https://github.com/snacsnoc/msmv
MicroVM specialized container build system in Python
https://github.com/snacsnoc/msmv
linux-build linux-containers linux-virtual-machine microvm python-container
Last synced: 5 months ago
JSON representation
MicroVM specialized container build system in Python
- Host: GitHub
- URL: https://github.com/snacsnoc/msmv
- Owner: snacsnoc
- License: gpl-3.0
- Created: 2024-06-14T08:32:27.000Z (over 1 year ago)
- Default Branch: main
- Last Pushed: 2024-07-02T09:45:15.000Z (over 1 year ago)
- Last Synced: 2025-01-22T18:33:33.346Z (about 1 year ago)
- Topics: linux-build, linux-containers, linux-virtual-machine, microvm, python-container
- Language: Python
- Homepage:
- Size: 143 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
MuStack MicroVM
======================
MuStack MicroVM (msmv) is a specialized container build system designed to create and manage minimal, secure, and isolated Linux virtual machine images for microVM environments.
Using QEMU's [microvm](https://www.qemu.org/docs/master/system/i386/microvm.html) machine type, msmv provides rapid boot times, enhanced security through isolation, and efficient resource management, for developers and hobbyists who require lightweight virtual testing environments.
Imagine Docker, except requiring a lot more effort. It’s like building a ship in a bottle — complex, rewarding, frustrating, and yes, a bit intense!
# Usage
## Configuration
Start by creating a TOML configuration file for your VM. Please see the `recipes/` directory for examples.
# Building the Virtual Machine Image
Note: it is recommended to use a virtual environment (venv)
```bash
python3 -m venv venv
pip install -r requirements.txt
```
Run the build script with your configuration file:
```bash
python -m msmv.bin.msmv --config-file my_application.toml
```
__Options:__
* `-c` or `--config-file` - sets the build (recipe) configuration file
* `-b` or `--build-dir` - sets the build workspace and output directory
__Environment variables:__
* `MAKE_COMMAND` - specify the `make` command, defaults to `make`
* For Mac OS, I would recommend `lkmake` for building the kernel https://github.com/markbhasawut/mac-linux-kdk
* `CC` - specify the compiler to be used, defaults to `cc`
* `LD` - specify the linker to be used, defaults to `ld`
This script will:
* Set up the workspace
* Download and extract the Linux kernel
* Configure and build the kernel
* Download and build the specified applications
* Set up the root filesystem
* Put a smile on your face
* Create the final VM image
# Running the Virtual Machine
After building, you can run the VM with QEMU:
Examples:
```bash
qemu-system-x86_64 -M microvm -drive file=output_vms/output_image.qcow2,if=virtio -m 128 -nographic -append "console=ttyS0" -qmp unix:/tmp/qmp-socket,server,nowait
```
The `microvm` machine type is not supported for aarch64, so we must use `virt`:
```bash
qemu-system-aarch64 -M virt -cpu max -kernel Image -initrd rootfs.cpio -append "init=/init rdinit=/init console=ttyAMA0" -serial mon:stdio -nographic
```
# VM Management with VMTool
msmv includes `VMTool`, a script for managing the VM lifecycle through the QEMU Machine Protocol (QMP):
```bash
python -m msmv.bin.msmv start # Starts the VM
python -m msmv.bin.msmv stop # Sends a shutdown signal to the VM
python -m msmv.bin.msmv status # Queries the current status of the VM
```
# TODO
* Use build commands defined in recipe versus assuming `make`
* Simplify Linux kernel downloading and optionally specify the download URL
* Create `init` script framework to boot into desired application upon VM start
* Use script argument to use ccache
* Pass cross-compiler env vars (`CC`, etc) to build processes
* Use architecture defined in recipe TOML versus host OS's arch
* Apply predefined kernel config sets (serial only, framebuffer, debugging etc) before user's config
* To reduce requirement for the user to supply all necessary kconfig options in the recipe file
# Contributing
Please fork the repository, make your changes, and submit a pull request.
# License
This project is licensed under the GPL v3 license.