Data Integration / ETL | Forum profile

The Frog, That's what the second option is for (3.2.0GA) : Code: [ ] Output one row, concatenate errors with separator [_____] So the answer is that you get N rows for N validation errors per input row, unless you enable the option above.

Find a step that supports variables (the grey diamond with a red $ in it) and press CTRL-Space. That will show you a list of them. Most have very descriptive names. Other than that... Not that I know of. Feel free to write one and post it on the wiki

Quote: Originally Posted by gutlez Then I guess you're beyond where I can help you - without having access to your java program and the KTR files (and DBs that underly the java program), I can't try to diagnose any further. If you only got one row in the text File (excluding the header), then it *should* work. Sadly, should != is. hey np gutlez !! thanks for always trying to help, thanks for your time. Issue resolved. i'm a...

It has come up a couple of times (can we put it in the FAQ somehow?) - the basic problem is that you will need to have the key right next to the properties file itself. So if you need to protect the properties file your will have to equally protect the key. simple example? Imagine your live environment. There is a PDI installation, the application user 'app' has a .kettle/kettle.properties file with DB passwords and these things. Now...

or split it into two transformations....one transformation for storing output text file and another for sending email. Put both of them in a job..

Can some one give some tips/suggestions to improve performance. We have a transformation that supposed to load around 17000 rows from source(SQL Server) to target(Oracle). It is taking 90 minutes. by setting commit size to 10 K, it took 80 minutes. so changing the commit size is also not helping much. Tips/suggestions to improve performance would be appreciated.

Hi, I need to write a pretty complicated code such as loops on the columns on a file, calculations etc', that i don't think Kettle supports. I'm thinking of implementing it by reading a text file from javascript and do all the processing inside the javascript. is there a way to open a file for read/write from javascript? thanks

Hi Everyone, I am seeking clarity on an option in the Data Validation step. There is an option to 'report all errors, not only the first'. Seems pretty clear . My query is whether or not this is done on a per input row basis, ie/ each error for each row produces a new error (each validation), or do you get one error per row? I am just sketching out a design and wanted to know for sure. Any help appreciated. Cheers The Frog

regardless of this issue; I still dont feel i can use this in a production scenario because it'll break my whole transformation if there is malformed xml - there's no way i can handle it right? So i'll raise a bug of there's no work around?

What is the best transformation step to build a datawarehouse table with. Specifically, I need to add data each day to a datawarehouse table. Should I use Insert/Update? or Table Output? or............something else? and how do I make sure I am getting all the records since the last entry in the table? Thanks - Kinda Rookie, but you are going to save me a lot of time........