Re: [GENERAL] ANALYZE getting dead tuple count hopelessly wrong

["Pavan Deolasee" <pavan(dot)deolasee(at)gmail(dot)com>]

On Mon, Mar 31, 2008 at 9:02 PM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:

>
>  [ Please see if you can stop using the "redirected dead" terminology ]
>
>

Apologies, will keep that in mind. Seems like a hang-over from the past :-)

>  Yeah, I think I agree.  The page pruning code is set up so that changing
>  a line pointer to DEAD state doesn't change the count of dead tuples in
>  the table, so we are counting unreclaimed DEAD pointers as still being
>  dead tuples requiring VACUUM.  ANALYZE should surely not affect that.
>
>  It looks like there's no trivial way to get ANALYZE to do things that
>  way, though.  heap_release_fetch() doesn't distinguish a DEAD line
>  pointer from an unused or redirected one.  But in the current
>  implementation of ANALYZE there's really no benefit to using
>  heap_release_fetch anyway --- it always examines all line pointers
>  on each selected page, so we might as well rewrite it to use a simple
>  loop more like vacuum uses.
>

I agree. I would write a patch on these lines, unless you are already on to it.


Thanks,
Pavan

-- 
Pavan Deolasee
EnterpriseDB http://www.enterprisedb.com