Re: SQL*Loader bad file

Are you sure the bad file did not already exist and that it is only the write permissions on the file that are causing the trouble ?

Joyce A Moore wrote in message ...
>I'm finding that sqlldr sometimes can't write bad records to the bad
>file and errors off with a message in the log file something like:
>
>SQL*Loader-523 error: -2 writing to file (bad file name)
>
>The bad file always gets created, but the rejected records aren't
>always added.
>
>This isn't a simple sqlldr process. Two tables are involved using
>"when", a uniqueness constraint on the header table, foreign keys in
>the detail referring to the header, triggers on both of them, and calls
>to an Oracle Project Accounting (with the Grants Management System
>patch) validation routine from the trigger. In short, there are many
>ways for a record to fail.
>
>Simple format errors usually end up in the bad file and allow the log
>to finish. If the header fails for any reason, it will end up in the
>bad file and the log file finishes. By "finish", I mean it doesn't
>print the above error and abort. If a foreign key or a trigger error
>occurs, the error message ends the log file. Maybe this is unnecessary
>information, but since the nature of the error is the only variable, I
>thought I'd include it.
>
>Any help is really appreciated.
>
>Joyce Moore
>-------------------------------------------------
>Administrative Computing and Information Services
>Carnegie Mellon University
>
>
>
Received on Fri Nov 19 1999 - 01:31:39 CET