DSXchange

The &PH& directory contains logging files, usually the stdout and stderr for background processes. The error message you posted does not affect your job run and can be ignored.

You cannot stop output to &PH& files, but you can occassionally delete the directory contents.

This is confusing - the cut-and-pasted text is from a windows installation.

What is the datatype of the EFFECTIVE_TIME and TO_TIME columns? Can you convert these explicitly to timestamp before the filter stage?

You cannot do this directly in one job using DB2 stages. It is better to have a sequence which calls first one job to get the max value, then passes that value to the second job which then performs the SQL.

DataStage will automagically repartition the DataSet from whatever number of nodes is on disk to the number required by the job at runtime.

If you activated $OSH_PRINT_SCHEMAS then you will have a log entry entitled "main_program: Schemas: (...)." and that will contain the schema desciriptions for each and every runtime link (n.b. that the optimizer might combine some stages)

Is the only thing that changed the DataStage version -> or is there a new database or DS machine? Just changing DS versions should not have this impact and I would look elsewhere first.

I misunderstood that and didn't realize that the error occurs after running for while. That would indicate ORACLE limitations of some sort - rollback space (what is your commit frequency?), table space or the like. Can you involve your DBA?

I haven't seen that error before.

Is the Oracle database partitioned and does your error occur if you run in a 1-node configuration (or make the Oracle read stage "sequential")?

If, from the designer, you try a "import table definitions -> odbc table definitions" from the menu do you get a list of DSNs?

ODBC to Oracle is slow. Why not use the enterprise stage, but don't do a LOAD - that should be faster. The correct solution is to get access so that the fast bulk load can be used.

OK, I get the point :D

Do you have other jobs which read from or write to Oracle and that are working (or could this be a general OCI problem)? Can you make a copy of the job that outputs to a PEEK stage, then take that first data row and manually perform the insert or update command with your preferred tool to see if tha...

Also, what method are you using to write the data? There really must be more to the error message(s) and you might try to reset the job and see if the entry "from previous run..." has more information.

Lookup stages do not have stage variables.