And which manual solution are you using instead of the cascade thing? A
trigger created by yourself or a procedure that does the deletion first in
the child and then in the parent?

I created a before delete trigger that does the deletion. But it is
interesting to note that the table that was having all the deletions done to
it DIDN'T have a cascade index pointing to it, but just a standard FK
reference.


Unfortunately, Bill, you would have to be our guinea pig to discover the
real cause. Maybe it only happens with multiple users connected? I never was
able to reproduce it in the past.

There were two users connected to the database, with one of the
connections just idling most of the time (and not involved in the data
writing at all). I'd imagine that my problem was exacerbated by the fact
that I deal with fairly large databases, this test one was around 8 gigs.