How to add basic authentication for Tensorflow serving

Question

I am currently trying to serve a model built with Tensorflow's dnn premade estimator with TF serving. I am very happy to see the release of the restful API with Tensorflow Serving 1.8.

However, if we can specify what will contain the input and the output, I don't see how to manage basic authentication.

Thanks in advance for your advices. Regards.

Aruwimi

Ok, I don't know, who downvotes my post, but the least he should have done is explain why he did that. The fact is, according to the tutorial in the website of tensorflow, we know how to call a specific model via the URL. This is the following code : POST host:port/<URI>:<VERB> with URI: /v1/models/${MODEL_NAME}[/versions/${MODEL_VERSION}] And VERB: classify|regress|predict. – Aruwimi

However, nowhere in the tutorial, they explain how to add basic authentication. In this tutorial of [tensorflow] (tensorflow.org/serving/serving_basic), they told us that we just launch tensorflow_model_server (so a black box) that will use our models. SO I don't see how to add the authentication. So again thank you for your advices, and for the one who downvotes my post, thank you to explain why you did that. – Aruwimi

It is not explain in the doc how to do that. – Aruwimi

Did you ever find a solution? – Carrycarryall

What do you exactly mean by "manage basic authentication"? Can you elaborate – Bulganin

@user961954, As I did not find a way, I used a flask server which managed the basic authentication and which launch the tensorflow models. – Aruwimi

@tensorflow-support, what i wanted was to add a login password to access my model with Tensorflow serving – Aruwimi

Can you take a look at this comment from a similar SO thread? Also you can use this link as a reference and let us know if it helps in adding authentication to tensorflow serving. Thanks! – Bulganin