Thread: multixact optimization patches

I have just pushed two optimization patches for multixacts to HEAD only.
I hesitate to backpatch them to 9.3 right away, but will consider doing
so if people think differently.

I also have a patch that supposedly fixes the performance regression
reported in bug #8470, but it's considerably more obscure so I'm not
pushing it right now.  It's attached.  I'd need to spend some more time
with it to ensure it doesn't break other cases before pushing.  I know
some people is interested in this fix and thought I'd throw it out there
to gather some input.

Since it affects much of the same code as the two patches I just pushed,
using it in 9.3 requires cherry-picking those patches, but they apply
without conflicts so it should be easy.

--
Álvaro Herrera                http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services

On Tue, Jul 29, 2014 at 03:56:19PM -0400, Alvaro Herrera wrote:
> I have just pushed two optimization patches for multixacts to HEAD only.
> I hesitate to backpatch them to 9.3 right away, but will consider doing
> so if people think differently.
> 
> I also have a patch that supposedly fixes the performance regression
> reported in bug #8470, but it's considerably more obscure so I'm not
> pushing it right now.  It's attached.  I'd need to spend some more time
> with it to ensure it doesn't break other cases before pushing.  I know
> some people is interested in this fix and thought I'd throw it out there
> to gather some input.
> 
> Since it affects much of the same code as the two patches I just pushed,
> using it in 9.3 requires cherry-picking those patches, but they apply
> without conflicts so it should be easy.

Wow, this is some very complex code.  Are you closer to a decision on it?

--  Bruce Momjian  <bruce@momjian.us>        http://momjian.us EnterpriseDB
http://enterprisedb.com
 + Everyone has their own god. +

Bruce Momjian wrote:
> On Tue, Jul 29, 2014 at 03:56:19PM -0400, Alvaro Herrera wrote:
> > I have just pushed two optimization patches for multixacts to HEAD only.
> > I hesitate to backpatch them to 9.3 right away, but will consider doing
> > so if people think differently.
> > 
> > I also have a patch that supposedly fixes the performance regression
> > reported in bug #8470, but it's considerably more obscure so I'm not
> > pushing it right now.  It's attached.  I'd need to spend some more time
> > with it to ensure it doesn't break other cases before pushing.  I know
> > some people is interested in this fix and thought I'd throw it out there
> > to gather some input.
> > 
> > Since it affects much of the same code as the two patches I just pushed,
> > using it in 9.3 requires cherry-picking those patches, but they apply
> > without conflicts so it should be easy.
> 
> Wow, this is some very complex code.  Are you closer to a decision on it?

I am not, but I note that the input (such as benchmarks) I have received
on it  is nil -- which is surprising given that a number of people had
pinged me about this issue previously.

-- 
Álvaro Herrera                http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services