DSXchange

I just designed a simple job having Source Oracle stage which has around 5 lakh records. A transformer stage and a target stage Dataset. blah blah blah Could you please let me know which is giving me best performance in this case. In your example, you simply allowed PX to run more processes to do t...

Or consider that you're able to ask questions and get answers from Premium Posters who used to work as consultants for Ardent/Informix/Ascential/IBM and have worked with the product since version 1.0. Or consider that the membership is helping to pay for a dedicated Webmaster to keep this site opera...

Some value isn't matching the appropriate data type and cannot be promoted, like a character string going into a numeric data type.

Without installing 3rd party software? You can't. You could use command line FTP or SCP to move files around or in your example get a directory browse.

Don't touch that. If you're just issuing DOS commands then your problem is 100% the operating system. If it's DSExecute that's failing, that a DataStage problem. I can't tell you where this message originates because you didn't post the exact message.

First, if you know the DATA.30 and OVER.30 are going into the 6-8GB size every time it runs, you should create the file manually using mkdbfile and a minimum modulus like 2000000 or something. The files may never grow, the thing to watch is the timestamp on the files. If they're changing, that means...

Don't use Unidata stage, that's totally something else. The hashed file must have been created first as a hashed file, not as a Universe table. DataStage is looking in an internal file for the table, and doesn't see it. If you use the Universe stage to create the hashed file, you can still reference...

You could always look at the last two characters and remove them if they are <CR><LF>. You could write a DS function that repeatedly strips off the last two characters of a string until the last two characters are not <CR><LF>. Something like (off the top of my head, please test): Text=Arg1 Done = @...

Sequential files, delimited by far. You'll have to process chronologically so that inserts precede updates, but that should be no big deal. In fact, you could take all of the inserts files and concatenate them together and load in one operation, then loop thru the updates.

Your options are really just to change your ETL that loads the tables to prepare insert and update load files. Triggers to audit tables are out if you're DIRECT path loading, plus, you'd have to write sophisticated triggers that only fire if they're in the background "mode" and not synchronizing. Ma...

Reset the job and see if more messages pop up, also, did you create the pipe first?

9 million is nothing like 900 million, you have some easier options. It's a shame you can't store load ready files from your ETL. You would simply load the background table, and when appropriate, move it into the foreground. Then, just take the same load files and apply it to the switched background...

$ORACLE_HOME is wherever the client is installed. bin is under that, so just try $ORACLE_HOME/bin/sqlplus

Sure, the Oracle bin is not in the dsenv file PATH environment variable. As Arnd so wisely states, fully qualify or update the path in your script, but better yet, update the dsenv file.

As long as the Routine is called from a Sequencer and not used in a derivation of a column, the sqlplus command line is the route. It bypasses all ODBC licensing and need to learn how to handle SQL calls in DS BASIC.