DSXchange

Make your sequence job restartable. Everytime you start your sequence job, check for the "Abort" message. If its other than "Job aborted after x warnings" then start the sequence. It will automatically run from the point of failure. If the error message is something else, reset the sequence job and ...

A 32 bit hashed file (default) will hold upto 2.2 Gigs of data. If you need to store more, then you need to create a 64 bit hashed file. Performance can be tweaked/increased by using different methods. One is to create multiple instance jobs and select a set of keys to be stored in each hashed file ...

Check your target table again. Make sure all columns are present.

If you have the info in afile, you can certainly write a small routine to type the value of the file, capture its return and send it back [using DSExecute()]. Call this routine in the job parameter's area,the return value of the routine will then be populated in the job parameter. I think thats what...

Forget about the error in Red. Its not an error. The one in blue and green means that your source and target datatypes are not identical. If they are not, you need to explicitly do datatype conversions, using either the transformer functions or modify stage. Do an "search on all terms" on keywords, ...

Use the Macro DSJobStartTimestamp.

WHere does your repository sit? As in, what database?
It seems like there is something else holding the block that you are trying to access. Get your dba to monitor what exactly happens when you try to save.

Well, check for IsNotNull() for each column and if true, take that value. Something like

OPEN/CLOSE commands get executed only once before or after the sql body, respectively.
Open your enterprise stage, go to the 'properties' tab. Its under 'Options Category'

If you go into the parallel transformer, expand the derivation, right click and go to "Functions". There are different functions for Strings like Convert(), Field(), INDEX() which are present in parallel jobs. But FMT and ICONV/OCONV are not. IF you can tell us what you are trying to achieve, someon...

What if you chop off the milliseconds, does it work then?

What happens when you send it directly to oracle without any conversion? Do you get any errors/warnings? If yes, can you paste it here?

What is the rule of conversion? In your first two examples, if you divide your input with 10, you will get your desired output. But your third example needs division by 1000.
You can use functions like StringToDecimal() to get your data converted.

If your DB2 timestamp is in YYYY-MM-DD 24HH:MI:SS then no conversion is required. You can send it as it is.

Use the OPEN command to pass a delete with the where predicate that contains those ids.