On Mon, Nov 24, 2008 at 10:45:42AM -0500, Tom Lane wrote:
> Sam Mason <sam@samason.me.uk> writes:
> > It's a normal 32bit Intel Debian system, nothing much special done
> > to increase the kernel/user split or anything like that as far as I
> > remember on this box. If I try with larger sizes it falls over with
> > "out of memory", but up until around 755MB (760MB fails) it gives back
> > "INSERT 0 1" which I've always read as inserting a row. A select on the
> > table gives this inserted row containing a zero length string.
>
> Well, I can't reproduce that here. Something strange about your
> configuration maybe?
Not that I know of. I've just created a test cluster to make sure and I
get the same behaviour. Minimal set of commands are:
LANG=C /usr/lib/postgresql/8.3/bin/initdb pg83
vi pg83/postgresql.conf
changed port to 5444 to prevent conflicts, everything else default.
/usr/lib/postgresql/8.3/bin/postgres -D pg83 -k /tmp
in a seperate shell:
echo "create database smason;" | psql -p 5444 -h /tmp template1
echo "create table test ( col text );" | psql -p 5444 -h /tmp
./test 64 | psql -p 5444 -h /tmp
results in:
INSERT 0 1
Time: 3354.269 ms
./test 512 | psql -p 5444 -h /tmp
results in:
INSERT 0 1
Time: 50.452 ms
echo "select length(col) from test;" | psql -p 5444 -h /tmp
results in:
length
----------
67108864
0
Anything else to try? All PG packages are from debian backports with
version "8.3.4-2~bpo40+1", anybody else getting this?
Hum, maybe I should try building from source and see if it's something
to do with the Debian packages. [/me twiddles thumbs while code builds]
... done ... I get the same result for both 8.3.4 and 8.3.5. I'm kind
of stuck for ideas now!
Sam