Re: [HACKERS] Hot Standby utility and administrator functions

[Emmanuel Cecchet <manu(at)frogthinker(dot)org>]

Simon Riggs wrote:
> On Mon, 2008-10-27 at 11:42 -0400, Robert Treat wrote:
>
>   
> > On Monday 20 October 2008 05:25:29 Simon Riggs wrote:
> >     
> > > I'm looking to implement the following functions for Hot Standby, to
> > > allow those with administrative tools or management applications to have
> > > more control during recovery. Please let me know if other functions are
> > > required.
> > >
> > > What else do we need?
> > >
> > >       
> > Is it possible to give the master/slave knowledge about each other?
> >     
>
> Yes, but for what reason?
>
> The project I'm working on is Hot Standby, not streaming replication.
> That will link things together better than they are now, so I'd probably
> rather not prejudge/duplicate that.
>   
I think this could make sense in the case of a network partition (split 
brain).
If we need to perform a reconciliation after a partition we will need to 
know when a slave has switched to master and what transactions have been 
processed on each side.
Another use case is probably to implement failback once the former 
master has been restarted to just send the diff since failover happened.
Actually to complement pg_last_recovered_xact_xid(), 
pg_last_completed_xact_xid() or pg_last_recovered_xact_timestamp(), I 
would like to have something like pg_xact_xid_status(txid) that would 
return something of a xid_status type that can be completed, recovered 
or not_found with a timestamp that would only be meaningful if the 
status is recovered or completed.

Regards,
Emmanuel

--
Emmanuel Cecchet
FTO @ Frog Thinker 
Open Source Development & Consulting
--
Web: http://www.frogthinker.org
email: manu(at)frogthinker(dot)org
Skype: emmanuel_cecchet