EKS Getting Started Guide Configuration

I took that from https://github.com/terraform-providers/terraform-provider-aws/tree/master/examples/eks-getting-started

This is the full configuration from https://www.terraform.io/docs/providers/aws/guides/eks-getting-started.html

Setup EKS via Terraform

1. terraform apply

terraformoutputs something like

config_map_aws_auth = 

apiVersion: v1
... 

kubeconfig = 

apiVersion: v1
clusters:
... 

copy the block after config_map_aws_auth into the file configmap.yml

copy the block after kubeconfig into the file ~/.kube/config-aws

I have setup my KUBECONFIG environment variable to support two config files, so that an additional file is supported

KUBECONFIG=/home/christian/.kube/config:/home/christian/.kube/config-aws

Use the new kubeconfig in config-aws with this command

kubectl config use-context aws

apply the configmap file

kubectl apply -f configmap.yml 

Check if the nodes come up

kubectl get no --watch

Create Kubernetes Dashboard

execute file createDashboard.sh to deploy the dashboard, heapster and the necessary influx. The scripts start kubectl proxy and displays the url and access token to access the dashboard

./createDashboard.sh

https://docs.aws.amazon.com/eks/latest/userguide/dashboard-tutorial.html

Create Ingress Controller

cd ingress

execute file createNGINXIngressController.sh to install an ingress controller in aws

this creates a new namespace ingress-nginx with a deployment of the image quay.io/kubernetes-ingress-controller/nginx-ingress-controller. the ingress controller automatically creates an aws load balancer for external access

./createNGINXIngressController.sh

https://kubernetes.github.io/ingress-nginx/deploy/

https://github.com/kubernetes/ingress-nginx

Install some services and corresponding ingresses

I`m experimenting with services and ingress, this is just some wip to see how I can deploy and access services

This is with an ingrewss for sveral services

kubectl apply -f echo1.yml -f echo2.yml -f ingress_echo1-2.yml 

And this command is for a service with a dedicated ingress resource.

kubectl apply -f echo3withIngress.yml 

Looks like everything works together.

More than one Ingress Controller

You can have more than one ingress controllers, e.g. when you want have ingresses mapped to internal and external loadbalancers.

To do that, first create an internal load balancer with

kubectl apply -f internal-lb.yml

The trick is the annotation service.beta.kubernetes.io/aws-load-balancer-internal: 0.0.0.0/0 that tells aws to create an internal loadbalancer.

Then create a new Ingress Controller with

kubectl apply  -f nginx-controller-deployment.yml

The argument --ingress-class=ingress-nginx-internal tells the ingress controller that it should only treat ingresses that specify

kubernetes.io/ingress.class: "nginx-internal"

Now, deploy an internal ingres with

kubectl apply -f echo4InternalIngress.yml 

and you should see the external loadBalancer echo3 as well as the internal loadBalancer

kubectl get ingress
NAME      HOSTS     ADDRESS                                                                   PORTS     AGE
echo3     *         aa9133a67d45411e8b9e902cc5740c06-1837280603.eu-west-1.elb.amazonaws.com   80        4m
echo4     *                                                                                   80        55s

https://docs.giantswarm.io/guides/services-of-type-loadbalancer-and-multiple-ingress-controllers/

Helm

INstall Helm https://github.com/helm/helm/releases

kubectl create clusterrolebinding add-on-cluster-admin --clusterrole=cluster-admin --serviceaccount=kube-system:default helm ls helm install stable/mysql

##storage EKSnsupports EBS out of the box. To create a default storage class execute

cd storage
kubectl apply -f storageclass.yml -f storageclass_cheap.yml

after that new pv's and pvc's can be create with the command

kubectl apply -f pvc.yml 
kubectl get pvc --watch
kubectl get pc --watch

and mount this pv into a new container:

kubectl apply -f pod.yml 

myserver: kubectl exec -it mypod bash
root@mypod:/# mount | grep /var/www/html
/dev/xvdbd on /var/www/html type ext4 (rw,relatime,debug,data=ordered)
root@mypod:/# root@mypod:/# df -h | grep /var/www/html
/dev/xvdbd      5.8G   24M  5.5G   1% /var/www/html
root@mypod:/# 

everything taken fro here: https://docs.giantswarm.io/guides/using-persistent-volumes-on-aws/

https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/EBSVolumeTypes.html

https://docs.aws.amazon.com/eks/latest/userguide/storage-classes.html

efs

https://github.com/kubernetes-incubator/external-storage/tree/master/aws/efs

helm

helm init

# give tiller cluster-admin role.... not the best idea for production

kubectl create serviceaccount --namespace kube-system tiller
kubectl create clusterrolebinding tiller-cluster-rule --clusterrole=cluster-admin --serviceaccount=kube-system:tiller
kubectl patch deploy --namespace kube-system tiller-deploy -p '{"spec":{"template":{"spec":{"serviceAccount":"tiller"}}}}'

helm repo add incubator http://storage.googleapis.com/kubernetes-charts-incubator

helm list
helm delete ...

filebeat, elastic

see https://github.com/elastic/beats/tree/master/deploy/kubernetes

and https://github.com/elastic/beats/blob/master/deploy/kubernetes/filebeat-kubernetes.yaml

cd filebeat
kubectl apply -f filebeat-kubernetes.yaml

Before installating elasticsearch, install a storage class (see storage)

helm install stable/elasticsearch
helm install stable/kibana
helm install --name chrissi-elastic incubator/elastic-stack --set logstash.enabled=false,kibana.env.ELASTICSEARCH_URL=http://chrissi-elastic-elasticsearch-client.default.svc.cluster.local:9200