faas-netes is an OpenFaaS provider which enables Kubernetes for OpenFaaS. It's part of a larger stack that brings a cloud-agnostic serverless experience to Kubernetes.
The existing REST API, CLI and UI are fully compatible. With OpenFaaS Pro, you have an optional operator mode so that you can manage functions with kubectl and a CustomResource.
You can deploy OpenFaaS to any Kubernetes service - whether managed or local, including to OpenShift. You will find any specific instructions and additional links in the documentation.
OpenFaaS (Functions as a Service) is a framework for building serverless functions with Docker and Kubernetes which has first class support for metrics. Any process can be packaged as a function enabling you to consume a range of web events without repetitive boiler-plate coding.
Pictured: OpenFaaS conceptual architecture
- Platform for deploying serverless-style workloads - microservices and functions
- Native Kubernetes integrations (API and ecosystem)
- Built-in UI portal
- Scale to and from zero
- Built-in queuing and asynchronous invocations
- Custom routes and domain support
- Commercial support available
Additional & ecosystem:
- A range of event-connectors and cron-support
- helm chart and CLI installer
- Operator available to use Custom Resource Definitions (CRDs) (See also: OpenFaaS Pro)
- IDp integration with OIDC and commercial add-on
Community:
- Over 30k GitHub stars
- Independent open-source project with over 300 contributors, with commercial option available
Commercial options:
- Support from full-time team
- Commercial add-ons and integrations with events like Kafka, Postgres, AWS SQS and Cron
- Multiple namespace support
- gVisor support and runtimeClass for isolation
- Affinity and advanced scheduling / security constraints
- Tutorial: Deploy OpenFaaS to Kubernetes with its helm chart
- Read news and tutorials on the openfaas.com blog
- Chat with the community on OpenFaaS Slack
OpenFaaS can be used as complete stack for Cloud Native application development called PLONK. The PLONK Stack includes: Prometheus, Linux/Linkerd, OpenFaaS, NATS/Nginx and Kubernetes.
Read more: Introducing PLONK.
The rest of this document is dedicated to technical and operational information for the controller.
There are two modes available for faas-netes, the classic mode is the default.
- Classic mode (aka faas-netes) - includes a REST API, multiple-namespace support but no Function CRD - available in Community Edition and OpenFaaS Pro/Enteprise
- Operator mode (aka "The OpenFaaS Operator") - includes a REST API, with a "Function" CRD and multiple-namespace OpenFaaS Pro/Enterprise
See also: README for "The OpenFaaS Operator"
The single faas-netes image and binary contains both modes, switch between one or the other using the helm chart or the flag -operator=true/false.
faas-netes can be configured with environment variables, but for a full set of options see the helm chart.
| Option | Usage |
|---|---|
httpProbe |
Boolean - use http probe type for function readiness and liveness. Default: true |
write_timeout |
HTTP timeout for writing a response body from your function (in seconds). Default: 60s |
read_timeout |
HTTP timeout for reading the payload from the client caller (in seconds). Default: 60s |
image_pull_policy |
Image pull policy for deployed functions (Always, IfNotPresent, Never). Default: Always |
gateway.resources |
CPU/Memory resources requests/limits (memory: 120Mi, cpu: 50m) |
faasnetes.resources |
CPU/Memory resources requests/limits (memory: 120Mi, cpu: 50m) |
operator.resources |
CPU/Memory resources requests/limits (memory: 120Mi, cpu: 50m) |
queueWorker.resources |
CPU/Memory resources requests/limits (memory: 120Mi, cpu: 50m) |
prometheus.resources |
CPU/Memory resources requests/limits (memory: 512Mi) |
alertmanager.resources |
CPU/Memory resources requests/limits (memory: 25Mi) |
nats.resources |
CPU/Memory resources requests/limits (memory: 120Mi) |
faasIdler.resources |
CPU/Memory resources requests/limits (memory: 64Mi) |
basicAuthPlugin.resources |
CPU/Memory resources requests/limits (memory: 50Mi, cpu: 20m) |
The readiness checking for functions assumes you are using our function watchdog which writes a .lock file in the default "tempdir" within a container. To see this in action you can delete the .lock file in a running Pod with kubectl exec and the function will be re-scheduled.
By default all OpenFaaS functions and services are deployed to the openfaas and openfaas-fn namespaces. To alter the namespace use the helm chart.
To configure ingress see the helm chart. By default NodePorts are used. These are listed in the deployment guide.
By default functions are exposed at http://gateway:8080/function/NAME.
You can also use the IngressOperator to set up custom domains and HTTP paths
By default, deployed functions will use an imagePullPolicy of Always, which ensures functions using static image tags are refreshed during an update.
If this is not desired behavior, set the image_pull_policy environment variable to an alternative. IfNotPresent is particularly useful when developing locally with minikube.
In this case, you can set your local environment to use minikube's docker so faas-cli build builds directly into minikube's image store.
faas-cli push is unnecessary in this workflow - use faas-cli build then faas-cli deploy.
Note: When set to Never, only local (or pulled) images will work. When set to IfNotPresent, function deployments may not be updated when using static image tags.
faas-netes maintainers strive to support as many Kubernetes versions as possible and it is currently compatible with Kubernetes 1.11 and higher. Instructions for OpenShift are also available in the documentation.
You can quickly create a standard development environment using:
make start-kindThis will use KinD to create a single node cluster and install the latest version of OpenFaaS via the Helm chart.
Check the contributor guide in CONTRIBUTING.md for more details on the workflow, processes, and additional tips.
This project is licensed under the MIT License.