Horton

Re: Unhappy about API changes in the no-fsm-for-small-rels patch

Amit Kapila <amit.kapila16@gmail.com>

View thread View on postgr.es
From: Amit Kapila <amit.kapila16@gmail.com>
To: Alvaro Herrera <alvherre@2ndquadrant.com>
Cc: John Naylor <john.naylor@2ndquadrant.com>, Andres Freund <andres@anarazel.de>, Tom Lane <tgl@sss.pgh.pa.us>, PostgreSQL-development <pgsql-hackers@postgresql.org>
Date: 2019-05-02T03:24:47Z
Lists: pgsql-hackers 
On Thu, May 2, 2019 at 3:42 AM Alvaro Herrera <alvherre@2ndquadrant.com> wrote:
>
> On 2019-May-01, Amit Kapila wrote:
>
> > On Tue, Apr 30, 2019 at 7:52 PM Alvaro Herrera <alvherre@2ndquadrant.com> wrote:
>
> > > Hmm ... so, if vacuum runs and frees up any space from any of the pages,
> > > then it should send out an invalidation -- it doesn't matter what the
> > > FSM had, just that there is more free space now.  That means every other
> > > process will need to determine a fresh FSM,
> >
> > I think you intend to say the local space map because once FSM is
> > created we will send invalidation and we won't further build relcache
> > entry having local space map.
>
> Yeah, I mean the map that records free space.
>
> > > but that seems correct.  Sounds better than keeping outdated entries
> > > indicating no-space-available.
> >
> > Agreed, but as mentioned in one of the above emails, I am also bit
> > scared that it should not lead to many invalidation messages for small
> > relations, so may be we should send the invalidation message only when
> > the entire page is empty.
>
> I don't think that's a concern, is it?  You typically won't be running
> multiple vacuums per second, or even multiple vacuums per minute.
>

That's right.  So let's try by adding invalidation call whenever space
is reduced.  Is there a good way to test whether the new invalidation
calls added by this patch has any significant impact?

-- 
With Regards,
Amit Kapila.
EnterpriseDB: http://www.enterprisedb.com

Commits

  1. Revert "Avoid the creation of the free space map for small heap relations".

  2. Improve code comments in b0eaa4c51b.

  3. During pg_upgrade, conditionally skip transfer of FSMs.

  4. Add more tests for FSM.

  5. Fix typos in commit 8586bf7ed8.

  6. tableam: introduce table AM infrastructure.

  7. Clear the local map when not used.

  8. Doc: Update the documentation for FSM behavior for small tables.

  9. Make FSM test portable.

  10. Avoid creation of the free space map for small heap relations, take 2.