DSXchange

As well as what Arnd said, resetting a job will roll back any uncommitted transactions, disconnect from data sources, close files and hashed files, and generally return the job to the state it was in prior to the last run request (as nearly as possible - transactions already committed remain committ...

Welcome aboard. If you installed DataStage using all defaults, pretty much everything is in one directory (for example /usr/Ascential/DataStage) so backing that up (while there is no DataStage activity) should suffice. This will change if you have created other directories on other file systems for ...

Unfortunately, using a File Pattern read assigns multiple reader processes that are reading simulaneously; it is this that is affecting your sort order and requiring you to re-sort the data; the rows from individual readers arrive in no particular order. You might like to try forcing the stage to ex...

Unless you have permission to remove them, you should not. These are part of the source data. Assuming, then, that you do have permission to remove them, my preference in a server job is to create a stage variable (called svCRLF for example) initialized to Char(13):Char(10) and to use a Convert() fu...

The answer is the same for server and parallel jobs. YOU need to specify, in the stage (format) or column properties, how NULL is to be represented in the file. After all, a text file does not have data types - it has only text, so that NULL is not a valid concept for text files.

The best partition method is the right one. For example, if you need to do something in the stage that requires key values to be on the same partition as all other instances of that value, then you need a key-based partitioning algorithm (for example hash or modulus). Otherwise, the best algorithm i...

Try typing in the name of the DB2 server. It may be that the drop-down list contains only those server names that you have already used.

Are you really running a 64-bit version of DataStage? If not, it (DataStage) might be reporting the wrong half (the high order bits) of the Oracle version number.

This warning is not related to the compiler - indeed, it is occurring during the pre-compilation phase and is alerting you to the fact that you are relying on an implicit conversion. That being the case, I tend to agree with Sainath's conclusion about a local message handler.

Have a look at a Table Definition record in the repository, a table that has a key. Go to the Layout tab and choose the Parallel option; you will see the table definition as a record schema and it will show you how to specify the key.

This information is not published anywhere.

Except on DSXchange, of course. But you will need to search.

Well done. "Could not connect to..." usually is a network issue.

chilli wrote:If it is -00005 then out put should be 0000A

Why?

chilli also wrote:and if it is 00009 then 0000X.

Why?
What other possibilities exist?
Please fully document the transformation rule.
Note also that your target data type can not be Decimal.

Not in a parallel job using a parallel Transformer stage, however.

The Key Management routines are suited only to server or BASIC Transformer stages.

1) The value "155" occurs with greater frequency in the data and therefore contributes less "information content" which is manifested as a lower agreement weight. 2) You are right. The value "ABCDEFG" occurs with lower frequency in the data and therefore contributes mor...