DSXchange

All rows go to one node. Hash means same values stay together on a node.

I don't know why the Dynamic stage can't read and write on a Unix platform, it's just ODBC. As for bulk loading, you have to have the command line utilities and since those are Windoze based, you need to manually handle calling it locally on the database server. Beware the marketing gimmick about to...

No, you can't delete jobs in any fashion outside of the Clients.

DSguru2B wrote:Are we invited

Sorry, paying customers only.

Of course, there's always IBM's conference this year in Anaheim.

HASH-->XFM--->SEQ means ALL rows output from the HASHed file to get to the XFM to be thrown away. Split the jobs if your logic is messy. As for the first row out of the hashed file, how do you qualify the first row? By order written into the file? Sorry, hashed data is randomized throughout the file...

What's your hardware setup? A single SMP box? Use glance, vmstat, memstat, top, prstat, whatever you can to monitor cpu, disk, and memory utilization. If something else is running, you're completing for resources.

Put on a big pot of coffee, we'll cover as much as we can.

Well, you made me laugh today. I wouldn't use the priceless, but thanks for the sentiments. Naked Fish, is that still open?

You need to know if the job threads are still active, because the status hasn't changed on the Director screen. Telnet to the DS server and login. Use DS.TOOLS to check the processes in a job and see if any are listed. These will have to be killed if the job doesn't die using Director. Then, you'll ...

This post is number 4000, so thanks. I've been answering fewer questions lately, trying to encourage others to pipe up first. I think folks are doing a pretty good job! We've got some superstars in the making.

It appears you need to correct set the file format in the stage as fixed-width. Then, prepare for misery as PX expects PERFECTLY datatyped data. You cannot have NULLs in non-nullable fields. You must edit each column's metadata and make sure all datatyping/nullability is handled. You must explicity ...

bcarlson wrote:you may need to break up your job.

Shocking! In some circles, you would be burned at the stake for such heresy. Don't worry, I'll be nearby to throw some water on you.

There's no metadata in a fixed-width file to describe the width of each column. If delimited DataStage can count the number of delimiters on the first line and guess at the number of columns. If you have the column names on the first line, that just confirms it.

Are you monitoring the resources on the database server and the DataStage server? Create a work table and DIRECT path bulk load into the table, then execute a MERGE statement.

Oh yes! I have the luxury of version 7.5.1, so View Data is not a problem with $PROJDEF. Yeah but it sucks if your desired directories are not the current project defaults. You have to keep overriding the $PROJDEF fr every stinking parameter if you can't change the project defaults. It's a pain if ...