DSXchange

You could use an execute command activity to update the content of the values file utilizing your favorite scripting language perl/sed/awk before invoking your job from a job activity.

Mike

I've done a pretty extensive amount of testing of the surrogate key generator using flat file state files over the years. The behavior that the OP observed is perfectly normal and most likely not a problem. As long as all of the key values in the table were generated from the same surrogate key stat...

You have no problem. That is the nature of parallel key generation. Each processing node acquires a block of keys to use. When the block is used up, a new block will be acquired. Surrogate keys are only guaranteed to be unique. Gaps are perfectly fine.

Mike

Also check the permissions at every level of /fullpath.

Mike

That's pretty straight forward these days with a couple of parallel transform functions: one to subtract a month and another to get the number of days in a month. What I don't remember though... in what version was the DaysInMonth function added? It's relatively new and I don't remember if it existe...

I will assume that you imported the MDMQS example jobs and ISD project that were bundled with MDM. Back when I did this on version 9.1, I had to edit the ISD project before importing it. It referenced an IBM host that I had to replace with our host. See if that might still be the case in your versio...

Use the join stage with a full outer join. You will have access to all of the columns from the left link as well as all of the columns from the right link. Keep the ones that you want.

Mike

Activities in a job sequence are executed sequentially. You will see that if you investigate the generated BASIC code. A job activity starts a job without waiting for it to finish. That is why you obtain the parallel execution of your commands by placing them inside a job sequence and executing them...

Also consider what NLS character set is used.

Are all characters a fixed number of bytes?

I believe the Len() function will be counting bytes rather than characters.

Mike

You can use NextSurrogateKey() as many times as you'd like in a single transformer.

But you can only define a single surrogate key state file per transformer.

Mike

You've written something that is not a valid date into a date field in your dataset.

You need to look upstream from the dataset to determine where the invalid date originates.

Mike

My current project is not on the latest 11.x fix pack, so maybe that issue has been fixed.

Mike

Actually, I've seen the warning Joshi is referencing several times on my current version 11.3 project... it might be new behavior since I don't ever remember encountering that before. Your options are to include an extra dummy record in the reference input that will never match anything or use a mes...

Change capture stage needs unique keys on both inputs. As noted by qt_ky, it is a simple 2-file match process on ordered inputs... once a record is consumed on an input it is no longer available for the next key comparison.

Mike

StringToDecimal is an explicit type conversion function for the transformer stage, so it won't generate a warning in going from varchar(10) to decimal(10,0) if your source string is nothing but numeric digits. Did you isolate a specific example row that causes a warning? Using NOWARN is likely maski...