This is a kops wrapper script for creating Kubernetes clusters on AWS and maintaining them. Certain addons are also installed into newly created clusters:

• Heapster
• Kubernetes Dashboard
• Elasticsearch/Fluentd/Kibana (EFK) logging stack
• Prometheus Operator monitoring stack

The cluster is additionally set up with RBAC authorization for heightened security.

We use Python 3. To install dependencies, issue the following command at the root of this project: pip3 install -r requirements.txt.

Write your secrets in secrets.yaml, see secrets.yaml.sample.

The ./kops script supports a few subcommands and a number of arguments:

The create subcommand allows you to create a Kubernetes cluster within your AWS account.

Example:

./kops --aws_profile example create app staging [email protected] email-smtp.eu-west-1.amazonaws.com:587 example.com clusters.example.com

In order to access Kibana, start kubectl --kubeconfig $KUBECONFIG proxy 8001 and open http://localhost:8001/api/v1/proxy/namespaces/kube-system/services/kibana-logging in your browser. The username/password is elastic/changeme.

We use Prometheus, via CoreOS' Prometheus Operator, as our monitoring solution. In addition to Prometheus Operator itself, we install the whole kube-prometheus stack that among other things ensures that the Kubernetes cluster itself and the host infrastructure are monitored.

When upgrading to a newer version of the Prometheus Operator stack, we need to first upgrade Prometheus Operator itself (by applying its manifests). Once this is ready, we can apply the manifests for the rest of the stack. Prometheus Operator will take care of upgrading corresponding components in the cluster.

We use Grafana for visualizing monitoring information.