Hi,

At September 27, 2004, 07:53, tmiravet wrote:

> Here is the result of gstat. I did it this morning. The size of .gdb
> file was 2GB, last friday was only 20M
>
> -----------------------------------------------------------
>
> Database "grader.gdb"
>
> Database header page information:
> Flags 0
> Checksum 12345
> Generation 1151
> Page size 4096
> ODS version 10.0
> Oldest transaction 1079
> Oldest active 1080
> Oldest snapshot 1042
> Next transaction 1145
> Bumped transaction 1
> Next header page 0
> Database dialect 3
> Creation date Sep 25, 2004 8:26:23
> Attributes
>
> Variable header data:
> Sweep interval: 20000
> *END*
>
> In your mail you mentioned the need of sweep to free up space
> occupied by deleted record. Well, some days before the problem was
> detected, there was a deletion of around 67000 record in a table.
> Usually no deletions are done in the database, they are done only to
> clean historical logs. I think, that was the cause of the problem.
> The system is running about 1 year ago and never detected something
> similar before. So, this morning I did a manually sweep, it took
> around 4 minutes -in a test database was only a few seconds- and
> after that, it seemed that the database was not growing.

The time that the sweep will take is proportional to the garbage that
it has to eliminate and free up the corresponding space.

>
> Do you think I should change the sweep interval, or just keep that in
> mind and do manually sweeps from time to time?

One thing that you might want to do is to schedule a manual sweep to
be run during an idle time. So, this way, the sweep is performed on a
regular basis without someone having to think about it, and would be
run at a time that would affect the minimum of users(or none at all).

--
Best regards,
Daniel Rail
Senior System Engineer
ACCRA Group Inc. (www.accra.ca)
ACCRA Med Software Inc. (www.filopto.com)