RuntimeError: Java gateway process exited before sending its port number when deploying Pyspark model to Azure Container Instance (Issue #23158)

Question

Toby Tan 1

Hi, I raised my issue in a GitHub repo for azure-sdk-for-python too:

romungi-MSFT 49,096 Reputation points Microsoft Employee Moderator

2022-02-23T08:13:37.143+00:00

@Toby Tan I think you need to use mmlspark environment after registering the model with Azure ML workspace to deploy it as ACI endpoint. Sample notebook available here.
I am not sure if azureml-mlflow package supports this framework. I think you have raised the issue on the right repo to check if this is a supported scenario.
Toby Tan 1 Reputation point

2022-02-23T10:20:24.14+00:00

Hi @romungi-MSFT . thank you for your suggestion. Do you know which repo I can raise this on?
Toby Tan 1 Reputation point

2022-02-23T15:37:19.137+00:00

Hi @romungi-MSFT , I tried the sample notebook and it doesn't work for me....
romungi-MSFT 49,096 Reputation points Microsoft Employee Moderator

2022-02-28T06:47:06.157+00:00

@Toby Tan Ok, In the first case did you try to register the model with Azure ML workspace and deploy it from ml.azure.com as a real time endpoint.
Register the downloaded model or model folder with Azure ML workspace using the models tab and then click the deploy option to deploy it as real time or batch endpoint. I believe this process will also setup your dependencies and scoring scripts.

Sample screen shot:

Your answer

romungi-MSFT 49,096 Reputation points Microsoft Employee Moderator

2022-02-23T08:13:37.143+00:00

@Toby Tan I think you need to use mmlspark environment after registering the model with Azure ML workspace to deploy it as ACI endpoint. Sample notebook available here.
I am not sure if azureml-mlflow package supports this framework. I think you have raised the issue on the right repo to check if this is a supported scenario.
Toby Tan 1 Reputation point

2022-02-23T10:20:24.14+00:00

Hi @romungi-MSFT . thank you for your suggestion. Do you know which repo I can raise this on?
Toby Tan 1 Reputation point

2022-02-23T15:37:19.137+00:00

Hi @romungi-MSFT , I tried the sample notebook and it doesn't work for me....
romungi-MSFT 49,096 Reputation points Microsoft Employee Moderator

2022-02-28T06:47:06.157+00:00

@Toby Tan Ok, In the first case did you try to register the model with Azure ML workspace and deploy it from ml.azure.com as a real time endpoint.
Register the downloaded model or model folder with Azure ML workspace using the models tab and then click the deploy option to deploy it as real time or batch endpoint. I believe this process will also setup your dependencies and scoring scripts.

Sample screen shot: