On Apr 18, 2006, at 2:57 PM, Tom Lane wrote:
> "Thomas F. O'Connell" <tfo@sitening.com> writes:
>> In the meantime, should I leave the database online while I attempt
>> to recover? It's still unclear to me whether postgres will restart
>> with invalid page headers. I certainly can't run pg_dumpall, for
>> instance.
>
> Has the database successfully checkpointed since the last error
> occurrence? If so, I wouldn't foresee any problem with a shutdown.
> If checkpoints are failing then you probably won't be able to shut
> down without getting the same error. (Use pg_controldata if you're
> not sure about when the last checkpoint happened.)
The last checkpoint was at Tue Apr 18 15:30:24 2006, so it looks like
we're fine on that front.
>> I'm glad that there have so far been only 3 relations affected... :(
>
> You mean, three that you know about :-(
True enough. The RAID is rebuilding with new drives right now. Then
we'll recover the relations known to be corrupt.
(If a pg_toast relation had an invalid page header, would restoring
the table to which it's connected solve the problem?)
Then I'll do a pg_dumpall. If that works, that should be a pretty
good indicator that the database is in a stable state, right?
--
Thomas F. O'Connell
Database Architecture and Programming
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)