount points.
>
> Via tablespaces you mean?
correct, to try to mitigate i/o contention. Doesn't this make
sense to do this?
> pg_ctl only cares about the top-level data
> directory (basically, so that it can find the postmaster.pid lock file).
good, maybe a note to this effect is the pg_ctl doc page is in order.