Skip to content

README-OPERATOR.md

Latest commit

 

History

History
397 lines (277 loc) · 9.15 KB

README-OPERATOR.md

File metadata and controls

397 lines (277 loc) · 9.15 KB

faas-netes

faas-netes operator README

Example Function for openfaas.com/v1 CRD

  • Minimal example:
apiVersion: openfaas.com/v1
kind: Function
metadata:
  name: nodeinfo
  namespace: openfaas-fn
spec:
  name: nodeinfo
  image: functions/nodeinfo:latest
  • Extended example:
apiVersion: openfaas.com/v1
kind: Function
metadata:
  name: nodeinfo
  namespace: openfaas-fn
spec:
  name: nodeinfo
  handler: node main.js
  image: functions/nodeinfo:latest
  labels:
    com.openfaas.scale.min: "2"
    com.openfaas.scale.max: "15"
  annotations:
    current-time: Mon 6 Aug 23:42:00 BST 2018
    next-time: Mon 6 Aug 23:42:00 BST 2019
  environment:
    write_debug: "true"
  limits:
    cpu: "200m"
    memory: "256Mi"
  requests:
    cpu: "10m"
    memory: "128Mi"
  constraints:
    - "cloud.google.com/gke-nodepool=default-pool"
  secrets:
    - nodeinfo-secret1

Example adapted from artifacts/nodeinfo.yaml

  • Generate CRD from stack.yml

You can generate CRD entries separated by --- by running faas-cli generate in the same folder as a stack.yml file. Each function in the file will be outputted. If you want a CRD for one function only then you can also pass --filter=function-name.

# create a go function for Docker Hub user `alexellis2`
faas-cli new --lang go --prefix alexellis2 crd-example

# build and push an image
faas-cli build -f crd-example.yaml
faas-cli push -f crd-example.yaml

# generate the CRD entry from the "stack.yml" file and apply in the cluster
faas-cli generate -f crd-example.yaml | kubectl apply -f -
  • Generate CRD from Function Store
# find a function in the store
faas-cli store list
...


# generate to a file
faas-cli generate --from-store="figlet" > figlet-crd.yaml
kubectl apply -f figlet-crd.yaml

Get started

You can deploy OpenFaaS and the operator using helm, or generate static YAML through the helm template command.

Deploy OpenFaaS with the operator (arkade)

Two namespaces will be created - openfaas for the core services and openfaas-fn for functions.

arkade install openfaas --operator

You will see the connection info and password after the command has completed, you can get the info back at any point in time with the arkade info openfaas command.

Deploy OpenFaaS with the operator (helm 3)

Two namespaces will be used - openfaas for the core services and openfaas-fn for functions.

# create OpenFaaS namespaces
kubectl apply -f https://raw.githubusercontent.com/openfaas/faas-netes/master/namespaces.yml

# add OpenFaaS Helm repo
helm repo add openfaas https://openfaas.github.io/faas-netes/

# generate a random password
PASSWORD=$(head -c 12 /dev/urandom | shasum| cut -d' ' -f1)

kubectl -n openfaas create secret generic basic-auth \
--from-literal=basic-auth-user=admin \
--from-literal=basic-auth-password="$PASSWORD"

# get latest chart version
helm repo update

# install with basic auth enabled
helm upgrade openfaas --install openfaas/openfaas \
    --namespace openfaas  \
    --set basic_auth=true \
    --set functionNamespace=openfaas-fn \
    --set operator.create=true

Note: If you are switching from the OpenFaaS faas-netes controller, then you will need to remove all functions and redeploy them after switching to the operator.

If you want to enable multiple namespaces feature which enables you to create functions across namespaces in the cluster, set clusterRole=true.

Deploy a function with kubectl:

kubectl -n openfaas-fn apply -f ./artifacts/nodeinfo.yaml

List functions, services, deployments and pods

kubectl -n openfaas-fn get functions
kubectl -n openfaas-fn get all

Deploy a function with secrets

kubectl -n openfaas-fn create secret generic faas-token --from-literal=faas-token=token
kubectl -n openfaas-fn create secret generic faas-key --from-literal=faas-key=key

Add the secrets section in nodeinfo.yaml and re-apply the function:

  secrets:
   - faas-token
   - faas-key

Test that secrets are available inside the nodeinfo pod:

kubectl -n openfaas-fn exec -it nodeinfo-84fd464784-sd5ml -- sh

~ $ cat /var/openfaas/faas-key 
key

~ $ cat /var/openfaas/faas-token 
token

Test that node selectors work on GKE by adding the following to nodeinfo.yaml:

  constraints:
    - "cloud.google.com/gke-nodepool=default-pool"

Apply the function and check the deployment specs with:

kubectl -n openfaas-fn describe deployment nodeinfo

Development build

The OpenFaaS Operator runs as a sidecar in the gateway pod. For end to end testing you need to update the sidecar to use your development image.

  1. Build, tag and push your image to your own public docker repository:
export USERNAME="username"

make build \
 && docker tag openfaas/openfaas-operator:latest $USERNAME/openfaas-operator:latest-dev \
 && docker push $USERNAME/openfaas-operator:latest-dev
  1. Update your helm deployment
helm upgrade openfaas --install openfaas/openfaas \
    --namespace openfaas \
    --set functionNamespace=openfaas-fn \
    --set operator.create=true \
--set operator.image=$USERNAME/openfaas-operator:latest-dev

Run locally, without Docker

You will a KUBECONFIG file and accessible in your path.

  • Create the OpenFaaS CRD
$ kubectl apply -f artifacts/operator-crd.yaml
  • Build and launch the OpenFaaS controller locally
$ go build \
  && ./openfaas-operator -kubeconfig=$HOME/.kube/config -v=4

As an alternative, you can use go run

$ go run *.go -kubeconfig=$HOME/.kube/config

To use an alternative port set the port environmental variable to another value.

Create a function:

$ kubectl apply -f artifacts/nodeinfo.yaml

Check if nodeinfo deployment and service were created through the CRD:

$ kubectl get deployment nodeinfo
$ kubectl get service nodeinfo

Test if nodeinfo service can access the pods:

$ kubectl run -it --rm --restart=Never curl --image=byrnedo/alpine-curl:latest --command -- sh
/ # curl -d 'verbose' http://nodeinfo.default:8080

Delete nodeinfo function:

kubectl delete -f artifacts/nodeinfo.yaml

Check if nodeinfo pods, rc, deployment and service were removed:

kubectl get all

REST API

The operator also implements the OpenFaaS REST API which provides an additional way to manage functions and secrets in addition to using the CRD with kubectl directly.

If OpenFaaS is configured with the basic_auth=true flag then Basic Authentication is enabled on the REST API. If that is the case then reformat each curl command to also include the credentials.

  • Basic Auth off
curl -s http://localhost:8081/system/functions | jq .
  • Basic Auth enabled
export USER=""
export PASSWORD=""

curl -s http://$USER:$PASSWORD@localhost:8081/system/functions | jq .

Function management

Create or update a function:

curl -d '{"service":"nodeinfo","image":"functions/nodeinfo:burner","envProcess":"node main.js","labels":{"com.openfaas.scale.min":"2","com.openfaas.scale.max":"15"},"environment":{"output":"verbose","debug":"true"}}' -X POST  http://localhost:8081/system/functions

List functions:

curl -s http://localhost:8081/system/functions | jq .

Scale PODs up/down:

curl -d '{"serviceName":"nodeinfo", "replicas": 1}' -X POST http://localhost:8081/system/scale-function/nodeinfo

Get available replicas:

curl -s http://localhost:8081/system/function/nodeinfo | jq .availableReplicas

Remove function:

curl -d '{"functionName":"nodeinfo"}' -X DELETE http://localhost:8081/system/functions

Secret management

Create secret:

curl -d '{"name":"test","value":"test"}' -X POST http://localhost:8081/system/secrets

List secrets:

curl -X GET http://localhost:8081/system/secrets

Update secret:

curl -d '{"name":"test","value":"test update"}' -X PUT http://localhost:8081/system/secrets

Delete secret:

curl -d '{"name":"test"}' -X DELETE http://localhost:8081/system/secrets

Configure a service account for your function

Example service account:

apiVersion: v1
kind: ServiceAccount
metadata:
  name: build-robot
  namespace: openfaas-fn

Set the following in your function spec:

annotations:
  com.openfaas.serviceaccount: "build-robot"

Logging

Verbosity levels:

  • -v=0 CRUD actions via API and Controller including errors
  • -v=2 function call duration (Proxy API)
  • -v=4 Kubernetes informers events (highly verbose)

Instrumentation

Prometheus route:

curl http://localhost:8081/metrics

Profiling is disabled by default, to enable it set pprof environment variable to true.

The pprof UI can be access at http://localhost:8081/debug/pprof/. The goroutine, heap and threadcreate profilers are enabled along with the full goroutine stack dump.

Run the heap profiler:

go tool pprof goprofex http://localhost:8081/debug/pprof/heap

Run the goroutine profiler:

go tool pprof goprofex http://localhost:8081/debug/pprof/goroutine