DKNY error

[Don Coleman]

> -----Original Message-----
> From: Jeff Harrison [mailto:jeffaharrison at yahoo.com]
> Sent: Monday, August 14, 2006 7:41 PM
> To: Don Coleman; 'filePro'
> Subject: Re: DKNY error
> 
> --- Don Coleman <dcoleman at dgcreact.com> wrote:
> 
> > For the past week or two the following table has
> > been causing DKNF errors on
> > one index in this file (E).  This table runs as the
> > second part of a script
> > where part one is an ascii import process.  As you
> > can see there is no
> > delete functions in this table yet the field index E
> > is built on (149) is
> > modified.  However, each night an archive process
> > runs on this file which
> > deletes 30,000-40,000 records from this file.  Total
> > number of records in
> > this file is approx. 70,000.  Each night via
> > scheduled tasks FREECHAIN.EXE
> > plus each index in this file is rebuilt.  Yet each
> > day we are having to have
> > each user exit fP and manually rebuild this one
> > index once or twice per day.
> > Does anyone see anything in this table that I am
> > missing which could be
> > causing the DKNY error?  fP v5.0.13, Windows 2000 &
> > XP Pro clients, WIN2000
> > Advanced Server.  TY,
> >
> >
> 
> Your code looks ok to me Don.  It should not be
> causing a DKNF error.  Also, the freechain program
> should not have anything to do with the indexes.  I'm
> wondering why you have each user rebuild the indexes?
> I would think that the indexes would only need to be
> rebuilt once after a DKNF error.
> 
> Are you sure the delete process works without any
> errors?  Are you rebuilding the indexes after the
> deletion process?
> 
> At any rate you should not be getting these errors -
> you may want to send the file and processing to fptech
> to see if they can duplicate this.
> 
> Good Luck.
> 
> Jeff Harrison
> jeffaharrison at yahoo.com
> 
> __________________________________________________
> Do You Yahoo!?
> Tired of spam?  Yahoo! Mail has the best spam protection around
> http://mail.yahoo.com

Jeff:

Thanks for the reply.  The index rebuild is only run by one person when the
error appears.  One big problem is the users are scattered across a rather
large building.  Ensuring that every fP user has exited the application is
problematic and undependable.  For example, yesterday I attempted to delete
the index before rebuilding it.  I couldn't delete it so I knew at least one
user was still in the file (or Windows once again didn't handle file locking
properly).  The IT Dept. checked access on the file and found 21 users still
accessing it.  No wonder.  They have now set up share permissions on that
file were they can bounce every one out.  However this is still problematic
because what is happening to the other indices they may be accessing/have
open when IT "exits" them.  

I run the freechain.exe nightly as part of the index rebuild because they
get automatic freechain rebuild messages as well.  Adding it to the nightly
script has reduced the occurrence of these.  My index rebuild runs before
the archive (delete).  It used to run after the archive but would frequently
error out during the archive process and stop.  Moving it before the archive
process mostly eliminated this.  Perhaps I should rebuild the indices after
the archive as well?  Any thoughts???  

BTW, my archive process is straightforward; the records to be archived are
selected via -v processing, I do a lookup free to a mirror file whose last
line is:
Copy file;write file;delete;end

Don Coleman
Donald G. Coleman, Consultant
402 Andrew Circle
Indiana, PA 15701
dcoleman at dgcreact.com
(724) 349-6302