Vagrant AWS Lambda dev env

A very simple dev environment for AWS lambda using Vagrant.

I created this while playing around with the latest version of SAM (mostly to avoid conflicts with existing installations).

SAM

The latest version of sam-cli is installed in the VM.

In the guest, it is possible to use the following command to mock the API Gateway (using SAM) locally so that it is forwarded correctly to the host on port 3000:

lambda

It will be then possible to get the results from the API in the host, eg:

curl localhost:3000/my_function

Localstack

The VM uses the Localstack docker container via the localstack command (installed in the virtual environment via pip).

Some commands to work with Localstack on the VM:

# starting Localstack
localstack start

# starting Localstack in the background:
localstack start -d

# Checking the localstack logs (when Localstack is running in the background):
log

Using awslocal and samlocal with Localstack

To speed local development up, the following 2 packages are installed via pip:

• awslocal
• samlocal

They just offer a simplified method to run aws-cli commands without having to specify the localstack's host name with every command.

For instance:

awslocal s3 mb s3://just-a-test-bucket

will create the just-a-test-bucket in Localstack (if it is running, of course :-)):

awslocal s3 ls
# 2023-04-13 13:49:30 just-a-test-bucket

Using Boto3 with Localstack

Boto3 can easily be used with Localstack, for example:

import boto3
import os
boto3.setup_default_session(profile_name='localstack')
s3 = boto3.client('s3', region_name='us-east-1',                                
        endpoint_url=os.environ.get('LOCALSTACK_ENDPOINT_URL'))

You should then be able to use the s3 Boto3 client against your running instance of Localstack. For example:

s3.list_buckets()

Serverless

The machine has the serverless framework installed in case that is preferred over sam-cli.

The following commands are available:

serverless
sls

Also, the following plugin are installed:

• serverless-offline
• serverless-localstack

Before you run the machine

Set up the following environment variables:

# This is the directory that will be shared with the guest VM (this will be where
you store store the code of your Lambdas)

VAGRANT_LAMBDA_SHARED_DIR

To install and run the VM

Then:

vagrant up
vagrant ssh

Enjoy!

Some Gotchas

Note: Vagrant/Virtualbox does not work well with symlinks (for security reasons). Instead of allowing symlinks to VB, I decided to simply create a virtual env in the guest VM, which gets activated once the session starts (~/VM_VENV/bin/activate) is run via the bashrc)

Therefore, you will need to install the dependencies in your VM by using something like:

python3 -m pip install -r path/to/requirements.txt

If you are editing your code in an IDE in your host, you will need to create a virtual environment on the host, or pass the VM's interpreter via SSH if your IDE allows to do so (eg Pycharm and Intellij do).

Warning: The venv created on the host does not work on the guest and vice-versa (due to the symlinks issue with VB).

It is advisable to always using virtual environments when using Python.

One way to enforce this is to ensure that the following local env is set to true:

PIP_REQUIRE_VIRTUALENV

Note: At the time of writing, SAM and AWS Lambda fully support Python 3.9, but not 3.10. Therefore, I used the deadsnakes PPA to download Python 3.9. This should be updated once Python 3.10 will be updated.