How to set recipe container through dataikuapi

fmonari

Hi,

I would like to change dynamically the container for the execution of a recipe through the dataikuapi-external-client. I could not find any referenece in the documentation.

How can I achieve that (if possible)?

Regards,

Filippo

HarizoR

Hi again,

In that case you will need to use the get_settings() method of the Recipe API, retrieve the existing parameters for container configuration, modify the dictionary in-place and finally call the save() method on the recipe settings handle. Here is a simplified example:

import dataikuapi
client = dataikuapi.DSSClient(host=HOST, api_key=API_KEY)
project = client.get_project(PROJECT_KEY)
recipe_settings = project.get_recipe(RECIPE_ID).get_settings()
recipe_advanced_params = recipe_settings.raw_params
recipe_advanced_params["containerSelection"]["containerMode"] = "EXPLICIT_CONTAINER"
recipe_advanced_params["containerSelection"]["containerConf"] = YOUR_CONTAINER_CONF_NAME

Of course this assumes that you already have an existing functioning containerized execution configuration in your instance's administration settings.

Best,

Harizo

HarizoR

Hi Filippo,

Do you mean to programmatically change the "container configuration" settings for a given recipe ?

Best,

Harizo

fmonari

Exactly. How can I do that? What methods in the DSSRecipe or DSSRecipeSettings classes should I use?

Thanks!

Torben

How can the number of concurrent jobs be specified? E.g. if a compute resource is set to a 'large configuration' how can it be limited to, say, 10 concurrent jobs, while for a standard compute be reset to the default?