On Mon, Oct 19, 2009 at 11:21 AM, Alvaro Herrera
<alvherre@commandprompt.com> wrote:
> Gokulakannan Somasundaram escribió:
>
>> Actually this problem is present even in today's transaction id scenario and
>> the only way we avoid is by using freezing. Can we use a similar approach?
>> This freezing should mean that we are freezing the sub-transaction in order
>> to avoid the sub-transaction wrap around failure.
>
> This would mean we would have to go over the data inserted by the
> subtransaction and mark it as "subxact frozen". Some sort of sub-vacuum
> if you will (because it obviously needs to work inside a transaction).
> Doesn't sound real workable to me.
Especially because the XID consumed by the sub-transaction would still
be consumed, advancing the global XID counter. Reclaiming the XIDs
after the fact doesn't fix anything as far as I can see.
...Robert
