Dataiku Application Test Instance

emilieg Dataiku DSS Core Designer, Dataiku DSS ML Practitioner, Dataiku DSS Adv Designer, Registered Posts: 1


I'm running some tests of my dataiku applications and I encounter the following problem. I have several sql recipes in the middle of my flow, using "intermediate" BigQuery tables, created by the flow. When I look at the associated scenario run details, it seems that when running a test, it creates additionnal BigQuery tables which have the original name "enriched" with a unqiue id changing at each new test. My problem is that when looking at run details of my sql recipes, it seems that the sql query executed don't update the table names I use in the recipes, in order to use the bigquery tables specific to the test instance, with the dedicated id suffix (instead of the global one). The global impact of such a behavior is that the test instance executes a flow that uses a mix of global tables and test instances specific tables. Unfortunately, I can't just replace my sql recipe by a "built-in " dataiku recipe. If I could, I think this would solve the problem. Is there any solution other than that ? Thanks in advance for your help on that


  • Zach
    Zach Dataiker, Dataiku DSS Core Designer, Dataiku DSS Adv Designer, Registered Posts: 153 Dataiker

    Hi @emilieg

    From my understanding, you have SQL recipes in your application's flow that are using a hardcoded table name, which is causing issues when an app instance is created because the project key changes.

    You can use variables in your SQL recipe so that the table name is updated automatically. For example, in the below screenshot, I'm using the "${input:tbl:0}" variable, which expands to the name of the first input dataset. A list of variables is available in the Variables tab.


    Does this solve your problem? If not, could you please provide one of your SQL queries so that I can see what the issue is?



Setup Info
      Help me…