DSXchange

Ofcouse
If i were you, i would run the job and in the second step run a small script that checks for 0 bytes in the file. If it returns TRUE, exit gracefully by adding a line in the job log using the dsjob command, else continue further processing.

Its not a third party component. Etlstats are jobs which do this kind of reporting for you. Kim has made that available for us. They are just jobs and basic code.

No. As far as i know.
Querying a hashed file via universe stage will be slow, especially if you have 7 keys defined. For lookup, you dont need to use a universe stage, you can just use the hashed file stage. Any particular reason why you opted to go for the universe stage ?

Ok i am glad there is a rule defined. Go for Ray's solution. See if it passes out all your tests. It should, Ray's solutions are pretty much all inclusive.

This is insane. This is like giving you a timestamp 04/05/06 and tell you to determine if its MM/DD/YY or DD/MM/YY. Next to impossible. If you can assure that the 'A' only shows up at the end of numeric values then you can use DIGITS() transform and get done with it by testing the first value and do...

Monitor the job from the director. And click on 'CPU utilitzation'. You should see four cpus.
Also, how do you know that only one cpu is being used?

That totally depends upon your job design. Each active stage will utilize a cpu. If you have
SEQ STAGE ----->XFM-------->HASHEDFILE

This has only one active stage and hence will use only one cpu.

You know what the experts say, "If you cannot find atleast two ways of achieving a task, you just havent thought hard enough"

If you tested it out and it works with partitioned databases then thats really good, as it cleared my misunderstanding too. I was thinking in terms of, data being passed in different channels to the database table. It was a blurry thing in my mind, thats why i advised and warned the OP about it in t...

You said the query is taking about an hour to finish in production env. Was that the same case in dev. and uat etc.? How many cpus do you have in your prod. box. You can probably create a multi-instance job which does selects and handles nvl in the transformer. Extract your data in chunks depending ...

Thats a nasty OCI API error as it retrieves a field that requires to be defined differently in DS. The annoying part is, its trial and error. ie, you have to try changing the sql type and the length. Good luck.

Its pretty obvious that column TypeA is of length 30 and type char. That means if you have "Alpha" as a value in TypeA, it will still have length of 30, whereas if you have the value "Alpha" in a colum varchar of length 30, its length will only be 5. Krazykoolrohit is right, trim the inputs and chan...

and the sqltype as char.

Forget all this duck and goose chase. Just request your dba to create two temp tables, load them with your data and do a database full outer join.