> It would seem about that the only currently viable technique for creating
> a standby database in case of a primary database failure is to ship output
> from pg_dump(all) to an awaiting postgres instance and reload them.
>
> I'm currently working on a few scripts to accomplish this which use cron
> to schedule the backup which then gets shipped and reloaded.
>
> This is not efficient since since the desired frequency of backups will
> often be greater than the frequency of data changes. However, I can't
> know when exactly a user will change data, so I was looking for a better
> way.
>
> Is there a way to determine if a particular database has changed, and
> therefore needs to be backed up?
>
>
> Thanks,
> Bryan
>