Re: Lazy xid assignment V3

On 9/3/07, Alvaro Herrera <alvherre@commandprompt.com> wrote:
> Marko Kreen escribió:
> > On 9/3/07, Alvaro Herrera <alvherre@commandprompt.com> wrote:
> > > Florian G. Pflug wrote:
> > > > Since we didn't really reach an agreement on how xid_age should behave,
> > > > I've reverted it back to the original version. So with this patch,
> > > > xid_age will just force assignment of a xid.
> > >
> > > Is this really a good idea?  I'm repeating myself, but a query like
> > >
> > > select age(xmin) from bigtable
> > >
> > > could accelerate Xid wraparound.  If the server is running close to the
> > > limit it could cause a shutdown to prevent the actual wraparound.
> >
> > Such query would take only one xid, which should not be a problem?
>
> My guess is that it would execute age(xid) once per tuple?  Even if all
> the tuples had the same xmin, there's no cache therefore it would
> consume as many Xids as there are tuples.
>
> Am I missing something?

First age() assigns CurrentXid, rest reuse it.

--
marko