Skip to content

Manage discovery of cloudlets and deployment of backends for edge-native applications

License

Notifications You must be signed in to change notification settings

cmusatyalab/sinfonia-tier3

Repository files navigation

Sinfonia

Manages discovery of nearby cloudlets and deployment of backends for edge-native applications.

The framework is a 3 tiered system. Tier 1 is located in the cloud and tracks availability of the Tier 2 instances running on the edge of the network (cloudlets) where backends can be deployed. Tier 3 is the client application that mediates the discovery and deployment process for edge-native applications.

This repository implements an example Tier3 client which can be used both as a command-line application and as a Python library.

Installation

You probably don't need to install this directly, most of the time it should get installed as a dependency of whichever edge-native application is using the Sinfonia framework to discover nearby cloudlets.

But if you want to run the standalone command-line application, you can install this with installable with pipx install sinfonia-tier3 or pip install [--user] sinfonia-tier3.

Usage

The sinfonia-tier3 application would normally be called by any edge-native application that uses the Sinfonia framework to deploy its application specific backend on a nearby cloudlet.

The information needed by the application are the URL of a Tier1 instance and the UUID identifying the required backend. The remainder of the arguments consist of the actual frontend application and arguments that will be launched in an seperate network namespace connecting back to the deployed backend.

$ sinfonia-tier3 <tier1-url> <uuid> <frontend-app> <args...>

An example application with UUID:00000000-0000-0000-0000-000000000000 (or the convenient alias 'helloworld') starts an nginx server that will be accessible with the hostname 'helloworld'.

$ sinfonia-tier3 https://tier1.server.url/ helloworld /bin/sh
sinfonia$ curl -v http://helloworld/
...
sinfonia$ exit

When the frontend application exits, the network namespace and WireGuard tunnel are cleaned up. Any resources on the cloudlet will be automatically released once the Sinfonia-tier2 instance notices the VPN tunnel has been idle.

Installation from this source repository

You need a recent version of poetry

$ pip install --user pipx
$ ~/.local/bin/pipx ensurepath
... possibly restart shell to pick up the right PATH
$ pipx install poetry

And then use poetry to install the necessary dependencies,

$ git clone https://github.com/cmusatyalab/sinfonia-tier3.git
$ cd sinfonia-tier3
$ poetry install
$ poetry run sinfonia-tier3 ...
... or
$ poetry shell
(env)$ sinfonia-tier3 ...

Why do we need a sudo password when deploying

Actually you should not need a password if wireguard4netns works correctly But if for some reason it fails to create the tuntap device and launch wireguard-go, the code will fall back on the older sudo implementation.

The older sudo implementation uses the in-kernel Wireguard implementation and needs root access to create and configure the WireGuard device and endpoint. All of the code running as root is contained in src/sinfonia_tier3/root_helper.py

It runs the equivalent of the following.

    ip link add wg-tunnel type wireguard
    wg set wg-tunnel private-key <private-key> peer <public-key> endpoint ...
    ip link set dev wg-tunnel netns <application network namespace>

About

Manage discovery of cloudlets and deployment of backends for edge-native applications

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages