Say I have a table that I want to completely
repopulate. It may hold 1/2 to 1 million rows.
I'd like to directly populate it in the
database using the SQL "select into", but
don't want a logged transaction.
I can create the table with "not logged initially",
then when I'm ready to populate the table, at
the start of a transaction, issue the command
alter table MINE.MYTABLE activate not logged initially ;
I can then do the SQL INSERT INTO, and it will
put unlogged rows in the table. (And then I do
a COMMIT, of course.)
Should something go wrong in the insert statement,
the table is left in a bad state, and I'll get a
SQL1477N error in trying to access it. The documentation
that I've seen mentions that the table must be dropped
and recreated at this point.
My question is:
Is there a way to restore the table to an
normal state again, truncated, without having
to drop and recreate it?
I'm wondering about this because I see situations
in certain batch programs where use of the "not
logged initially" would make sense, except that
the batch programs are not allowed to drop and
create tables in our production environment.
If there were a command that I could use, I could
run it just before the alter table activate
not logged initially and SQL insert. This would
insure the table was always accessible regardless
of the status from a previous job.
-- Eldon Tucker
=====
To unsubscribe, send 'unsubscribe' to [EMAIL PROTECTED]
For other info (and scripts), see http://people.mn.mediaone.net/scottrmcleod
