Re: Autovaccuum vs temp tables crash

On Sat, Feb 23, 2019 at 4:18 PM Michael Paquier <michael(at)paquier(dot)xyz> wrote:

> On Sat, Feb 23, 2019 at 02:48:58PM +0100, Magnus Hagander wrote:
> > I think we need to either prevent dropping of temp namespaces *or* we
> need
> > to create a new entry in pg_namespace in this particular case.
>
> Perhaps I am missing something, but it would be just more simple to
> now allow users to restrict that?
>

I can't parse what you are saying here. Now allow users to restrict what?

> I wonder if other "fun" things could happen if you go rename the
> namespace,
> > haven't tried that yet...
>
> In this case the OID remains the same, still there are some cases
> where we rely on the namespace name, and one is CLUSTER.
> objectaddress.c uses as well get_namespace_name_or_temp(), which would
> be messed up, so it would be better to prevent a temp namespace to be
> renamed. Could ALTER SCHEMA OWNER TO also be a problem?
>

Or possibly altering permissions on it?

--
Magnus Hagander
Me: https://www.hagander.net/ <http://www.hagander.net/>
Work: https://www.redpill-linpro.com/ <http://www.redpill-linpro.com/>