leehanchung

#!/usr/bin/env bash

MODEL_NAME=<<name your model>>
ROLE_NAME=AmazonSageMaker-ExecutionRole-XXXXXXXXXXXXXXXX

# the name of the image created with
ECS_IMAGE_NAME=<<your image name>>

# the role arn of the role
EXECUTION_ROLE_ARN=$(aws iam get-role --role-name ${ROLE_NAME} | jq -r .Role.Arn)

leehanchung

# Using the official tensorflow serving image from docker hub as base image 
FROM tensorflow/serving

# Installing NGINX, used to rever proxy the predictions from SageMaker to TF Serving
RUN apt-get update && apt-get install -y --no-install-recommends nginx git

# Copy our model folder to the container
COPY <<local model directory>> /<<model directory inside Docker image>>

# Copy NGINX configuration to the container

leehanchung

events {
    # determines how many requests can simultaneously be served
    worker_connections 2048;
}

http {
    server {
    # Increase JSON POST payload to bypass 413 Client Error: Request Entity Too Large for URL
    client_max_body_size 100M;

leehanchung

#!/usr/bin/env bash

IMAGE_NAME={name your local Docker image here}

REGION=$(aws configure get region)
ACCOUNT=$(aws sts get-caller-identity --query Account --output text)

# If the ECS repository doesn't exist, creates it.
aws ecr create-repository --repository-name ${IMAGE_NAME} > /dev/null 2>&1