DSXchange

Yes, that should work just fine. Is your XML Input stage setup to take that filename column as the URL/File path to read the files from?

Just to be pendantic, that didn't "install" anything. You simply set NLS_LANG globally for the server. Also note you could have set this up as a Project level environment variable and (if needed) you can still override it at the individual job level.

Sounds to me like something to take to your official support provider. I seriously doubt NLS anything controls that.

I was thinking you could use the string "DATE RUN" to indentify the start of each report but that's only true for two of them. A line that starts with "TIME :" looked like another candidate but it's only good for three of the four. Perhaps look for the actual report name you need...

Are you looking on your client pc?

battaliou wrote:Take a look at the UNIX tr -s '\n\' ',' command.

It's also simple enough to convert this in job, as Ray noted, if desired.

India2000 wrote:it would be easier track issues with files if I use this method

Why? What would be easier about it? The stage can import the current filename along with the data, if that is your concern. To me, this seems like more work for zero gain. [shrug]

Can you not leverage something like gunzip to std out in the Filter of the Sequential File stage?

Still looking to understand what "load" means here, so simply repeating the word doesn't help. What I specifically mean is load how - what do your target tables look like, do you have to pivot rows to columns, that kind of thing. Or do you simply take each line as a string and stick that i...

That viewer wouldn't help when their server is on a UNIX box.

I would imagine one could find the dsrpcd process and its start time would be the start time of "the services":
ps -ef | grep dsrpcd

Ask your DBA to trace the process in the database, see what (if anything) that it is doing.

Why not simply use the File Pattern capability of the Sequential File stage and be done with this? Is there something in what you are doing that requires you to process each file one at a time?

As I said, typically it would be right but not always, depends on the job design and also if buffering is involved. Would be curious what one of the Grognards think...

Use the various dsjob options to:

1. Find all projects on the server
2. Find all jobs in a project
3. Pull the most recent (or full) logs from each job
4. Lastly, search for your target string.

Loops within loops.

Can you post a (small) representative example of your input file? Also, clarify exactly what "load a report" means here.