On Apr 30, 2013, at 8:13 PM, Tom Lane wrote:
> Appears to be sheer blather, or at least not tempered by any thoughts
> of whether it'd work in special cases.
As the author of that blog post, I'd go with option (b). :) And that lack of tempering was actually the point.
The point there was that having the tablespace directories disappear shouldn't be considered a normal operational
model. (Like, for example, putting a tablespace in a RAM disk.) There's no question that you can patch the database
backtogether if the underlying storage of a tablespace disappears, but that should be thought of as disaster recovery,
notas a "oh, third time this week" operation.
--
-- Christophe Pettus
xof@thebuild.com