That's pretty darn old. As noted, I've seen similar things. Ask your DBA to install a more recent release, 9.2.0.6 or 9.2.0.8 seem to be stable for us. No guarantees but it just may do the trick.
If you do it in the Filter, there is no filename. You cat them to standard out and the stage reads it as a stream. However, since it requires a value, I would put something like '/dev/null' as the 'Filename'.
So... you've noticed that -logsum gives a 'summary' of the log and not the 'detail' from each log entry? Isn't the difference between the two obvious and exactly what you posted?
You could also try looking both options up in the documentation.
Yes, I would think you should be able to use the Incomplete Column option set to a none 'warning or error' option (like Replace) and then constrain your Transformer to only pass a row on where a 'key' field (say the first) is not null. Pretty sure I've done that in the past. Dont' worry about the ca...
I would still cat them. You could pass them thru something like 'sed' at the same time to strip the unwanted lines at the same time or simply skip processing them inside the job.
Before someone replies to the other post, you can delete it. The document you read and listed numbered questions from must have been written from a much lower level standpoint than DataStage - i.e. directly leveraging the ODBC or OCI protocols in C++, for instance so is not valid in this context. A...
No errors with no output means your expressions are 'valid' but don't match what is coming in from the xml. Can you post a sample of the xml you are processing and your corresponding xpath expressions?