DSXchange

Hello and welcome to DSXchange! This is a problem that has been encountered and discussed in this forum numerous times in the past. This is the reason why it is recommended to always use the Search facility prior to posting a question. I used the keywords "mutex_unlock() - Not owner" and received 14...

The Float to String conversion can also be done implicitly, anytime your source is float and your target is string in a stage the engine will perform an implicit conversion for you. It is better to perform explicit conversions so that you know when and how numbers are being converted, but it isn't r...

or use an absolute path to your tr command in your job to see if that works.

Just as an info, I did a test run with with 5% of the data into a dataset and it was 17.661Gb large, the same data going to unbounded VarChar columns was only 6.109Gb on disk (120 columns, 80 Varchar with 1-50 lengths). The saving in disk space were great but runtime was even better on this I/O boun...

Thread Closed ---- see this duplicate thread

Did you specify a fully qualified name in your DB2 import?

Question 1 - Sorry, the lookup filesets won't allow extensions. You can use normal datasets and append to them, though. Performance will be similar unless the datasets are huge. Question 2 - PX won't allow anything like that. You can split the job into 2, but the data would need to be written to dis...

I just did a test - if I declare a dataset with one column of VarChar(1024) and fill it with just a single character per line, my test job of 500K rows generated a data set 263479K, whereas if I remove the length then the same number of rows only takes up 2752K. I wasn't aware of that change in beha...

I am going to have to check that up to see if using unbounded strings changes the DataSet behaviour on my system as well. If so, it is an interesting way of making sure that datasets use less space.

Does your job have the path to the "tr" executable set? Check the 1st line of the log file where it shows the environment and compare it to the environment setting for your terminal command session.

DataSets store the data in fixed width format internally - so "hello" stored in a VarChar(1000) field will use up 1002 bytes.

It isn't easy to determine which users are in which projects on a busy system with many users and projects. I sometimes use the device and inode to locate the physical file; but that is a lot of effort. Or you can look at the age of the processes to see what has been around a while. In addition, do ...

...but it is gving me the error..

What error message are you getting? Remember, your current working directory will be that of the DataStage project, you should always use fully qualified pathnames.

See this post for the answer to this duplicate post.

Your understanding is correct, and by that token a Hashed file stage (notice the difference in spelling, otherwise Ray will pounce on you) is a passive stage.