Re: Learning: Freeing of Unused Records

On 26 May 2003 at 12:52, Michael Owens wrote:
> My questions is: when the update commits, and all transactions that depend on
> the version of these (old/updated) tuples all commit, could these tuples be
> automatically marked for reuse during the course of normal processing
> (without adding significant overhead)? --- rather than waiting for a vacuum?
> I am sure the short answer it 'No'.

Well, update is one of the special case. Dead tuples generated can be freed
using vacuum analyze which is not non-blocking and cheaper on resources.. so
you can use it every now and then..

HTH

Bye
 Shridhar

--
Captain's Log, star date 21:34.5...