The only change that could affect between 42.2.10 (which works based
on your first message), and 42.2.11+ is the introduction of the tuple
abstraction.
The autocommit=false + fetchSize combo should help with large tables:
https://jdbc.postgresql.org/documentation/head/query.html#fetchsize-example
On Thu, Jul 23, 2020 at 4:19 PM Christophe combet <chrisc_pro@yahoo.fr> wrote:
>
> Dear Dave,
>
> The table definition is:
>
> CREATE TABLE mytable (
> column1 character varying(14) NOT NULL,
> column2 integer NOT NULL,
> column3 bytea NOT NULL
> );
>
> ALTER TABLE ONLY mytable
> ADD CONSTRAINT mytable_pkey PRIMARY KEY (column1);
>
> Column column3 contains zipped data (BLOB and toast table).
>
> The full dump of the table (pg_dump as SQL text) is 24GB.
>
> Thanks.
> Le jeudi 23 juillet 2020 à 15:24:54 UTC+2, Dave Cramer <davecramer@postgres.rocks> a écrit :
>
>
>
>
> On Thu, 23 Jul 2020 at 09:21, Christophe combet <chrisc_pro@yahoo.fr> wrote:
>
>
> autocommit=false & fetchSize=100 or 10000 : same exception.
>
> 42.2.10 still running.
>
>
> Is there a very large column in there
>
>
> Dave Cramer
> www.postgres.rocks
>
>
>
>
>
> Le jeudi 23 juillet 2020 à 14:30:56 UTC+2, Christophe combet <chrisc_pro@yahoo.fr> a écrit :
>
>
>
> Thank you for pointing the link between auto commit an buffering.
>
> autocommit=false: same exception.
>
> We try fetchSize.
>
> Thanks.
> Le jeudi 23 juillet 2020 à 13:47:14 UTC+2, Vladimir Sitnikov <sitnikov.vladimir@gmail.com> a écrit :
>
>
> >can you try switching to autocommit=false?
>
> + you probably need to configure fetchSize as well.
>
> Vladimir
>