At 07:01 PM 14/08/2006, you wrote:

>I am not sure if this topic got posted as I recived an email error
>
>--- In firebird-support@yahoogroups.com, "davidrevill69"
><datarun@...> wrote:
> >
> > We have been trying to run an automatic restore of our Firebird
> > Database using a batch file run as a scheduled task on a Win2003
> > server.With verbose output redirected to a file the restore runs
>until
> > just before the 'Finishing and going home message'.
> >
> > At this point FBserver becomes unresponsive (0% CPU in Task
>manager),
> > the guardian will not terminate it and it cannot be stopped in the
> > service control window. The only recovery is to kill the process in
>the
> > task manager.
> >
> > The database appears to have restored ok, but I do not want to
>trust it.

Does it do the last bit (which you think is a problem) if you run it
with the "no garbage collection" (-g) switch? If it doesn't, then I
suggest you either start doing a regular manual sweep (which can be a
nightly batch job) or else you let gbak finish doing its Good
Stuff. The benefit of daily sweeping would be that you can let your
backup focus just on backing up and ignore garbage.

In any case, it's not a great idea to try to kill the server when
it's busy doing GC. The garbage collector's work will get rolled
back and the *same* garbage, plus more, will be there next time a GC
task runs.

If it seemed not to be so in the past, it's possibly time you
reviewed your application's transaction handling. It's not a cosy
situation when you get to the point where garbage is accumulating
faster than the engine can clean it up.

btw, I thought I saw a similar reponse to this from Dmitry S. last week...

./heLen