On Fri, Apr 26, 2019 at 12:11 AM Robert Haas <robertmhaas@gmail.com> wrote:
>
> On Tue, Apr 23, 2019 at 7:09 AM Masahiko Sawada <sawada.mshk@gmail.com> wrote:
> > For the second issue, I've changed lazy vacuum so that it reports both
> > the number of kilobytes we freed and the number of kilobytes can be
> > freed after index cleanup.
>
> I am not very convinced that this reporting is in any useful to users.
> Despite N kilobytes of tuples having been freed, the pages themselves
> are still allocated and the actual ability to reuse that space may be
> dependent on lots of factors that the user can't control like the
> sizes of newly-inserted tuples and the degree to which the free space
> map is accurate.
Hmm, it's a term problem? The phrase 'x bytes vacuumed' would solve it?
>
> I feel like we're drifting off into inventing new kinds of reporting
> here instead of focusing on fixing the reported defects of the
> already-committed patch, but perhaps I am taking too narrow a view of
> the situation.
I should have divided the patches into two: fixing assertion error and
the reporting. I think we could think the latter issue also is a kind
of bug because it can report something like "1000 index tuples
vacuumed but 0 heap tuple vacuumed" in case where the vacuumed table
had only dead line pointers. Maybe I should add an another open item
for the latter.
Attached the split version patches.
Regards,
--
Masahiko Sawada
NIPPON TELEGRAPH AND TELEPHONE CORPORATION
NTT Open Source Software Center