DSXchange

How much physical memory does the machine have?

Please advise where you found this term. Then we may be able to assist.

I expect you are running into a configured limit in SAS rather than a DataStage limitation.

Lacking sufficient information I am assuming your job design ends something like the following. -----> Transformer -----> Database | out | rej V RejectFile You use link variables on the rej link: out.SQLSTATE VarChar(12) out.DBMSCODE VarChar(12) out.LASTERR VarChar(2000)

Become fluent with UNIX scripting. It will stand you in good stead not just with DataStage but with all your interactions with UNIX systems.

If your script exits with a non-zero exit status, ExecSH will cause execution of the job to be halted. So that part of it's fairly automated.

Tell 'em from me...

According to the documentation, OCCURS DEPENDING ON is not supported

Did you try doing the conversion using a Modify stage? Was it a Transformer stage or a BASIC Transformer stage? A BASIC Transformer stage has a limit to the precision that can be represented, which may explain the rounding errors. Indeed, you can still get rounding errors with decimal[18,2] if, at s...

Search (Google?) for SQLSTATE = 01602. This is a message coming back from the database server; it is not a DataStage message.

Rewrite the query more efficiently?

Welcome aboard. It's bad practice to have your scratch disk resource on an NFS mount. This is because network latency can lead to exactly the problem you encountered - "stale NFS handle" is another way of saying that your connection is old. Always have scratch disk on local disk. If possible also ha...

Try using square brackets.

You can not resize VOC while anything is active in that project. You would need to SETFILE from a different project. Whence did you get this suggestion? I disagree heartily with it. While it is true that VOC can be a dynamic hashed file, there is no great gain, and higher risk of corruption. Resizin...

To do this from a job sequence on one server to run on a separate server you could run dsjob from an Execute Command activity. You only need to specify the foreign server name and authentication information once.

Perhaps that's just a "feature" of View Data ?

There may be no need to bring the separate streams together - database servers can support multiple connections. If they are all inserts, for example, with different keys, it doesn't really matter how many "clients" are doing the inserts - up to a certain limit, of course. This is one way to effect ...