On Mon, 15 Mar 2004 08:26:14 +0100 Edwin Pauli <edwin@epauli.dyndns.org> sat down, thought long and
then wrote:
> Very trange is that the server will start without any problem when i
> copy a 3 weeks old backup to /usr/local/pgsql/
>
Seems we have been hunting the wrong bugs in the beginning. Could it be possible that there was a
major version update of the database engine (7.2.x -> 7.4.x for example), without a database reload?
I have noticed several new "features" for queries, I think, 7.4 is much closer to SQL standards than
7.3, which, in turn, is closer than 7.2, especially regarding the necessity of type casts. In one of
our applications we had to rewrite most of the queries, because they did not work properly with
newer PostgreSQL releases. So that could have caused the PHP-Errors, and the restart of the database
without a dump/restore could then cause the start up failure. Just a theory.
Well, if there was no database engine update, this theory is rubbish, of course. It could also be a
simple coincidence, that there were query errors and server problems at the same time.
Did you compare the actual binaries and the binaries of the backup? Are they identical?
Regards, Frank.