DSXchange

Fewer than 20 at a time!

With 20 four-node jobs simultaneously you're generating over 200 processes:

20 conductor processes

80 section leader processes

80n player processes (one per stage)

Do you have more than 200 CPUs?

There's an appendix in the Advanced Developer's Guide, which ships as part of the standard manual set.

Saad, Please post examples of what you've got as source, what you expect to get as result, what you're actually getting as output and why that doesn't meet your expectations. Which kind of match process are you specifying? Also have you performed any investigation or conditioning (standardization) o...

The OCI stage only uses SQL. So, if you can specify how you'd use it in SQL, you can use it in the OCI stage. Presumably as a "before" command rather than every row.

And prepared to be embarrassed when the other user turns out to be yourself, in a minimized Designer, Manager or Director window that you'd forgotten you had opened!

Specifically, check that the data type if RNG_CODE is the same in DataStage as it is in UniVerse. This is the column name associated with the second parameter marker.

Always read everything in the error message.

DataStage is NOT UniVerse - hasn't been since version 5.2. Therefore, if you want to use UV/Net, you will need to install UniVerse and UV/Net on the DataStage server machine. But this isn't actually necessary. DataStage can use the UCI to access a remote UniVerse database. You need to set up a data ...

Sort (and hash partition) the data set based on the key column if it's not already thus sorted. Load up three separate stage variables depending on MOD(key,3) value. Use these values in the three separate non-key columns in the output of a Transformer stage. Run the result through an Aggregator stag...

I can't take credit for Arnd's work.
Indeed, this week I'm working on a Windows-only, server-only site (with ProfileStage and QualityStage).

UNDUP is a one-file match. Use a two-file match if there exists a one-to-one relationship between the rows in the two files, or use a GEOMATCH if the relationship is one-to-many.

OK, install the UniVerse ODBC driver (it's on the UniVerse clients CD), configure it, and away you go.

The dsdlockd.log file records the activity of the deadlock daemon. Every time it wakes, every time it checks for deadlocks, every time it cleans up after defunct processes - they're all recorded, and there's no auto-purge. Hence Mr Kim's problem.

You still don't need a driver. You need to specify the hostname (and maybe some additional information) in uvodbc.config where you define the UniVerse data source.

You have to create a script that puts its result onto stdout (to which the output link from the Wrapped stage is connected). Only in this way can you prevent a file from being created.

Are there always three records per group, or is the number variable? This will affect the solution.