DSXchange

Anupam, your post was a bit unclear about the current status. 1. How did you change the DS_JOBS hashed file (a program? TCL "update" command?)? 2. You need to make sure that all users are logged out when you do a reindex. Also, do your indices need rebuilding (do a LIST.INDEX DS_JOBS ALL and see if ...

Oddly enough, I am getting the same error at the moment on a new installation where the DB/2 database hasn't been set up correctly. You will get the same problem when doing a "db2 connect" from a shell and it is not a DataStage problem. Get your DB/2 people involved and once it works from UNIX then ...

That looks OK. I just looked at your query - are you sure that is the correct one, since it only returns 1 row of data?

Why do you think this is a problem with the controlling sequence? It will not generate any log messages while it waits for the job(s) that it started to return some status other than running. It continued once the XX job returned a problem status.

Put the dump score attribute in your job parameters, set the value to "true" and look at the dump score in your log file.

I just noticed that I posted a SERVER response in the EE forum. Ignore my previous post!

Matthieu's response is a good approach in a PX job.

Sounds like you have very many columns in your transform stage! You could look at the generated source code for your transform to find line 4300 (+- 5) and see what is being done. Or, if most of your column mappings straight-through derivations you can concentrate on those derivations that you have ...

These speeds also depend greatly on what order the incoming data is already in.

Do a reset of job XX and look at the message "from previous run" that gets created in the log file.

Please do not request "ASAP" help unless you are willing to pay for it. This is a volunteer site.

If that is all you need then a simple transform stage and not an aggregator is sufficient for you. Create a stage variable "GroupCount" with a derivation of "IF In.Col1=LastGroup THEN GroupCount += 1 ELSE GroupCount" and an initial value of 0 Creat another stage variable after that called "LastGroup...

the command is "dsjob -run" with options. You can call it up from the command line to get the options or use teh documentation.

With TOAD did you actually list all the data, or did the selection part just start returning data after 15 minutes? Another test you could try is to do the same user defined SQL in a server job which writes to /dev/null or passes through a transform stage that has a false constraint. How many nodes ...

What speed does your job run when you only have your Oracle stage and write to a dataset? What rate does your select statement run at from your favorite client access tool?

So your DataStage is on a UNIX machine, and you are writing to a text file on the UNIX machine and also viewing it there, yet you specified

NLS Setting MS1256-WIN2K-CS

for the file output?

I'm not at a DataStage machine right now, but think that the SDK has such a routine in there or at some time did have one. Making a COMP-3 out of an integer is not difficult, just a couple of lines of code in BASIC and use of the OCONV function and the "MB" conversion.