DSXchange

/.dshome is used if $DSHOME is not set. It therefore provides the "default" server to be used.

That's like asking for a spanner that only works for left-handers. The Execute Command activity behaves the way it does. That's how it was written. The only way you'll get that to change is via an enhancement request. In the mean time, you need a workaround, such as passing these arguments via envir...

Build the rows in Transformer stage then run them through an Aggregator stage set to deliver the Last row from each group.

What you said about hashed files applies equally to all database tables. Physical storage is irrelevant to the user. This is a fundamental tenet of database design; it permits the engine to use an optimal storage strategy. Hashed files are used by a number of different database products (such as Uni...

There is another possibility, particularly if you are on Solaris, where there is a hard limit on the number of subdirectories in a directory (from memory 32K). Looking at your job number (9949) and knowing that each job creates five or six subdirectories, some of which are dynamic hashed files, I wo...

For DataStage users the umask should be relatively open, for example 002. Definitely not 022, unless you want no-one else in the group to be able to edit things you create.

The data storage structure is called a hashed file because key values are "hashed" (passed through a hashing algorithm) to determine their physical location. A hash file is a tool for shaping a block of hash. There is no such thing as a hash file in DataStage. (There is also a "hash table". This is ...

Another argument for always maintaining table definitions in the Repository!

If you've specified hash partitioning on the same key, the Transformer input will partition the data, identically to how they were partitioned upstream. You would be better off specifying Auto or Same.

The DataStage macro DSJobStartTimestamp should be available in a Transformer stage. Or simply initialize a stage variable to CurrentDate() and don't derived it as each new row is processed.

Remove duplicates on both inputs.

You may have a file called /.dshome. I don't know whether the unmount/mount technique will work. DataStage internally does not may use of inodes except in the lock tables at run time or when a client is actually connected, so that's a positive for you. The main thing is that everything must be in ex...

Wrong, Arndo-san! With no C++ compiler any job that does not have a Transformer or Build stage will compile and run happily on 7.5x2.

There is no such thing as a hash file stage, therefore your question is moot.