DSXchange

As a server job, there will almost certainly be an entry in the log file after resetting which will show the termination cause and address. Are you certain that those are the only 2 log entries generated by the "reset", and did you reset from an aborted state?

A quick look at V8 shows that you can drop and create ODBC connection tables; I am pretty certain that is built in at V7. What stage are you using to write to SQL Server?

Do a "reset" of the job, and post the log entry that starts with "from previous run..."

There are hidden complexities involved in putting together such information; particularly when you have multiple projects and multi-instance jobs. Mr. Kim Duke has written EtlStats which is available freely and gathers such data as you require.

I don't know how one could, with surety and without risk, physically migrate the XMETA storage space. I gather you are not increasing capacity on one machine but are migrating from one machine to another. In that case it is safer and cleaner to create a new and empty XMETA and use the DataStage Expo...

I'm a bit confused, as PICture clauses are COBOL related and the COMP-3 data type is only relevant for seqeuential file I/O. When processing a database entry you would just define this as numeric with 11 spaces with an assumed 2 decimal places.

There are several steps to a multiple job compile, which step does this occur at? Does it happen regardless of which job or jobs you try to compile?

Yes, it is possible to drop and create tables - in principle.
The question is what kind of tables using what kind of database access?

You can specify whether or not the data is already sorted or grouped by a column or columns as well as specifiying the sort method and memory usage when using an explicit sort stage, whereas those options are not available at the stage sort level.

Since you are on Windows, using <cr><lf> would be more correct, unless your file is actually in UNIX format.

From your Windows DS Server, try the DOS command "telnet localhost", what output do you get?

from UNIX, what does the command return?

Do you have other workstations from which to try to connect, perhaps it is just a problem on your PC. Have you checked on the server to see if all the Windows processes are running (there is a tab in the System menu, but I don't recall what it is called)?

From your description it isn't quite clear what you are doing. Do you have 1 row with over 4Mb, or a group of rows totalling over 4Mb?

...All the data put to queue as single message...

, then

...split it in say 1000 records / file...

Manager (which doesn't exist at V8)? Do you have the MCM installed and are using it - perhaps the Designer just has the incorrect version? Can you connect to the project from your workstation using another user, and can you connect to another project using your user?