Re: Postgresql concern of effect of invalid index

Matheus de Oliveira <matioli(dot)matheus(at)gmail(dot)com> wrote:
> Burgess, Freddie <FBurgess(at)radiantblue(dot)com> wrote:
>
>> My question is? What are the ramifications of having this
>> spatial indexes remain in this state?

>> update pg_index set indisvalid = false
>>   where indexrelid = 'sidx_sponser_report_y2014m06'::regclass;
>
> If you marked it as invalid, then it is not going to be updated
> anymore.

I think you are confusing indisvalid with indisready:

http://www.postgresql.org/docs/current/interactive/catalog-pg-index.html

Flagging it is invalid should suppress its use for queries, yet
still do all the work of maintaining it.

> BTW, you'd better take this to -performance list and discuss
> about why it is choosing a *bad* index instead of messing with
> the catalog.

+1

In any event, this is most definitely *not* a bug report, so it
belongs on a different list.

--
Kevin Grittner
EDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company