Re: How to fast the REINDEX

Hi Kevin,

Thank you for the update,

>>What does the table look like? What indexes are there?
Table has a combination of byteas. Indexes are b-tree and Partial

>>Why are you doing that?
Our table face lot of updates and deletes in a day, so we prefer reindex to
update the indexes as well overcome with a corrupted index.

>> How long?
More than 4 hrs..

>>What run time are you expecting?
Less than what it is taking at present.

>>It's hard to answer that without more information, like PostgreSQL
>>version and configuration, for starters. See:
version
------------------------------------------------------------------------------------------------------------
PostgreSQL 8.4.2 on i686-pc-linux-gnu, compiled by GCC gcc (GCC) 4.1.2
20080704 (Red Hat 4.1.2-44), 32-bit
(1 row)

>>http://wiki.postgresql.org/wiki/SlowQueryQuestions
Expected the performance question..

Regards
Raghavendra

On Thu, Apr 1, 2010 at 2:32 AM, Kevin Grittner
<Kevin(dot)Grittner(at)wicourts(dot)gov>wrote:

> raghavendra t <raagavendra(dot)rao(at)gmail(dot)com> wrote:
>
> > I have a table with 40GB size, it has few indexes on it.
>
> What does the table look like? What indexes are there?
>
> > When i try to REINDEX on the table,
>
> Why are you doing that?
>
> > its take a long time.
>
> How long?
>
> > I tried increasing the maintenance_work_mem, but still i havnt
> > find a satisfying result.
>
> What run time are you expecting?
>
> > Questions
> > =======
> > 1. What are the parameters will effect, when issuing the REINDEX
> > command
> > 2. Best possible way to increase the spead of the REINDEX
>
> It's hard to answer that without more information, like PostgreSQL
> version and configuration, for starters. See:
>
> http://wiki.postgresql.org/wiki/SlowQueryQuestions
>
> My best guess is that you can make them instantaneous by not running
> them. A good VACUUM policy should make such runs unnecessary in
> most cases -- at least on recent PostgreSQL versions.
>
> -Kevin
>