Markus Ostenried wrote:

> On 12/20/06, heineferreira <heineferreira@...> wrote:
>> I haven't had a corrupt firebird database....
>
> Other reasons include faulty RAM, defunct disk drives, and running out
> of disk space.

Bad RAM possibly, dead disk drives certainly, but running out
of disk space will not be a problem if you have forced writes
on. Forced writes causes pages to be written in careful order
so each write leaves the database in a consistent state.

> One of the most nasty ways would be to hit the file
> size limit - most likely the 2gb/4gb limit on a 32bit file system

That isn't a problem on any Firebird version. We use 64bit file
I/O. InterBase 5.6 would happily write over the front of the file,
but it's eight years old and should be retired.

>
> That being said I never had a corrupt db even though my customers are
> quite careless and a hardware reset and the occasional power loss
> happened more often than I liked. Just be sure to have the Forced
> Writes setting enabled on windows machines. Normally all you would
> lose then are the changes of the transactions that were not committed
> at the time of the failure.

Which is as it should be. Committed changes are preserved, but
an interrupted uncommitted transaction fails and its changes are
rolled back.

Regards,


Ann