DSXchange

Your Char column IS padded, with ^@ (\x00 if you prefer that notation). These are invisible except with a hex editor. The pad character is set using APT_STRING_PADCHAR environment variable.

Is the partitioning correct?

If you've set "All rows = one transaction" then you would expect rows for 4, 5 and 6 to be rolled back (not quite the same as dropped, but the same effect). Use one row per transaction to cure that particular issue.

The second key is red because the GUI doesn't permit doing this. Even so, it will compile and run successfully if the key expression is syntactically valid. Right click and choose Validate Expression.

You can't connect to a UNIX server with Omit checked. RT_CONFIG1303/DATA.30 has incompatible header information and physical size. Restarting DataStage may well cure this, but not for the reason singhald imagines. The header information is in shared memory, and is hanging around (out of date) becaus...

Does your job design look like this? HashedFile1 HashedFile2 | ^ | | V | insert Source ---> TransformerStage -----> DB2Stage -----> update If not, it should. This design works. I've implemented it dozens of times. The same hashed file is referred to by both Hashed File stages, and all caching is dis...

Force closing the Adminstrator was not really a good plan. Connect to the server via telnet, get into a DataStage shell in the project and investigate. Re-indexing won't help; the RT_... tables are not indexed. Run UVFIXFILE over RT_CONFIG1303 to determine whether there is any internal corruption. R...

DataStage logs are database tables in the Repository. In the Administrator client you can set default log purge properties which can be overridden for individual jobs. Ordinarily you set the auto purge fairly tight, so that the log tables are kept reasonably small. In this case no other maintenance ...

I'm sure I mentioned that earlier. Oh well.

Click the More button to get full output from the compiler. Read it carefully; try to understand the cause. There is usually a lot of useful diagnostic information in the compiler output. If you don't understand what it is telling you, post the message here.

You may need to re-attach the named message handlers to each jobs for which they are specified. Have you set up project.msh as the default message handler for the project (in Administrator)?

If you have any local message handlers, these need to be moved separately. Search the forum for details.

Welcome aboard. :D

Without knowing a lot more about your source data it's impossible to say. For example, how many columns are in each source row? Maybe even a different stage type (such as Complex Flat File) can accomplish your normalization.

That's true in DOS as well - at least back to ftproot. You gave us no inkling in your original post that the file you wanted was not in a subdirectory of the ftproot directory.

Please mark this thread as Resolved.

Try Decimal(12,n) - your source number is unsigned, and DataStage can be finicky about having space for the sign.

Welcome aboard. :D

Please expand on what you mean by "log maintenance". A well set-up project will have had automatic log purging enabled from day one.