DSXchange

Ummm... that *is* the whole message.

Still not much in the way of more information.

If you Reset the aborted job, is there a 'From previous run...' message added to the log?

So, I assume you have to actually generate this data rather than load it from a source, yes? What have you tried so far? What kind of approach are you thinking of taking? Please tell us that your efforts so far have been more than to simply post here... The Parallel Job Developer Guide pdf has a sec...

The stage is metadata driven and thus really only meant to transfer flat files.

Without getting into the gory details of how the positional metadata works in a hashed file, just wanted to let you know you can get away with what you are doing because the job that clears the hashed file isn't writing any data to it. For a lookup I would clear the hashed file in the job that uses ...

Just want to clarify - there's no "record type" field or any way to know which rows are which type, it is strictly driven by record number?

Why do you have 1GB XML files? Are you cat'ing your files together to process them? If so, that's completely unnecessary and counter-productive as you've found out. The upper limit on file size seems to vary from system to system and perhaps version as well but seems to be in the 250-500MB range. Fo...

I wouldn't try to use the stage for a VSAM file since the stage is metadata driven. I'd stick with a plain old command line ftp or whatever tool your company may use for file transfers on/off the mainframe.

I would think the 'performance gain' comes when you use the Lookup Fileset in multiple jobs or steps in a job. Take the time to build it once and then leverage it many times.

Server job. Folder stage.

Define "slow". And have you determined where the bottleneck is?

For years though, directly writing to a cube wasn't in big demand because most of the cube vendors prefered defining and manipulating those things in their own tooling, so creating a clean .csv file or staging rdbms table with DataStage was enough. Yup, that's how I've had to do it in the past. As ...

As asked - I want to pass this to three outputs seperately from transformer without constraints - this must be done with constraints. They are the only thing stopping rows from going down all output links, Server or Parallel.

Very nice - and excellent formatting as well. :D

It might be your DB2 version being a little 'ahead of the curve'. Probably best to ping your official support provider and see what they say.