Re: 9.2.1 & index-only scans : abnormal heap fetches after VACUUM FULL

On Thu, Nov 28, 2013 at 05:17:07PM -0500, Robert Haas wrote:
> > I need to know this is the right approach, and need to know what things
> > are wrong or missing.
>
> The fact that you've needed to modify SetHintBits() to make this work
> is pretty nasty. I'm not exactly sure what to do about that, but it
> doesn't seem good.

Hey, if that's the worst of the problems, I will be very happy. There
is a sense that because we are not using the shared buffer cache and not
WAL logging, we have to skip some stuff.

> This patch also has the desirable but surprising consequence that hint
> bits will be set as a side effect of update_page_vm(), which means
> that you'd better do that BEFORE checksumming the page.

Yes, I already saw that and fixed it in my git tree.

> I wonder if we ought to mark each page as all-visible in
> raw_heap_insert() when we first initialize it, and then clear the flag
> when we come across a tuple that isn't all-visible. We could try to
> set hint bits on the tuple before placing it on the page, too, though
> I'm not sure of the details.

I went with the per-page approach because I wanted to re-use the vacuum
lazy function. Is there some other code that does this already? I am
trying to avoid yet-another set of routines that would need to be
maintained or could be buggy. This hit bit setting is tricky.

And thanks much for the review!

--
Bruce Momjian <bruce(at)momjian(dot)us> http://momjian.us
EnterpriseDB http://enterprisedb.com

+ Everyone has their own god. +