In the last couple of weeks, I've started having more and more issues with
floating point errors. I've NEVER had this error before, in any system I've
worked on (12 years and counting...). Why would it start all of the sudden?
Is there something in our data that's causing it to happen? I posted a
question with some DDL a couple of weeks ago but got sidetracked and never
followed up on that particular occurrence.
However, since I had a scheduled task fail last night, I think it's time to
try to understand why this is happening. The job that failed last night
gave me this error:
Executed as user: svc_sql. A floating point exception occurred in the user
process. Current transaction is canceled. [SQLSTATE 42000] (Error 3628)
[SQLSTATE 01000] (Error 7312). The step failed.
What I was trying to do is something that I do every night, and have for
several months. I restore our LIVE db, truncate several tables, shrink the
db, detach it, copy it to our DEV server, then re-attach it.
I'm baffled as to why it would have worked so well for so long, then failed
with a floating point error last night. We've been bit by this same bug
several times recently, both in scheduled tasks and also in proc calls.
Any assistance in understanding, and hopefully resolving this, is
appreciated.
Andre