Vlad, the large files have fb_sort prefix. But we also see a lot of small
fb_lock files.

Lester, we are a Windows-based shop and most of our customers have FB
databases are under 200MB. There are just a few that have 10-20GB databases.

Our customers need the ability to put those files on a separate disk and
restrict their cumulative size. TempDirectories setting would have been
perfect for it, but I can't get it to work with embedded Firebird.

If its is supposed to work in embedded mode then I will spend more time
making sure that I am not doing something wrong before taking any more of
your time.

Thanks,

Alec

On Sun, Jun 3, 2012 at 3:15 AM, Lester Caine <lester@...> wrote:

> **
>
>
> Alec Swan wrote:
> > We are working on tracking those queries down, but fixing them will
> require
> > an upgrade which the customer does not want to do. Moreover, their db
> size
> > is 12GB, so it may not be possible to optimize the queries a lot further.
> >
> > The customer also ran out of disk space trying to rebuild database
> indexes
> > which left firebird in a corrupted state. We had to add more space and
> run
> > gfix to repair it. This is a different problem, but it's also related to
> FB
> > disk usage.
> >
> > Is TempDirectories setting expected to work in embedded mode? What
> happens
> > when max allowed temp space is not big enough for a sort?
>
> Personally I would not be comfortable with the set-up you seem to be
> trying to
> run. 12Gb of database really needs it's own server rather than trying to
> run it
> with just a single embedded connection. I'd also be a lot happier having
> that on
> a Linux box which is just managing the database. Windows has it's own
> 'layers'
> of protection that keep files open while a program is running, so I
> suspect that
> it is windows that is retaining the disk space rather than Firebird. Until
> a
> program closes windows does not release some resources.
>
> But more important is to review the queries since a CTE approach may be
> able to
> apply filtering before you try sorting the extremely large results sets
> you seem
> to be working with. I have half million to one million record sets of data
> these
> days and have had to add extra indexes to allow some processing that was
> fast
> even on the slower machines from 10 years ago, but was getting slower.
>
> --
> Lester Caine - G8HFL
> -----------------------------
> Contact - http://lsces.co.uk/wiki/?page=contact
> L.S.Caine Electronic Services - http://lsces.co.uk
> EnquirySolve - http://enquirysolve.com/
> Model Engineers Digital Workshop - http://medw.co.uk//
> Firebird - http://www.firebirdsql.org/index.php
>
>
>


[Non-text portions of this message have been removed]