Issue calling deployed Dataiku model into Python code running in a container

suhail_shaik

Hello,

I'm facing an issue trying to read a Dataiku Visual Model into a Python code which is a part of a webapp running in a container. I think the model call from dataiku.Model is making the correct call but it's not able to process the get_predictor() command correctly and I'm seeing a file not found error.

if i am running the webapp using local execution then it seems to work just fine the issue raises only when the webapp runs on a container

Attaching the log of the error for refrence

File "/opt/dataiku/python/dataiku/core/saved_model.py", line 202, in get_predictor
    self._predictors[version_id] = build_predictor_for_saved_model(model_folder, self.get_type(), sm.get("conditionalOutputs", []))
File "/opt/dataiku/python/dataiku/core/saved_model.py", line 268, in build_predictor_for_saved_model
    core_params = dkujson.load_from_filepath(osp.join(model_folder, "core_params.json"))  
File "/opt/dataiku/python/dataiku/core/dkujson.py", line 36, in load_from_filepath
    with open(filepath) as fp:
FileNotFoundError: [Errno 2] No such file or directory: '/app/dataiku/DSS_DATA_DIR/saved_models/<PROJECTKEY>/<modelid>/versions/initial/core_params.json'

MehdiH

Hello @suhail_shaik

Thank you for reaching out.

On the attached log, the project key and model ids seem to be missing (look at the `<PROJECTKEY>` and `<modelid>` in the filepath)

The issue must probably lie in the definition of your `filepath`, could you attach the code that performs this operation ?

Cheers

suhail_shaik

Hi @MehdiH

The `<PROJECTKEY>` and `<modelid>` in the file path were deliberately added by me while posting the question here, in the actual error log the path is correct.

The code that i have used is referenced from the below documentation. and you can see that the file path is not manually given.
Interaction with saved models — Dataiku DSS 9.0 documentation

However the above code is running on a containerized execution.

Clément_Stenac

Hi,

Unfortunately, it is not currently possible to use these APIs when the webapp is running in a container. You'll need to run this particular webapp locally.

Tsfan_22

Hello Clement,

Would this limitation also apply to a notebook run in a container?

Thanks!

MehdiH

Yes it also applies in containerized notebooks

Sreekrishnan

hi is this fixed on Dataiku 12?

Sreekrishnan

I am trying to get the dataiku trained model into my python recipe on same project and trying to get the predictor

client = dataiku.api_client()
project = client.get_default_project()
#print(project.list_saved_models())
sm_id = 'Anomaly detection Model'
m = dataiku.Model(sm_id, ignore_flow = True)
predictor = m.get_predictor()

This works on notebook , but not on containerized execution. Getting error

Job failed: Error in python process: At line 33: <class 'Exception'>: Error fetching dir /opt/dataiku/data/saved_models/

MehdiH

Hello Sreekrishnan,

This should be fixed starting from Dataiku 12.3

Best,