On Fri, Jun 20, 2014 at 01:41:42PM -0400, Tom Lane wrote:
> Alvaro Herrera <alvherre@2ndquadrant.com> writes:
> > Bruce Momjian wrote:
> >> Why would we want this if the system functions fine without those files
> >> being created? I don't think we want pg_resetxlog to be doing anything
> >> except changing pg_controldata.
>
> > I don't understand why you say the system functions fine. If you move
> > the multixactid with pg_resetxlog to a point which doesn't have the
> > necessary file and page, the system will not start. Only pg_upgrade
> > knows to create the file appropriately, but normal operation doesn't.
>
> Yeah. In my mind, at least, there's been a TODO for some time for
> pg_resetxlog to make sure that pg_clog and the other SLRU-like files
> get populated appropriately when you tell it to change those counters.
> You have to have a segment file at the right place or startup fails.
>
> I'm not sure why Alvaro chose to fix this only for pg_multixact, but
> fixing that case is better than not fixing anything.
Well, if we are going to do this, we had better do all cases or the
behavior will be quite surprising, and when doing disaster recovery,
surprises are not good. I am a little worried that removing files as
part of pg_resetxlog might cause data to be lost as users try to figure
things out.
--
Bruce Momjian <bruce@momjian.us> http://momjian.us
EnterpriseDB http://enterprisedb.com
+ Everyone has their own god. +