Hyperglass Helm Charts

Warning

Please note that hyperglass is currently not maintained and uses a lot of software that is EOL since a long time (e.g. see this). Hyperglass should not be considered a safe piece of software. This helmchart protects you to some degree, as the application cant escape out of the containers, but you should still think about this, especially since hyperglass may have ssh keys to your core infrastructure!

Overview

This repository contains helmcharts for a hyperglass deployment. The deployment includes a redis caching instance as well as a configurable hyperglass.yaml configuration file which is stored as a kubernetes ConfigMap and is later mounted as a file into the containers.

Prerequisites

Consult the official documentation on how to setup hyperglass, especially in regards to creating a devices.yaml config and a hyperglass.yaml. Once you have created these configs, you are ready to deploy hyperglass using this helmchart.

Important

Please note that it is absolutely mandatory that you have the following lines in your hyperglass.yaml:

listen_address: "0.0.0.0"
listen_port: 8080
cache:
  database: 0
  host: 127.0.0.1
  port: 6379
  show_text: true
  timeout: 120

Deployment via helm and kubectl

Hyperglass is mainly configured through two files: hyperglass.yaml and devices.yaml.

As the devices.yaml and hyperglass.yaml file differs greatly from deployment to deployment, the user is expected to pass a valid yaml object in values.devices_config and values.hyperglass_config.

Installation

• Have your devices.yaml and hyperglass.yaml ready and in the current directory

Add the helm repo:

helm repo add mowoe-hyperglass https://mowoe.github.io/hyperglass-helm

And install the chart with the devices.yaml:

helm install \
    --set-file server_config=hyperglass.yaml \
    --set-file devices_config=devices.yaml \
    --create-namespace -n hyperglass \
    my-cool-lookingglass \
    mowoe-hyperglass/hyperglass

(Omit the --create-namespace -n hyperglass flags if youre on RedHat Openshift)

Network configuration

Note that the default deployment will only use a NodePort-Service because the intended way to deploy this, is to have a seperate loadbalancer like traefik in place. If you want to directly assign a loadbalancer to the service you can change this to LoadBalancer.

Additional Information

Due to some unfourtunate design choices in hyperglass itself, the pods take quite some time (~1 min) to be ready, as they will always build the ui again upon start. After the first deployment, this shouldnt be an issue though, as the service will only hand off traffic to new replica sets as soon as the pods are fully ready (like you would expect).