Netifi Helm Charts

Kubectl

Install Kubectl

brew install kubernetes-cli

Minikube

Install Minikube with HyperKit.

brew cask install minikube
brew install docker-machine-driver-hyperkit

# docker-machine-driver-hyperkit need root owner and uid
sudo chown root:wheel /usr/local/opt/docker-machine-driver-hyperkit/bin/docker-machine-driver-hyperkit
sudo chmod u+s /usr/local/opt/docker-machine-driver-hyperkit/bin/docker-machine-driver-hyperkit

Start Minikube

minikube start --cpus 4 --memory 8192

Helm

Install Helm locally on your machine.

brew install kubernetes-helm

Install Tiller Role-based Access Control Service Account

kubectl create -f tiller-rbac-config.yaml

Initialize Tiller on the cluster

helm init --service-account tiller --history-max 200

For Kubernetes 1.16 here's a known issue.

Use Chart

Install the repo:

helm repo add netifi https://download.netifi.com/helm-charts/

List available chart:

helm search netifi/

Example Deployments

Local Install:

helm install netifi/netifi-helm-charts -f ./setFiles/local.yaml

Open the minikube dashboard:

minikube dashboard

You should see a Pod with a single Netifi Broker node deployed in it.

Using the Kubernetes Cluster for local development

Go clone our Pinger example project.

git clone https://github.com/netifi/pinger.git pinger
cd pinger

Configure your shell to use the docker daemon running in minikube:

eval $(minikube docker-env)

In that same shell, build the ping and pong docker containers:

./gradlew dockerBuildImage

Find your minikube IP address so that you can properly deploy the containers to point to the IP address of the Netifi Broker:

minikube ip
# 192.168.64.8

Start a Pong service that connects statically to the Netifi Broker over the TCP connection, be sure the change the ADDRESSES value with the address of your minikube cluster:

kubectl run pong1 --image=netifi/pinger-pong --image-pull-policy=Never \
--env="NETIFI_CLIENT_DISCOVERY_ENVIRONMENT=static" \
--env="NETIFI_CLIENT_DISCOVERY_STATICPROPERTIES_CONNECTIONTYPE=tcp" \
--env="NETIFI_CLIENT_DISCOVERY_STATICPROPERTIES_ADDRESSES=192.168.64.8" \
--env="NETIFI_CLIENT_DISCOVERY_STATICPROPERTIES_PORT=8001"

Now start Ping service that connects statically to the Netifi Broker over the TCP connection, and again be sure to replace the ADDRESSES value with the address of your minikube cluster:

kubectl run ping1 --image=netifi/pinger-ping --image-pull-policy=Never \
--env="NETIFI_CLIENT_DISCOVERY_ENVIRONMENT=static" \
--env="NETIFI_CLIENT_DISCOVERY_STATICPROPERTIES_CONNECTIONTYPE=tcp" \
--env="NETIFI_CLIENT_DISCOVERY_STATICPROPERTIES_ADDRESSES=192.168.64.8" \
--env="NETIFI_CLIENT_DISCOVERY_STATICPROPERTIES_PORT=8001"

You can run:

minikube dashboard

to find the deployments and read the logs, or you can view the ping and pong endpoints by doing this:

kubectl port-forward deployment/pong1 8080
open http://localhost:8080/actuator/info

or

kubectl port-forward deployment/ping1 8081
open http://localhost:8081/actuator/info

By refreshing the pages, roughly every second, you should see the total counters go up.

Now let's get the name of the service account this tutorial created so that we can use it to dynamically discover the Netifi Broker our container should connect to:

kubectl get serviceaccounts
# NAME                       SECRETS   AGE
# default                    1         7m13s
# odd-dog-netifi-broker   1         5m10s

Now let's launch another Ping service using our Kubernetes service discovery mechanism. Be sure to substitute the serviceaccount and DEPLOYMENTNAME values with your service account name:

kubectl run ping2 --image=netifi/pinger-ping --image-pull-policy=Never --serviceaccount=odd-dog-netifi-broker \
--env="NETIFI_CLIENT_DISCOVERY_ENVIRONMENT=kubernetes" \
--env="NETIFI_CLIENT_DISCOVERY_KUBERNETESPROPERTIES_CONNECTIONTYPE=tcp" \
--env="NETIFI_CLIENT_DISCOVERY_KUBERNETESPROPERTIES_DEPLOYMENTNAME=odd-dog-netifi-broker"

You should see that the Pong service is now incrementing roughly twice as fast, while the Ping services are maintaining their individual rates.

Let's build these same Docker images on our local computer:

eval $(minikube docker-env -u)
./gradlew dockerBuildImage

And now start an additional Pong service:

docker run --rm -p 8080:8080 \
-e NETIFI_CLIENT_DISCOVERY_ENVIRONMENT=static \
-e NETIFI_CLIENT_DISCOVERY_STATICPROPERTIES_CONNECTIONTYPE=ws \
-e NETIFI_CLIENT_DISCOVERY_STATICPROPERTIES_ADDRESSES=192.168.64.8 \
-e NETIFI_CLIENT_DISCOVERY_STATICPROPERTIES_PORT=8101 \
netifi/pinger-pong:latest

You can now open: http://localhost:8080/pong to see this Pong service taking requests.

Now launch more Ping services:

for i in {1..5}
do
   docker run -d -P \
   -e NETIFI_CLIENT_DISCOVERY_ENVIRONMENT=static \
   -e NETIFI_CLIENT_DISCOVERY_STATICPROPERTIES_CONNECTIONTYPE=ws \
   -e NETIFI_CLIENT_DISCOVERY_STATICPROPERTIES_ADDRESSES=192.168.64.8 \
   -e NETIFI_CLIENT_DISCOVERY_STATICPROPERTIES_PORT=8101 \
   netifi/pinger-ping:latest
done

You can use docker ps to find the ephemeral ports, and launch their respective counter pages in a browser, or follow their logs, with docker logs -f <container id>

When you're done exploring, you can use these commands to remove everything:

docker rm -f $(docker ps -f "ancestor=netifi/pinger-ping:latest" -q)
docker rm -f $(docker ps -f "ancestor=netifi/pinger-pong:latest" -q)
kubectl delete deployment ping2
kubectl delete deployment ping1
kubectl delete deployment pong1
helm delete odd-dog
minikube delete

Using the Helm chart with GKE

Open Google Cloud Console and build yourself a Kubernetes cluster with 3 default nodes with the Kubernetes label: key: type and value: broker with 2 vCPUs each, and a second node pool of 3 nodes with the Kubernetes label: key: type and value: general with 2 vCPUs each. To access the Kubernetes labels you will need to click the 'More options' button and scroll down.

Then get the Kubernetes credentials installed on your local machine via the Google Cloud SDK:

gcloud container clusters get-credentials netifi-demo --zone us-central1-a

Install the Tiller RBAC policy:

kubectl create -f tiller-rbac-config.yaml

Initialize Tiller on the cluster

helm init --service-account tiller --history-max 200

And finally setup the and install the Netifi Helm Chart:

helm repo add netifi https://download.netifi.com/helm-charts/
helm install netifi/netifi-helm-charts -f ./setFiles/gkePublicWS.yaml

Now go to the GKE Console, open the Workload tab, and wait for the Netifi Broker cluster to come up. Now let's find our service account name and launch some Ping containers:

kubectl get serviceaccounts
# NAME                          SECRETS   AGE
# default                       1         61m
# rafting-quoll-netifi-broker   1         56m

Replace all the references to rafting-quoll-netifi-broker in the file: ./setFiles/gkePing.yaml with value of the service account you just found.

Now launch the Ping containers:

kubectl apply -f ./setFiles/gkePing.yaml
# replicaset.apps/ping-rs created

You should see 3 Ping containers launch and the logs should show that pings are being generated. Even though there is no Pong service yet, the requests are being buffered inside the application.

Use Kubectl to get the load balancer EXTERNAL-IP of the Netifi Broker cluster so that we can connect to it from our laptop:

kubectl get services
# NAME                          TYPE           CLUSTER-IP     EXTERNAL-IP     PORT(S)                                                                      AGE
# kubernetes                    ClusterIP      10.52.0.1      <none>          443/TCP                                                                      68m
# rafting-quoll-netifi-broker   LoadBalancer   10.52.15.197   34.67.191.203   8001:32660/TCP,8101:30443/TCP,7001:31100/TCP,6001:30060/TCP,9000:30623/TCP   62m

We can now start a Pong service from our laptop. Be sure to change the ADDRESSES value with your EXTERNAL-IP from the previous command:

docker run --rm -p 8080:8080 \
-e NETIFI_CLIENT_DISCOVERY_ENVIRONMENT=static \
-e NETIFI_CLIENT_DISCOVERY_STATICPROPERTIES_CONNECTIONTYPE=ws \
-e NETIFI_CLIENT_DISCOVERY_STATICPROPERTIES_ADDRESSES=34.67.191.203 \
-e NETIFI_CLIENT_DISCOVERY_STATICPROPERTIES_PORT=8101 \
netifi/pinger-pong:latest

You should see our local Pong service is able to process messages from the Ping services.

Now scale up the Ping services:

kubectl scale --replicas=91 rs/ping-rs

You should see that our local service is still able to keep up with the 91 Ping services. You'll need to provision more Kubernetes nodes if you want more Ping containers.

Welcome to the Netifi cloud-native application platform!

Releasing this chart package

curl -O https://download.netifi.com/helm-charts/index.yaml
helm package .
helm repo index --merge index.yaml .
AWS_PROFILE=netifi-prod  aws s3 cp --acl public-read ./netifi-helm-charts-0.0.4.tgz s3://download.netifi.com/helm-charts/
AWS_PROFILE=netifi-prod  aws s3 cp --acl public-read ./index.yaml s3://download.netifi.com/helm-charts/
# invalidate /helm-charts/index.yaml in cloudfront