DSXchange

You can do this many ways. One method may be to select all of the primary keys in use in the table and write it to a hash file. Then, use a pre-generated file containing one row of all numbers from 1 to the max in your table. Use a DataStage job to read that file of numbers, referencing the hash fil...

It sounds like what you are doing is "filling in the gaps" in a surrogate key assignment sequence. Methinks you have a table who's primary key has reached it's maximum assignment value, so you have to find the unused numbers. If I'm right, you have quite a good little problem here. I'd turn the prob...

We figured out the problem. Since so many users were using this server we sometimes were trying to exceed the maximum number of files DataStage server can open at a time. The limit can be increased by increasing the Mfiles number. The other way to resolve this issue is to limit the number of users....

Whoa, did I ever get that one wrong...

Sorry folks, for some reason it stuck in my brain DYNAMIC hash files aren't resizable. I'm suffering from a severe lack of donuts...

Are you attempting a BETWEEN lookup using an ODBC/OCI stage? Or is your attempt to use hash files. Search this forum, as this topic is extensively covered.

Thread hijacker, DYNAMIC hash files do not use the resizing commands, only the static hashfiles. You must whack the file and recreate it. If it's important enough to save the data, then rename the hash file directory and recreate the new file under the old name. Then, use a job to copy the data from...

The worst problem is that 30K of data per row is going to bog down the processing. It will be like a pineapple going thru a garden hose. I GUARANTEE you will have terrible performance, this is too much data to juggle thru DataStage. Yes it will work, but what a performance penalty. This is simply a ...

Are you running a lot of jobs at that particular moment that are also creating hash files?

If it helps, unix has the command "basename" that will strip out the directory from a fully qualified filename.

Rats, no donut.

Uuugghh Diets suck. Watching me on the treadmill is like watching a walrus try to hoola hoop. Not pretty.

mmmm donut


My kingdom for a donut.

What was the problem? I'm not frustrated with you, just the problem. We need to know what you were doing wrong so that everyone following this post learns from your experiences. Plus, they will see the train of thought we try to take people thru in order to troubleshoot issues.

This is getting frustrating. I'm guessing you just pasted the format returned from the OCI/ODBC stage you're using, not what truly exists in the transformer at the time the math is happening. I asked you to write out the values from the transformer stage variables into a file and look at them there....

Why not create a temporary output file containing the results of each of your stage variables and run it for a couple of rows? The ICONV statement returns an integer number that equates to the number of days passed since Jan 1, 1968. So, for May 17, 1995 the number is 10000. If ICONV cannot "interna...