Thanks for the suggestions.
For some strange reason all seems to be fine now. DB Workbench finds the
record
(and looks normal data) and works fine via ODBC as well. No idea what went
on there.

RBS

On Fri, Feb 3, 2012 at 6:43 PM, Thomas Steinmaurer <ts@...>wrote:

> **
>
>
> > Firebird 1.5 on Windows, accessed via ODBC.
> > Have a large table (some 50 million records) with the following
> structure:
> >
> > CREATE TABLE AUDIT_TRAIL
> > (
> > AUDIT_ID INTEGER NOT NULL,
> > AUDIT_TIME DATE,
> > USER_ID VARCHAR( 12) COLLATE NONE,
> > TABLE_ID SMALLINT,
> > ACTION SMALLINT,
> > PRIMARY_KEY VARCHAR( 128) COLLATE NONE,
> > DATA VARCHAR( 5000) COLLATE NONE,
> > PATIENT_ID INTEGER
> > );
> >
> > When I run this query via ODBC:
> >
> > select
> > *
> > from
> > audit_trail
> > where
> > audit_id>= 46262765
> >
> > The query will hang and I need to kill the app.
> >
> > I have figured out that the problem is with one record with AUDIT_ID
> 46270901.
> >
> > This suspicion is based on the following:
> >
> > select
> > *
> > from
> > audit_trail
> > where
> > (audit_id>= 46262765 and
> > audit_id<= 46270900) or
> > audit_id>= 46270902
> >
> > This will run fine.
> >
> > select
> > *
> > from
> > audit_trail
> > where
> > (audit_id>= 46262765 and
> > audit_id< 46270901) or
> > audit_id> 46270901
> >
> > This will hang/crash.
> >
> > If I run these queries in Workbench it will run fine, but the record
> > with AUDIT_ID 46270901 will not show.
> > In this table all AUDIT_ID's are consecutive, except for this 46270901
> > that is either missing or I think corrupt.
> >
> > Is this normal behaviour?
> > Could this record be fixed?
> > Thanks for any insight.
>
> - Have you run gfix to check for corruption?
> - What's happening when you fool the optimizer to not use the index on
> audit_id (I guess audit_id is the primary key column) by adding +0 to
> the field in the where clause.
>
> --
> With regards,
> Thomas Steinmaurer (^TS^)
> Firebird Technology Evangelist
>
> http://www.upscene.com/
> http://www.firebirdsql.org/en/firebird-foundation/
>
>


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