On Tue, Apr 17, 2007 at 10:33:21PM -0400, Bruce Momjian wrote:
> Alvaro Herrera wrote:
> > I think that is too strong an assumption, which is why I'm planning to
> > back-patch the change to reset statement_timeout to 0 on autovacuum till
> > 8.0, as discussed. I think I should also backpatch the change to set
> > zero_damaged_pages as well (which is not on 8.0 AFAIR).
> >
> > It's very very easy to change things in postgresql.conf. Actually
> > knowing what you are doing (i.e. thinking on the consequences on VACUUM
> > and such) is a whole another matter.
>
> Frankly, setting statement_timeout in postgresql.conf seems so risky in
> so many ways, perhaps we just need to document that the parameter
> probably should not be set in postgresql.conf, and why.
I'd suggest doing both. Tell people that it's dangerous (probably to the
point of a comment in the sample config file), but *also* force it in
pg_dump since you can't really expect people to read the documentation.
//MAgnus