DSXchange

This is an old problem. I don't remember the exact solution to the problem, but I think you need a new liborchsun4.o or the one you have needs to be symlinked to a different liborch. I'm sorry for not knowing the solution, but the good news is that if you call Ascential Support, they should be able ...

Nitin, For files ending with .osh, they are typically already "compiled". OSH is actually a scripting-like language that can be interpreted by the osh command. Typically, generated .osh files include an osh command followed by the OSH script, and these files can be run like any executable ...

Teej,

Thanks for your post. How are your lookups and joins holding up in 7.0?

- BP

In parallel extender, all datasets that are inputs to a single join must be partitioned and sorted on the same keys prior to the join. In Parallel Extender 6.0, I believe that partitioners and sorts are automatically inserted into the jobs, so that a novice PX user doesn't have to understand partito...

1. Do not rely on an extra transform to solve your update/insert timing issues. I recommend that you land your inserts to a dataset and then load that dataset into the database in a separate job. Use a sequencer to load the inserts after the updates. If you don't mind loading the database sequential...

LM, If I understand your question correctly, you want to know how to partition the non-DB2 data. You will want to use the DB2 partitioning method. I believe this will ensure that the partioning algorithm of your non-DB2 data matches that of the DB2 data. The partitioning methods are described in the...

Hi. You could build this sequence using a _sequential_ PX transformer stage. You can set a stage var to the alphabet string (e.g. "ABCDEFGH..") and use substrings to grab the letter that you need for your index. You would need two other stage vars to keep track of the indexes you pass to t...

Well, just to corroborate. There is no 'first' functionality in the PX aggregator. To get 'firsts', I've always written a very simple buildop that takes in sorted, grouped input and spits out the first record in each group. If performance is critical for you, buildop will probably be your best bet. ...

I believe that this sometimes also happens when the score file is too long.. Breaking up the score file might be the next step for you if the previous suggestions do not work.

I know this is a lot to ask but:

1. Could you post the score file/config file?
2. Could you post how much data is in each lookup?

Thanks,
BP

T.J, Firstly, PX used to automatically insert hashes and sorts into PX jobs. This was done to make it easier for people with no parallel experience to write performant parallel jobs. I'm not sure if PX still does this, but I recommend that experienced PX users turn off the automatic insertion of has...

T.J.,

Thanks for the clarification regarding Oracle and the APT_RDBMS_COMMIT_ROWS option.

Perhaps Ascential should rename the evvironment variable to APT_DB2_COMMIT_ROWS to clear up any confusion.

- B.P.

My two cents.. PX uses the C compiler to compile the transform stage and 'new parallel' stages (i.e. buildop). If you don't use the transformer or buildop stage in your PX jobs, I don't think PX needs the compiler to build the job, but PX might still insist on having one there anyway. If you run int...

Hi Tunde.

Can you tell us what your error is? Also, what is your runtime PATH and LIBPATH set to?

Thanks,
BP

Hkotze, The enviroment variable APT_RDBMS_COMMIT_ROWS can be set to a number equal to your desired commit rate. If you set it to 0, then PX will issue a single commit for all of the records being processed. I am fairly certain that this env var works with DB2, but I am not sure if it works with othe...