Hello all,

I am using SQL 2000 enterprise edition on a two machine
cluster. Everything is gigabit based network capacity.
The servers are dual processor, and over a gig of RAM
each. I am using a large disk array between the two
machines 0.13 TB capcity.

We are using a software package call CIMPLICITY to datalog
information from our production floor machinery. This
information comes very quickly, 15 SQL statements / .25
sec. Over 700 points of information are logged at a 5
second interval.

Twice a month the connection from CIMPLICITY fails and the
entire datalogging system reports a massive number of
connection failure attempts against the SQL databases. I
am getting nothing on the SQL side reporting these
rejected connection attempts that I can see. The windows
event log is empty as well.

The only way to get the datalogging system back online is
to restert the datalogging machine. Once this is done 90%
of the time the datalogging will continue as if nothing
happened.

Any thoughts?

One of my developers suggests that the ODBC connection can
be overwelmed with information and can reject futher
communications until a restart happens. We have perhaps 5-
7 open connections through for this machine in the SQL
enterprise manager. Each for a different location that
the datalogging machine is connected to. Can one of these
be fouled, which then causes a failure of them all?

Is this not making any sense to anyone else?

It sounds to me like the datalogging machine is catching a
busy signal from SQL and then forever after will reject
attempts to reuse that communication line. Remember this
only happens once every two months. The chances of
catching it in an analyzer is remote. I probably could
catch the after affects of the disconnection, but I havn't
been successful as of yet in getting anything.

Any more thoughts?

Thanks cjones