In this article I’d like to discuss how to add error handling for the new Job Executor and Transformation Executor steps in Pentaho Data Integration. A simple set up for demo: We use a Data Grid step and a Job Executor step for as the master transformation.
Let’s execute the master transformation now. Observe that everything seems to have run just fine: We get the nice green tick marks on our job entries:
The Execution Results show no errors:
But on inspecting the log we can see an error message:
Now that is interesting! So when there is a problem in the slave job, our Job Executor step just happily carries on … not quite what we want.
Luckily, the Job Executor step (as well as the related Transformation Executor step) has an optional error output, which will allow us to take further action:
- Let’s add a Dummy, Filter, Abort and another Dummy step to our master transformation.
- Connect the Job Executor step with the first Dummy step and choose This output will contain the execution results from the stream options.
- Set the Filter condition to
ExecutionNrErrors > 0.
- Point the
trueFilter output to the Abort step and the
falseoutput to the last Dummy step.
When we run the master transfromation now, we will see that the Abort step on the canvas gets a red highlight as well as the Abort entry in the log:
As we will be reusing this logic probably quite a lot, we can be efficient and store it in a dedicated transformation which we call via a Mapping (sub-transformation) step from the master transformation.
Our master transfromation looks like this now:
And our error-handling transformation looks like this:
We can reuse the error handling transformation now with any other Job Executor and Transformation Executor steps.
Note: For some strange reason it didn’t seem possible to connect the Job Executor step directly with the Mapping step, hence I kept the Dummy step still in the master transformation.