nmoutschen / devbox-spot

Tools to run a development environment on spot instances

Hi! 👋

My name is Nicolas Moutschen. I'm an engineer building things in Rust at Apollo GraphQL. You can find more about me on my personal website.

Here are some of the things I do (or contribute to):

🛠️ Tools

• Serverless Rules - opinionated linting rules on serverless recommended practices for your infrastructure-as-code templates
• AWS Lambda Rust Runtime - runtime for the Rust programming language on AWS Lambda
• Cookiecutter SAM templates for Rust - Templates to get started with Rust and serverless

🧑‍🏫 Samples and demos

• Serverless Ecommerce Platform - a sample application showcasing 10 microservices interacting over Amazon EventBridge
• Serverless Rust Demo - an application built with Rust and serverless technologies

devbox-spot's Issues

Add tests

Add tests for the Lambda function code

Hide make error message with make connect

Make returns an error when running the following commands, even though this is expected behaviour:

make connect

# Inside the instance
sudo shutdown -h now

Add support for whitelisting an IP through the API

Add a new API paths that would whitelist the IP address of the caller, even if the instance is already running.

Automatically shut down the instance after a period of inactivity

When no SSH connections are active on the instance for a certain period of time, a daemon should automatically shut down the instance to save costs.

Wait and auto-connect to instances on start

Right now, when a developer runs make start, it only triggers the start of the instance. The developer then have to type make status to check if the instance is running, and make connect to connect to the instance.

Preserve EBS volumes on instance change

When the instance is changed (e.g. changing the instance type), the EBS volume is lost. This would allow to reuse the EBS volume.

Add support for specifying a VPC ID

Right now, the instance and associated resources will launch in a default VPC.

Stack error when the service role doesn't exist

The stack fails to deploy with this error message when the service role for spot instances is non-existent:

An error occurred (AuthFailure.ServiceLinkedRoleCreationNotPermitted) when calling the RunInstances operation: The provided credentials do not have permission to create the service-linked role for EC2 Spot Instances.

Add checks for race conditions between OnRun and OnStop

When stopping/running the instance with a short time interval, there could be a race condition causing the record to be deleted from Route53 while the instance is running.

When using HostedZoneId, check if the DNS name is correct in Route53 in StatusFunction

The status function returns the instance status, but the DNS might not have propagated to Route53 yet. Therefore, the function should check if it's possible to connect to the instance as well.

Add the ability to grant IAM permissions to the instance

Right now, the instances don't have IAM permissions out of the box. This would create a role and allow the EC2 instance permissions to assume that role at creation time.

Add support for custom domains

As a user, I want to be able to use the same domain over and over so that I don't need to reconfigure tools that leverage SSH connections.

Therefore, it would be nice to have support for custom domain (e.g. devbox.mydomain).

Check if samconfig.toml exist on make deploy

Currently, make deploy assumes that there is a samconfig.toml file at the root of the repository, meaning that users need to run sam deploy --guided on a first run.