Tom Lane writes:
> Peter Eisentraut <peter_e@gmx.net> writes:
> > To alleviate the need for a start up script somewhat we could perhaps fix
> > up pg_ctl to serve that purpose better.
>
> > -w should be the default, as has been suggested.
>
> Agreed.
But we need to make up a better way to wait for postmaster startup. The
current 'psql -l' is no good, as we have discussed, but if this thing is
going into system startup scripts mainstream now we can't have that kind
of random behaviour.
Actually, trying to connect is wrong-headed anyway. If the user has a
peculiar setup to disallow connections from the local host (virtual host
maybe), even if it is only a temporary misconfiguration, we should still
be able to start up in good faith.
How is "successfully started up" constrained anyway? Even if we had a way
to check that, the postmaster could still choose to abort briefly after
our check, e.g., during the backend reaping phase.
Maybe waiting on startup isn't so important to worry about it.
--
Peter Eisentraut peter_e@gmx.net http://yi.org/peter-e/
