Hi,
On 2018-10-12 11:07:58 +0900, Michael Paquier wrote:
> On Thu, Oct 11, 2018 at 06:53:19PM -0700, Andres Freund wrote:
> > Hm. Maybe I'm confused, but how is it correct to use a blacklisting
> > approach here? It's far from absurd to have files inside a tablespacs
> > when using temp_tablespace that aren't written through the buffer
> > manager. Like, uh, temp files, when using temp_tablespaces.
>
> pg_verify_checksums assumes that the cluster has been cleanly shut
> down so temp files are not an issue, no? However...
We only remove temp dirs on startup, and I'm pretty sure there at least
not too long ago were a few error cases where we can leak temp files in
individual sessions. And there's talk about allowing
pg_verify_checksums when online. So this seems to be an angle that
needs to be thought about...
Greetings,
Andres Freund