US and Worldwide: +1 (866) 660-7555
Results 1 to 5 of 5

Thread: Evaluation of multiple steps within a kettle job

  1. #1
    Join Date
    Feb 2011
    Posts
    11

    Default Evaluation of multiple steps within a kettle job

    I have a job that runs multiple sub-jobs in parallel.
    Upon completion of ALL sub-jobs I would like to start a follow up job.
    How can I do this in KETTLE. There seems to be no good wait to do that.

    I would also like to use similar evaluation in case of on or more of the sub-jobs failing - for example:
    - if job A fails but B and C succeeds - > do action1
    - if job B fails but C and A succeeds -> do action2
    etc

    anyway to do that kind of logic?

  2. #2
    Join Date
    Jun 2007
    Posts
    387

    Default

    Personally I would create a tmp file after a success execution (one per each job) and then all those jobs lead to a Check if files exist condition so with the first 3 jobs, it will fail, but when the 4th and final job finishes it will detect the 4 files exist and it will continue. See attached job.
    Attached Files Attached Files
    Rodrigo Haces
    TAM / Enterprise Architect
    Pentaho

  3. #3
    Join Date
    Feb 2011
    Posts
    11

    Default

    Thanks Rodrigo - that looks like a workable solution.

    Maybe next version of Kettle will have a more robust Job evaluation selection?

  4. #4
    Join Date
    Jun 2007
    Posts
    387

    Default

    Quote Originally Posted by jerzy_s View Post
    Maybe next version of Kettle will have a more robust Job evaluation selection?
    Let me tell you that this is a very robust evaluation, I have used it in very complex and critical ETL solutions. Anyhow there are a lot of other ways of doing this, there might be another way of doing it. Anyhow I recommend you that if you want a single step to detect if all the jobs pointing to it have finished ok, you should create a JIRA (http://jira.pentaho.com) case requesting it as a feature request.
    Rodrigo Haces
    TAM / Enterprise Architect
    Pentaho

  5. #5
    Join Date
    Nov 1999
    Posts
    9,535

    Default

    The solution is obviously simply to put all the parallel work in a sub-job.
    The parent job will wait until it's all done.
    Matt Casters, Chief Data Integration
    Pentaho, Open Source Business Intelligence
    http://www.pentaho.org -- mcasters@pentaho.org

    Author of the book Pentaho Kettle Solutions by Wiley. Also available as e-Book and on the Kindle reading applications (iPhone, iPad, Android, Kindle devices, ...)

    Join us on IRC server Freenode.net, channel ##pentaho

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •