Inviso3d

Idea, Vision

Expanding perception and visibility of complex distributed computing systems through spatial visualization.

What does it do currently ?

Gets the list of objects from Kubernetes and displays it in 3d space with meta info embedded in visual representation. In the ProxyService, there are still classes to fetch data from GC Compute and OpenShift, but in the current branch, the focus is on Kubernetes.

Architecture

[3d client] <--websocket--> [proxy service] <--http(s)---> [APIs, data sources]

3d Client talks to ProxyService via JSON asynchronously.

Proxy Service

ProxyService is an intermediary that proxies data from K8s API. It provides a WebSocket and HTTP server and keeps track of client sessions. It is written in Java and Spark framework. Currently has Kubernetes, OpenShift, and GCP Compute data sources (plugins), where the focus is on Kubernetes. Apart from the WebSocket endpoint service also has an HTTP endpoint for static content. By default, the service runs on port 4567 (both HTTP and WS)

Proxy Service Plugin

Kubernetes

For the plugin to work, you need to have either:

• Configured kubectl on the machine where Inviso ProxyService will run or config file with at least one context. On Linux, usually in ~/.kube/config.
• If deployed using Helm, it will pick the config from within the cluster (Service account RO access rights defined in Helm needs to be applied during installation)

OpenShift Plugin [Legacy]

Gets and filters pods from OpenShift API. The plugin was tested with OKD v3.11 with API v1. Currently only works with token auth. To get your API key, use oc tool:

oc login -u your_username https://url_to_your_openshift.com:8443

oc whoami -t

Update config.properties accordingly.

Google Cloud Compute Plugin [Legacy]

Uses Google's API Client for Java and gcloud CLI authentication.

Howto for installing the gcloud command line tool: https://cloud.google.com/sdk/docs/quickstarts After setting it up and successfully authenticating, you need you set the environment variable GOOGLE_APPLICATION_CREDENTIALS to point to the gcloud credentials token.

In Linux/bash:

export GOOGLE_APPLICATION_CREDENTIALS = /home/YOUR_HOME_DIR/.config/gcloud/legacy_credentials/YOUR_USER@SOMEWHERE.COM/adc.json

Or use your preferred IDE to set up runtime environment variable.

Visualisation Clients

All static content (clients) reside in /src/main/resources/public/ Service serves this content on http://localhost:4567 (if you are running the service locally)

3DView

Technology: html, .js, WebGL, three.js

3D representation of nodes. Each box represents one node/container/VM where position represents state (higher = online / lower = offline) Colors represent different roles. Aiming and clicking on a specific box shows basic info in the properties box.

If run locally, you can access it on: http://localhot:4567/3dview.html

The current configured frontend is 3dviewV2.js, and it supports only Kubernetes visualization

• Kubernetes query string format: http://localhost:4567/3dview.html?provider=k8s&context=YOUR-K8S-CONTEXT If it is run within cluster context is ignored.

Move around by using standard WASD keys + mouse aiming Pressing ~ button will bring down quake-style console. Typing help into the console will show basic commands

Inspiration for the 3Dviewer was three.js example: (https://threejs.org/examples/#misc_controls_pointerlock)

Debug endpoint for testing server messages

location /debug.html

Displays server (ProxyService) messages and active clients

Quickstart: Build and Run

Local run

Tested with JDK 9 and OpenJDK 9

Update service config file

Copy dist.config.properties to config.properties and update authentication and config data.

cp dist.config.properties config.properties

Maven build

mvn clean compile assembly:single

Run

java -cp target/inviso-1.0-SNAPSHOT-jar-with-dependencies.jar ProxyService

If running locally you should go to http://localhost:4567 for index page

Running in Docker

Get the repository and change directory git clone git@github.com:rgombash/inviso3d.git & cd inviso3d

cp dist.config.properties config.properties Edit config.properties and update authentication and config data according to your setup Note: For viewing GC Compute or Kubernetes, you must get your configs into the container image by adding them to Dockerfile or using the 'docker cp' command. Refer to the plugins section for details.

Build the image docker build -t inviso .

Run the image docker run -p 4567:4567 inviso

Running in Kubernetes cluster

dependencies: installed and configured kubectl and helm

from the source root:

cd charts helm install inviso ./inviso

TODOs

• Enable HTTPS and WSS
• Add more object types like deployments, replica sets, services etc. and make hierarchical representation of the objects
• Test subscription model, client subscribes to periodic streaming updates and refreshes states periodically

Final notes

Inviso3d is a hobby project that I have been playing with sporadically for a few years now. The initial concept was written in Python and the Panda3d engine and was later rewritten to Java and Three.js engine. The code was prototyped fast and dirty (also my first Java service), so do not expect high-quality code.