This is a self-paced workshop on OpenFaaS running on Kubernetes.
Requirements:
- Functions will be written in Python, so prior programming or scripting experience is preferred
- Preferred OS: MacOS, Linux, Windows 10 Pro
- Kubernetes
- Docker for Mac Edge
- Or an account on DigitalOcean to provision a small Kubernetes cluster
Disclaimer: this is a work-in-progress.
- Install pre-requisites
- Docker Hub account
- OpenFaaS CLI
- Set up a single-node cluster with Docker Swarm
- Deploy OpenFaaS
- Use the UI Portal
- Learn about the CLI
- Deploy via the Function Store
- Find metrics with Prometheus
- Scaffold or generate a new function
- Build the space-counter function
- Add dependencies with
pip
- Troubleshooting: verbose output with
write_debug
- Troubleshooting: find the container's logs
- Use third-party language templates
- Extend timeouts with
read_timeout
- Inject configuration through environmental variables
- Use HTTP context - querystring / headers etc
- Call a function asynchronously vs synchronously
- View the queue-worker's logs
- Use an
X-Callback-Url
with requestbin and ngrok
- Make use of another function
- Director pattern
Triaging customer feedback from Twitter
- Receiving Tweets from IFTTT.com
- Filtering out unwanted messages
- Sorting good + bad with SentimentAnalysis
- Forwarding to two different Slack channels
You can start with the first lab Lab 1.
If you're taking an instructor-led workshop, we'll now take Q&A and cover some advanced topics too.
- What is auto-scaling and how does it work?
- How to monitoring functions with a dashboard in Grafana
- How to enable TLS for security
- How to lock down the gateway with Basic Authentication
- Object storage with S3/Minio
- How to build your own templates
- How to customise the templates