Has anyone ran across the following error: "WSOCK32.DLL The remote host has
aborted the connection (#10054) (#-1)" while using MS Access linked to an Image/SQL
database?  You normally see this when you "had" a connection established, and it
disappeared out from under the PC for some reason.  I've seen it when I've aborted
the ODBCLink/SE listener while a PC Client still had a connection open, but in this
case I'm stumped.  Here's the scenario:

.  I create a TI database, and attach it to a DBE.
.  In MS Access (95 or 97, both have the problem), I create a select query which
   gathers info from six separate tables (all linked via indexed columns).
.  I create a Link Table to my ODBC datasource, and select the target table
   which is the TI dataset.
.  I create an append query using the above select query as the source, and
   the linked table as the target.

  When I run the append query it processes for a little while and eventually returns
With the #10054 error.  While the query is processing, I see disk activity on the
3000, and SQLMON shows the active transaction.  When the query blows up,
the 3000 has a burst of activity, and then the ODBCLink/SE connection is gone
(I'm sure the disk activity is the rollback occurring).  If I try and rerun the query,
I get the #10054 error immediately (makes since, the PC-3000 link is gone).  If
I shutdown MS Access, and start it back up again, I can repeat the cycle.

It doesn't appear to be hitting any defined "timeouts" since it can run for a looooong
time if I turn on client side tracing (and watch it build multi-megabyte log files) before
getting the error.  When I look at the trace files, nothing appears wrong up until the
#10054 error shows.  On the host, nothing erroneous is being logged.  The last thing
shown is the INSERT.  I have bumped up the MS Access ODBC timeout values, and
added the CommandTimeout registry entry for ODBCLink/SE just in case, but it
doesn't appear to make any difference.

It does appear to be volume related in that if I limit the number of rows selected, the
exact same setup works fine without errors (I've had it work at around 600 rows,
but blow at around 700).  The rows are a little over 1K in length BTW.

I've checked the NMLG#### files, and nothing is being logged on the 3000.

I don't seem to have any "network type" errors with other processes (Samba, VT, etc.)
so I don't think it is a configuration issue on the 3K.. Unfortunately this #10054 error
seems to be a real generic one that isn't helping me get to the bottom of the problem
(oops, sorry but your connection to the datasource has gone "bye bye"!).  Anyone
have any ideas of where to look next?

Regards,
Michael L Gueterman
Easy Does It Technologies
Email: [log in to unmask]
http://www.editcorp.com
voice: (888) 858-EDIT -or- (509) 943-5108
fax:   (509) 946-1170