Loss of inserted records during/after an insert

We have a system that records a data record for each cycle of a machine in a
n MS SQL Server database. These cycles take place approximately once every
10-12 seconds, and there are four stations on the machine, so we are writing
approx. 24 records per min
ute. Our database contains four tables, one for each machine station. Each
record contains a unique sequential number generated by the machine control
software. Data is logged using SQL INSERT scripts in the application (Wond
erware) that Operators use
to control the machine. (Wonderware script, BTW is not VBA, but is a proprie
tary scripting language.)
Everything works fine, UNTIL the one of the stations encounters an operation
al fault, and stops. This brings up a window on the control screen that req
uires the Operator to manually enter data, and an UPDATE statement is execut
ed to modify the last recor
d generated. Occasionally when this update is processed, a single record wi
ll be lost (never written) in one or more of the data tables.
At first we had all of the records going to one table. Thinking maybe the u
pdate for one station was somehow locking an index in the table, we separate
d the tables so that each station has its own table. Since the station is s
topped, no new record is ge
nerated for that station until after the update is processed. The other sta
tions can still be running, so they are generating INSERT commands, which co
uld coincide with the UPDATE command. Both commands use the same connection,
which is always open.
We still occasionally lose ONE record in one or more of the other tables whe
n the UPDATE executes.
Any thoughts?
Message posted via http://www.droptable.comUse the profiler and watch the sql statements - the most likely culprit is a
logic error within the application. Based on your narrative, I would guess
that the problem lies in the error-handling logic.
"Lee Drendall via droptable.com" <forum@.droptable.com> wrote in message
news:9981fa1e356140a298c4ffa13b629920@.SQ
droptable.com...
> We have a system that records a data record for each cycle of a machine in
an MS SQL Server database. These cycles take place approximately once every
10-12 seconds, and there are four stations on the machine, so we are writing
approx. 24 records per minute. Our database contains four tables, one for
each machine station. Each record contains a unique sequential number
generated by the machine control software. Data is logged using SQL INSERT
scripts in the application (Wonderware) that Operators use to control the
machine. (Wonderware script, BTW is not VBA, but is a proprietary scripting
language.)
> Everything works fine, UNTIL the one of the stations encounters an
operational fault, and stops. This brings up a window on the control screen
that requires the Operator to manually enter data, and an UPDATE statement
is executed to modify the last record generated. Occasionally when this
update is processed, a single record will be lost (never written) in one or
more of the data tables.
> At first we had all of the records going to one table. Thinking maybe the
update for one station was somehow locking an index in the table, we
separated the tables so that each station has its own table. Since the
station is stopped, no new record is generated for that station until after
the update is processed. The other stations can still be running, so they
are generating INSERT commands, which could coincide with the UPDATE
command. Both commands use the same connection, which is always open.
> We still occasionally lose ONE record in one or more of the other tables
when the UPDATE executes.
> Any thoughts?
> --
> Message posted via http://www.droptable.comsql