Archives of the TeradataForum
Message Posted: Mon, 19 Jul 1999 @ 16:18:04 GMT
Have other sites seen this problem at V2R3? Has it been reported to NCR?
Yes, we ran into the same problem, usually with dates. It is an easy fix, but very aggravating especially since it is not consistent across other products.
We, at Cincinnati Bell Telephone, have upgraded to V2R3 this weekend and came across a Multiload error that we did not expect. I'm sending this as a heads-up to anyone who hasn't migrated yet and wondering if anyone who has already upgraded ran into the same problem.
---------- Forwarded by Amy Ferguson/CBT on 07/19/99 09:18 AM
I checked all of the FastLoads and MultiLoads that have run since the upgrade, and I also ran my own controlled tests to isolate the problem. My findings follow.
All FORMAT clauses seem to work fine in BTEQ.
All FORMAT clauses seem to work fine in FastLoad.
Date FORMAT clauses seem to work fine in MultiLoad.
This type of Numeric FORMAT clause no longer works in MultiLoad.
columnname = :columnname (FORMAT '999V99')
Now it is apparently required to include the exact data type and length and precision of the column being populated (don't ask me why).
If this column is defined in the DDL as, say, DECIMAL(6,2), then the syntax must be this:
columnname = :columnname (DECIMAL(6,2),FORMAT '999V99')
Seems redundant and unnecessary, but at least this appears to be a workaround.
I found a FastLoad example that is almost identical to the MultiLoad example above, and the FastLoad works.
I have restarted one production job with this workaround, with success. I'll do the rest after sending this note. I don't see me trying to catch all of these potential problems before they hit the fan, as it seems enough work for a mini project. The Multiload source would have to be scanned for all Numeric FORMAT statements, and for each one found, a SHOW TABLE will have to be done to determine what Numeric data type and length and precision to put in the FORMAT clause. So for now anyway, I'll just fix them as they happen. Perhaps a software patch will be in place before we have to change too many scripts (unless they don't consider this a software problem, although FastLoad and MultiLoad treating this situation differently seems odd).
|Copyright 2016 - All Rights Reserved|
|Last Modified: 27 Dec 2016|