ScalerCtl

CLI tool to interact with NodeFitter

About This Project

NodeFitter is a simple VM autoscaler created for the "Fog and Cloud Computing" course at the University of Trento, Italy.

The goal was to create a VM autoscaler capable of automatically deploying VMs and making them join a Kubernetes cluster for it to be able to schedule pods.

This repository hosts a practical CLI to interact with NodeFitter.

Installation Instructions

For the purpose of this project the CLI is meant to be run inside a Docker container alongside the autoscaler. From the Submission repository, run:

cd nodefitter
docker compose up

To run this project by itself, install the Go language package on your device (golag-go on Debian-based distributions), then run:

go install

Ensure the installation path (~/go/bin by default) is included in your $PATH

Usage

With the Docker container running, execute:

docker exec <container> /app/scalerctl <subcommand> <args>

The available subcommands are:

• get: returns information about VMs and threshold.
  • get vms: to get the current local list of VMs. The resource usage gets updated at every scheduling cycle, otherwise it will appear as "Not Available" (N/A) if currently not available.
  • get cpu: to get the current CPU threshold.
  • get ram: to get the current memory threshold.
• set: overwrite the current threshold of free resources below which a new VM will be scheduled. The initial values come from the schedulerConfig.yaml file required by the autoscaler.
  • set cpu <percentage> to set the free % of CPU usage (1–100).
  • set ram <Mb> to set the free memory in Mb.
• stop: pause the scheduling process.
• start: resume the scheduling process.