On Wed, Mar 21, 2018 at 12:00:52PM +0300, Arthur Zakirov wrote:
> On Tue, Mar 20, 2018 at 09:30:15PM +0100, Tomas Vondra wrote:
> > I wonder if these restrictions needed? I mean, why not to allow setting
> > max_shared_dictionaries_size below the size of loaded dictionaries?
> >
> > Of course, on the one hand those restriction seem sensible. On the other
> > hand, perhaps in some cases it would be useful to allow violating them?
> >
> > I mean, why not to simply disable loading of new dictionaries when
> >
> > (max_shared_dictionaries_size < loaded_size)
> >
> > Maybe I'm over-thinking this though. It's probably safer and less
> > surprising to enforce the restrictions.
>
> Hm, yes in some cases this check may be over-engineering. I thought that
> it is reasonable and safer in v7 patch. But there are similar GUCs,
> wal_keep_segments and max_wal_size, which don't do additional checks.
> And people are fine with them. So I removed that check from the variable.
>
> Please find the attached new version of the patch.
I forgot to fix regression tests for max_shared_dictionaries_size. Also
I'm not confident about using pg_reload_conf() in regression tests.
I haven't found where pg_reload_conf() is used in tests. So I removed
max_shared_dictionaries_size tests for now.
Sorry for the noise.
--
Arthur Zakirov
Postgres Professional: http://www.postgrespro.com
Russian Postgres Company
