Re: 8.3.3 stability ?

On Wed, 2008-09-17 at 16:07 -0400, Tom Lane wrote:
> "Scott Marlowe" <scott(dot)marlowe(at)gmail(dot)com> writes:
> > On Wed, Sep 17, 2008 at 11:31 AM, Gauthier, Dave
> > <dave(dot)gauthier(at)intel(dot)com> wrote:
> >> Time for an upgrade. How stable is 8.3.3 ?
>
> > If I was deploying to production today, I'd use 8.3.3 no questions.
>
> FWIW, 8.3.4, which is due out Monday, squashes about half a dozen
> interesting bugs that are not also present in 8.2.x. I'd venture that
> as of 8.3.4 we are about down to the point where 8.3.latest and
> 8.2.latest are of equivalent reliability. Which is not to say that
> 8.3.x won't still have some new bugs not present in 8.2.x, but that
> I think it'll be about a wash when you consider the problems 8.3 fixes
> that are unfixable in 8.2.

Reliability is a wide topic. 8.2 reliably works as intended, though 8.3
certainly has better "intentions" as to how things *should* work. I
would say that 8.3 has fewer "unintended consequences" for the average
user and a ton of benefits:
http://www.2ndquadrant.com/download/Postgres_Performance_Update.pdf
If your definition of software reliability includes good behavioural
characteristics as well as absence of bugs, then 8.3 is a must.

--
Simon Riggs www.2ndQuadrant.com
PostgreSQL Training, Services and Support