DSXchange

Lots of ways. My preferred way is as follows. Create a server job to catch all four in a single string and load that into the job's user status area, and read that from the Job activity's $UserStatus variable and parse with Field functions. SELECT v1 || ';' || v2 || ';' || v3 || ';' || v4 FROM table...

When you use a Code tag to open, don't use a Quote tag to close - use a Code tag to close. Please go back and edit your post. Not sure about the legality of using different column names in the SELECT and GROUP BY clauses - are they defined as synonyms? You certainly have not specified aliases. I als...

Yes, by creating a job that reads it or using a command that dumps it, with the APT_CONFIG_FILE environment variable set to the same configuration file as the one that was used to create the Data Set. Or a compatible one.

Please post either the generated OSH or the score for this job. We need to verify that the record schemas are correct.

I usually prefer 0 rows per transaction (= commit after all rows loaded) when using OCI. That makes it "all or nothing" and easier to recover. Beware with the ORABULK stage - it is woefully inefficient (has been around since version 1.0). Use it to write the CTL file but use a Sequential File stage ...

Prefer
Now tell us which one is generating the problem.

(I dont know to put the diagram exactly, dataset is looked up with another dataset)lem Surround your diagram with Code tags and use Preview until you get it right. ##E TOPK 000000 12:25:57(027) <_PEEK_IDENT_> Input dataset does not have field: "AR_PRIM_PYMT_TP_DT"....." I tried to look the columns ...

Copy the query from the job log and paste it here, surrounded by code tags.

The first thing you have to do is the read the error message carefully. It tells you precisely what's wrong. Second thing is Search DSXchange - maybe someone has solved this particular problem in the past. Indeed, I'm fairly certain they have. If you do both of these things the remedy will be obvious.

Re-read my post then start building jobs. You may be aware of common processing requirements, which mean you can create re-usable components. But, without moving to Enterprise Edition (parallel jobs) you have no easy strategy for dynamic metadata. You could, of course, do the parsing using routines ...

Or you may be trying to run the same job multiple times either in the same job sequence or in concurrently running job sequences.

Tell us a bit more about the database stages. Are you using array processing? What are your transaction handling strategies? Are all the writes inserts or are there updates as well?

Keeping it real.

Clearly it is not "running forever" not least because you killed it. Column analysis is the most computationally and I/O intensive analyses that IA performs - you must allow it to finish. I have had column analyses run hours, even tens of hours. For large data sets, we run them over the weekend.

Help yourself. Search