sousandrei / firesquid

Simple AWS Firecracker MicroVM Orchestrator

Simple AWS Firecracker orchestrator

Report Bug or Request Feature

Table of Contents

• About the project
• Features
• Assets
• Help Wanted
• Sponsors
• License

About the Project

FireSquid is a simple AWS Firecracker orchestrator. It abstracts the hard part making it a breeze to spawn your very own fleet of micro-vms!

Features

• REST API
• Multiple vms from the same kernel
• Lightweight
• Customizable

Upcoming:

• Fine grained networking
• Package release and distribution
• Choices between which kernel to use for which machine

Assets 📦

Here are some assets to get you started. The default folder for assets is just called assets in the same folder as FireSquid.

Grab the assets listed in the official firecracker guide. There is a script provided in this repo to set this up

curl -fsSL https://raw.githubusercontent.com/sousandrei/firesquid/main/scripts/getassets.sh | sh

Help wanted 🤝

All contributions are welcome!

If you are using or plan to use Firecracker or FireSquid please don't hesitate to reach out so we can foster a collaboration ecosystem around Firecracker.

Sponsors ❤️

you?

License

See LICENSE for more details.

firesquid's Issues

Upgrade to clap 4.x.x

Gracefully shutdown

Listen to interrupt signal
Send shutdown signal to all vms
Wait for their processes to end
Send shutdown signal to hyper server

Treatnig unwraps

Treating errors that are suppressed with unwraps throughout the code

Run as a system service

Investigate how to Run as a service on Linux
Create a .service file
Prioritize ubuntu initially

Migrate to Axum

Migrate to Github actions

Leave Travis in favor or github actions

Separate state

separate all state and state operations into a new module
investigate the going from String to &str in the state

Permissions

Hey, cool project!

I tried the application and had some issues with permissions.

The first issue is that I couldn't download Linux Kernel 5.9-rc2 that is listed in the Asset section of the README.

The error was:

<Error>
<Code>AccessDenied</Code>
<Message>Access denied.</Message>
<Details>Anonymous caller does not have storage.objects.get access to the Google Cloud Storage object.</Details>
</Error>

So I grabbed the kernel and rootfs from Getting Started with Firecracker instead (and Firecracker v0.23.0).

Then I started the app with cargo run and sent this request:

~/.../firesquid >>> curl --header "Content-Type: application/json" \
  --request POST \
  --data '{"vm_name":"test"}' \      
  http://localhost:3000          

The response was:

{"success":true}%

But this was printed to the console:

thread 'tokio-runtime-worker' panicked at 'Failed to spawn vm child process: Os { code: 13, kind: PermissionDenied, message: "Permission denied" }', src/vm/child.rs:29:10

Any idea why I got a permission denied error?

I would be happy to provide more information if needed.

Implement logging on instances

Initialize log dir if needed
Bind and listen to stdout and stderr on the vms
Write to files

Cleanup HTTP handlers

Lots of duplicated code in there, we can make it more elegant

Update documentation

Add a how to use section on readme
Add a guide on how to create a boot drive
Add a guide to compile the linux kernel for firecracker