DSXchange

Cache is always FIFO (first in, first out). Even though this is per node on Oracle RAC, your fears are unwarranted.

I did ask for information on how these were set. There's also a recent thread from U on exactly this question.

I performed a test a couple of days ago - scheduled a job from Director and examined the AT queue. There was nothing in the AT queue, but the job executed on time. I deduce from that that the Information Server scheduler service is being used (at least in version 8.5 FP1 on Windows Server).

Funnel stage must have two or more inputs. You can use a Copy stage with one input.

Then the original poster can mark your thread as Resolved using the green button at the top.

Yes, the name of the environment variable is APT_STRING_PADCHAR, not APT_STRING_PAD. This indicates that any Char data type will be right-padded with 0x0 characters.

I don't see any representation of the commas in the FD. Try 01 EMPLOYEE-RECORD. 05 EMP-RECORD-LENGTH PIC 9(4). 05 COMMA-1 PIC X. 05 EMP-HIRE-DATE. 05 COMMA-2 PIC X. 10 EMP-HIRE-MM PIC 99. 10 EMP-HIRE-DD PIC 99. 10 EMP-HIRE-YEAR PIC 9999. 05 COMMA-3 PIC X. 05 EMP-SALARY PIC 9(9). 05 COMMA-4 PIC X. 05...

VS2008 is supported for DataStage version 8.5.

You still have not advised the settings of the four project environment variables in the Compiler folder, as requested.

And this is related to the problem reported by xch2005 how?

Not really. Try replacing the Lookup stage with a new Lookup stage (for test purposes).

Octal 376 is decimal 254. This is the "field mark" character which can be referenced via the Char() function but, to avoid recalculating the Char() function unnecessarily, also through the @FM system variable. This is the character to which line terminators are converted when reading.

I don't have access to a system right now, but does not dscmdexport or dsxport have the /XML switch?

In the "Job starting" event in the log, verify that the environment variable job parameter has a valid value.

Could it be that the mainframe expects "masked decimals" in which the metadata (for example PIC 9999V99) tells it where the decimal placeholder needs to be?

Looks like 64GB isn't enough to process that many rows, particularly if you're making them all into a single transaction.

How are the resources on the DataStage engine machine?