DSXchange

I'm thinking "hashed field" rather than "hashed file" and that they are referencing the partitioning.

Post some examples of what your incoming date values look like.

One at a time or the entire project, those seem to be our only choices.

... or that.

That or make sure you filter out any "duplicates" first.

You can't use sqlldr to insert duplicate records, as you've found they must all be "new" for the bulk loader to work properly.

Hmmm... DATE datatypes don't have formats, people.

Check with your official support provider but I believe any 7.5.x can talk to any 7.5.x. I used to believe the same thing. 'Talk' is pretty generic and yes, they can talk but I was in the exact same situation back in the day and had... issues. When I called support, the answer I got surprised me - ...

Entries don't typically need to carry the db/user/password information as that's provided at runtime to any job that uses them via job parameters. And native stages don't use ODBC.

As noted.

So, in other words... there wasn't one? The log message should specifically say "From previous run..." and will add additional diagnostic information, otherwise we're all in the dark and pretty much just guessing.

No, I've used it back on a 7.5.1a system, so I'd say in the 7.5 release.

<Begin Standard Message>

Reset the aborted job and let us know the contents of a "From previous run..." log entry if one shows up when you reset it.

<End Standard Message>

"User variables" would need to be passed in as job parameters. You can't update job parameters, rather you use them in other derivations who's values can change.

No, it can't work with nulls, so as Ravi posted you need to handle those first.