README.md

How to deploy an AWS Lambda function

Prereqs

1. Install okta-keyman and aws cli: pip install awscli aws-okta-keyman
2. aws_okta_keyman -o mdsol -u your_medidata_id -a 0oaj1kcx7aTjEmXas0x7/272 -r
3. Create a new conda environment to install all necessary packages. In our case our test example only requires flask, but feel free to include other libs that you need KEEP THE PACKAGES TO A MINIMUM
4. EXPORT VIRTUAL_ENV=/home/eyang/miniconda3/envs/(your conda_env) (for linux)
   • For windows: https://docs.oracle.com/en/database/oracle/machine-learning/oml4r/1.5.1/oread/creating-and-modifying-environment-variables-on-windows.html
   • If that doesn't work. I'm sorry you're on your own. HINT HINT learning linux isn't a bad thing
5. Install zappa pip install zappa

When creating a new project

1. zappa init
2. Modify the zappa_settings.json

{
    "dev": {
        "s3_bucket": "eyang-test-2",  # this is a temporary work space. you can create your own 
        "app_function": "example.app",# it's example because our primary file is example.py
        "lambda_handler": "example.lambda_handler",  #the primary function that is going to be run
        "http_methods": ["GET"],    # are we sending info as a GET or a POST Request
        "manage_roles": false,      # Use an existing role for execution, otherwise new roles are created 
        "role_arn":"arn:aws:iam::833923177614:role/acorn-dse-lambda", # the role that we as DS can use for our lambda
        "profile_name": "okta-orangeds"   #okta keyman generates this and allows us to actually do things on aws
    
    }
}

3. zappa deploy

The following error can be ignored

CloudFormation stack missing, re-deploy to enable updates
ERROR:Could not get API ID.
Traceback (most recent call last):
  File "/home/eyang/miniconda3/envs/lambda/lib/python3.9/site-packages/zappa/core.py", line 2420, in get_api_id
    response = self.cf_client.describe_stack_resource(StackName=lambda_name, LogicalResourceId="Api")
  File "/home/eyang/miniconda3/envs/lambda/lib/python3.9/site-packages/botocore/client.py", line 530, in _api_call
    return self._make_api_call(operation_name, kwargs)
  File "/home/eyang/miniconda3/envs/lambda/lib/python3.9/site-packages/botocore/client.py", line 960, in _make_api_call
    raise error_class(parsed_response, operation_name)
botocore.exceptions.ClientError: An error occurred (ValidationError) when calling the DescribeStackResource operation: Stack 'eyang-lambda-dev' does not exist
Oh no! An error occurred! :(

To invoke

aws lambda invoke --function-name eyang-lambda-dev response.json --profile okta-orangeds

This is just to test. If you need to use this operationally, you an use boto3 to wrap this in a python script more generally, or turn this into a actual URL (turning into a url is left as an exercise to the reader, since given the need to pass auth information over I find to be a pain in the ass, and for most of our purposes not really necessary, or it's going to be handled by DSE in the context of an app)