search

aws / sagemaker-inference-toolkit

Serve machine learning models within a 🐳 Docker container using 🧠 Amazon SageMaker.
Apache License 2.0
385 stars 82 forks source link

Ability to define custom log4j properties #52

Open arvarik opened 4 years ago

arvarik commented 4 years ago

Is your feature request related to a problem? Please describe.

Currently I can't define the log4j properties because the default log4j properties is hardcoded in.

https://github.com/aws/sagemaker-inference-toolkit/blob/ac1b164aed3c466876452202e5858b620a87c3cc/src/sagemaker_inference/model_server.py#L90

Would like to configure this so that the MMS logs are written.

Describe the solution you'd like Have the ability to define a log4j.properties file in my package.

I think the cleanest way here is to perhaps add more arguments to the start_model_server() function so that users can define their own properties.

At the very least I think the default log4j properties should match the default log4j properties in multi-model-server

https://github.com/awslabs/multi-model-server/blob/master/frontend/server/src/main/resources/log4j.properties

Additional context

A similar issue for config.properties is tracked here.

nadiaya commented 4 years ago

Thank you for the suggestion!