Re: [HACKERS] Index greater than 8k

On October 31, 2006 06:42 am, Joshua D. Drake wrote:
> Teodor Sigaev wrote:
> >> The problem I am after is the 8k index size issue. It is very easy to
> >> get a GIST index (especially when using tsearch2) that is larger than
> >> that.

The problem as I remember it is pg_tgrm not tsearch2 directly, I've sent a
self contained test case directly to Teodor which shows the error.

'ERROR: index row requires 8792 bytes, maximum size is 8191'

> >
> > Hmm, tsearch2 GIST index is specially designed for support huge index
> > entry:
> > first, every lexemes in tsvectore are transformed to hash value (with a
> > help of crc32), second, it's stripped all position infos, third, if size
> > of array is greater than TOAST_INDEX_TARGET then tsearch2 will make bit
> > signature of tsvector. Signature's length is fixed and equals to 252
> > bytes by default (+ 8 bytes for header of datum). All values on internal
> > pages are represented as signatures below.
> >
> > So, tsearch2 guarantees that index entry will be small enough. If it's
> > not true, then there is a bug - pls, make test suite demonstrating the
> > problem.
> >
> >> Is recompiling the block size the option there?
> >> What are the downsides, except for the custom build?
> >
> > Can you send exact error message?
>
> I am training this week, but Darcy can do it. Can you give them a test
> case on what we were working on with that customer?
>
> Joshua D. Drake

--
Darcy Buskermolen
Command Prompt, Inc.
Sales/Support: +1.503.667.4564 || 24x7/Emergency: +1.800.492.2240
PostgreSQL solutions since 1997
http://www.commandprompt.com/