Run Ray Task on AWS Lambda

Ray Core
1

Hello,

I have a basic Ray code that consumes resources.

My aim is to deploy Python code on AWS with dockerized. When I try to use the docker image on AWS Lambda Function, It not working.

Python code which simple consume resource:
import math
import time
import ray

#ray.init(“anyscale://ray-demo-cluster”, runtime_env={“pip”: “requirements.txt”})
#ray.init(“anyscale://anyscale-demo/kd-cluster-v1”)

@ray.remote
def factorial_func(n):
return math.factorial(n)

def lambda_handler(event, context):
start = time.time()
results = [factorial_func.remote(i*40) for i in range(2000)]
ray.get(results)
finish = time.time()
print(f’Total duration:',finish-start)

lambda_handler(None, None)

Issue:

image
image1115×138 14.6 KB

Why is this happening?

Many thanks for your time.

2

I’m not sure if the issue is related to Ray. Ray doesn’t use sudo permission.

3

Possibly the issue is related to Anyscale. Because I am using Anyscale base image when I am building a docker image.

Dockerfile:

# Use Anyscale base image
FROM anyscale/ray:2.0.0-py37

ADD demo.py .

RUN echo "Testing Ray Import..." && python -c "import ray"

CMD [ "main.lambda_handler" ]

Is it possible?

4

yea it’s possible. Are you using anyscale? If not, you should use the open source ray image from https://hub.docker.com/r/rayproject/ray