On Mar 3, 2006, at 12:00 PM, Simon Riggs wrote:
> On Fri, 2006-03-03 at 10:23 +0000, Andy Shellam wrote:
>
>> Then you would have to find a method of telling the script
>> that you wish to bring the database up and it will exit and allow
>> PGSQL to
>> come up at the current state with the latest data.
>
>> Replication sounds like a better option
>
> If you want auto failover, then you will need a package that detects
> that accurately and can initiate any required actions, whatever method
> you choose of keeping multiple systems in sync.
>
> IMHO auto failover needs *extensive* testing and failure analysis to
> make it more robust and practically useful than manual failover, so is
> not quite the magic bullet it may appear.
>
> Best Regards, Simon Riggs
Auto failover is not a requirement for me. My primary goals are to
reduce the need for pg_dump to occur nightly against a large
production database and to reduce the window of potential data loss
in a failover scenario, where the assumption is that the failover
would be managed manually.
There's too much application-inspired DDL and too many tables (as a
result of an inheritance-heavy) data model for Slony to be a viable
solution in the short term.
--
Thomas F. O'Connell
Database Architecture and Programming
Co-Founder
Sitening, LLC
http://www.sitening.com/
3004 B Poston Avenue
Nashville, TN 37203-1314
615-260-0005 (cell)
615-469-5150 (office)
615-469-5151 (fax)