grafana

Add Grafana for OpenFaaS function/service visualization

Grafana packages a pre-compiled dashboard for OpenFaaS to show metrics like throughput and latency.

• Apply the Deployment for Grafana

  kubectl apply -f ./grafana/deployment.yaml

• Port-forward the dasbhoard to your local computer:

  kubectl port-forward deploy/grafana -n openfaas 3000:3000

• View the dashboard when needed with:

  http://127.0.0.1:3000/dashboard/db/openfaas?refresh=5s&orgId=1

  The credentials are: admin/admin.

Run a load-test for testing

Below is a screenshot showing auto-scaling of the "echo-fn" in response to load being generated by the hey load-generation tool.

You can run this lab yourself, but this is optional.

Download (hey):

For MacOS:

curl -SLsf https://storage.googleapis.com/hey-release/hey_darwin_amd64 > hey
chmod +x hey
sudo mv hey /usr/local/bin

For Linux:

curl -SLsf https://storage.googleapis.com/hey-release/hey_linux_amd64 > hey
chmod +x hey
sudo mv hey /usr/local/bin

Run a load-test:

git clone https://github.com/alexellis/echo-fn
cd echo-fn
export OPENFAAS_URL=http://$PUBLIC_IP:31112
faas-cli deploy

Edit stack.yml, replace com.openfaas.scale.max: 1 with com.openfaas.scale.max: 20

Now run hey:

hey -c 10 -z 2m $OPENFAAS_URL/go-echo

You should see very low latency as the test runs for 2 minutes with 10 concurrent connections. View the Grafana dashboard for the visualization.