From: | ghaverla(at)freenet(dot)edmonton(dot)ab(dot)ca |
---|---|
To: | Andrew McMillan <andrew(at)catalyst(dot)net(dot)nz> |
Cc: | pgsql-novice(at)postgresql(dot)org |
Subject: | Re: index item size 4496 exceeds maximum 2713 |
Date: | 2002-03-06 10:13:19 |
Message-ID: | Pine.A41.3.95.1020306030937.69500D-100000@fn2.freenet.edmonton.ab.ca |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-novice |
On 6 Mar 2002, Andrew McMillan wrote:
> On Tue, 2002-03-05 at 13:27, Jon Hassen wrote:
> > "ERROR: btree: index item size 4496 exceeds maximum 2713"
> >
> > On another PGSQL database I only got that message when the size was above
> > 8192. How can I change my database to use the full 8192 size? Or how can I
> > get around this problem at all?
>
> The maximum _indexable_ field size is 1/3 of the blocksize. In most
> cases this will be 2713 which is 8192/3.
>
> In reality there is usually very little value in indexing fields larger
> than a few hundred bytes, unless you are doing keyword or full-text
> indexing.
[ ... ]
Interesting note, not that I have the problem right now, but in
another (nameless) dbase, I will also have long fields I want
to index (GIS metadata). Where does this number of 3 come from?
Just some small random integer? Is there some fraction/performance
curve somewhere, which shows 3 is in some sense optimal?
Gord
Matter Realisations http://www.materialisations.com/
Gordon Haverland, B.Sc. M.Eng. President
101 9504 182 St. NW Edmonton, AB, CA T5T 3A7
780/481-8019 ghaverla @ freenet.edmonton.ab.ca
780/993-1274 (cell)
From | Date | Subject | |
---|---|---|---|
Next Message | Tom Lane | 2002-03-06 16:14:23 | Re: index item size 4496 exceeds maximum 2713 |
Previous Message | Ivan Horvath | 2002-03-06 09:49:26 | referential integrity for insert |