heat-kubeadm

Heat stack template for installing Kubernetes with kubeadm.

Installation

On the machine that will be used to install this stack we need following packages:

python-openstackclient - for general OpenStack functions
python-heatclient - for heat-specific CLI functions
kubectl - to access Kubernetes cluster
helm - to install Spinnaker on top of Kubernetes

Basic steps are:

Ensure all necessary Floating IPs are created
Prepare template config
Create stack
Wait for it to finish
Install Spinnaker with Helm

Floating IPs

We assume that floating IPs with necessary parameters are already created.

First you need to find out IDs of these floating IPs:

$ openstack floating ip list
+--------------------------------------+---------------------+------------------+------+--------------------------------------+----------------------------------+
| ID                                   | Floating IP Address | Fixed IP Address | Port | Floating Network                     | Project                          |
+--------------------------------------+---------------------+------------------+------+--------------------------------------+----------------------------------+
| 24776f99-7599-4bdd-8a1a-1a551180eb4a | 172.17.50.93        | None             | None | 3e868882-d59e-416a-90a1-48cc04cab723 | bf35dfbf572e41dab0ceb9b8085da879 |
| 851660f0-3648-4560-9403-c68647e4aa99 | 172.17.49.222       | None             | None | 3e868882-d59e-416a-90a1-48cc04cab723 | bf35dfbf572e41dab0ceb9b8085da879 |
| 32cc96ef-8590-4785-b660-c6e773c97f3d | 172.17.48.254       | None             | None | 3e868882-d59e-416a-90a1-48cc04cab723 | bf35dfbf572e41dab0ceb9b8085da879 |
| 6c9d42cb-ad7e-46dd-aca8-9a5cc628e5de | 172.17.49.237       | None             | None | 3e868882-d59e-416a-90a1-48cc04cab723 | bf35dfbf572e41dab0ceb9b8085da879 |
+--------------------------------------+---------------------+------------------+------+--------------------------------------+----------------------------------+

Your values here will differ. You need to select IP addresses that you are going to use for your cluster.

Prepare template config

Here is an example of environment file (lets call it env.yaml) that can be used to create the stack:

parameters:
    key_name: mysshkey
    image: ubuntu-16-04-amd64-cloudimg    
    master_floating_ip: 24776f99-7599-4bdd-8a1a-1a551180eb4a
    slave_floating_ips: 851660f0-3648-4560-9403-c68647e4aa99,32cc96ef-8590-4785-b660-c6e773c97f3d,6c9d42cb-ad7e-46dd-aca8-9a5cc628e5de
    availability_zone: az1
    public_network_id: extnet

Here you must specify ID of one floating IP for master in master_floating_ip and 3 comma-separated IDs of floating IPs for slaves in slave_floating_ips.

You must also specify name of your SSH key in OpenStack in key_name, name or ID of image to use for nodes in image and availability zone for VMs in availability_zone.

If you need to use proxy with your environment, you can add these options:

    proxy_host: myproxy.local
    proxy_port: 8080

All other options and their descriptions can be found at the top of stack_full.yaml file.

Note that if you would like to deploy stack in existing network, you can specify internal_net parameter in environment file and then use template stack.yaml.

Create Heat stack

To create Heat stack, issue command:

$ openstack stack create -t stack_full.yaml -e env.yaml teststack --wait                                                                                        ~/src/github.com/YorikSar/heat-kubeadm

If you do specify --wait flag, output should look like this:

2018-02-02 19:01:57Z [teststack]: CREATE_IN_PROGRESS  Stack CREATE started
2018-02-02 19:01:57Z [teststack.random_string]: CREATE_IN_PROGRESS  state changed
2018-02-02 19:01:58Z [teststack.random_string]: CREATE_COMPLETE  state changed
2018-02-02 19:01:58Z [teststack.prefix_random]: CREATE_IN_PROGRESS  state changed
2018-02-02 19:01:58Z [teststack.prefix_random]: CREATE_COMPLETE  state changed
2018-02-02 19:01:58Z [teststack.internal_net]: CREATE_IN_PROGRESS  state changed
2018-02-02 19:01:58Z [teststack.internal_net]: CREATE_COMPLETE  state changed
2018-02-02 19:01:59Z [teststack.internal_router]: CREATE_IN_PROGRESS  state changed
2018-02-02 19:01:59Z [teststack.internal_subnet]: CREATE_IN_PROGRESS  state changed
2018-02-02 19:01:59Z [teststack.internal_router]: CREATE_COMPLETE  state changed
2018-02-02 19:01:59Z [teststack.internal_subnet]: CREATE_COMPLETE  state changed
2018-02-02 19:01:59Z [teststack.stack]: CREATE_IN_PROGRESS  state changed
2018-02-02 19:02:00Z [teststack.internal_router_interface]: CREATE_IN_PROGRESS  state changed
2018-02-02 19:02:00Z [teststack.internal_router_interface]: CREATE_COMPLETE  state changed
2018-02-02 19:02:35Z [teststack.stack]: CREATE_COMPLETE  state changed
2018-02-02 19:02:35Z [teststack]: CREATE_COMPLETE  Stack CREATE completed successfully
+---------------------+----------------------------------------+
| Field               | Value                                  |
+---------------------+----------------------------------------+
| id                  | 5957e776-d630-4072-bdbb-df69303b5d9f   |
| stack_name          | teststack                              |
| description         | Deploy Kubernetes cluster with kubeadm |
| creation_time       | 2018-02-02T19:01:57Z                   |
| updated_time        | None                                   |
| stack_status        | CREATE_COMPLETE                        |
| stack_status_reason | Stack CREATE completed successfully    |
+---------------------+----------------------------------------+

The line [teststack]: CREATE_COMPLETE Stack CREATE completed successfully means that stack has been created successfully.

Wait for Kubernetes to be installed

After stack creation is completed, you can SSH to master node with key you've specified in environment file:

ssh -i mysshkey ubuntu@172.17.50.93

There you can check if Kubernetes deploymentis is in progress:

$ journalctl -fu kubeadm-install

This command will output lines from log of kubernetes installation. Last lines are expected to be like these:

Feb 05 20:32:41 k8s-xnpetidzawduzisz-master sudo[5850]: pam_unix(sudo:session): session closed for user ubuntu
Feb 05 20:32:41 k8s-xnpetidzawduzisz-master bash[3199]: + cp /root/spinnaker.yaml /home/ubuntu/spinnaker.yaml
Feb 05 20:32:41 k8s-xnpetidzawduzisz-master bash[3199]: + chown ubuntu:ubuntu /home/ubuntu/spinnaker.yaml

Now you can verify that kubernetes is up:

$ kubectl get nodes
NAME                   STATUS    AGE       VERSION
k8s-qyyenxoh-master    Ready     17m       v1.8.7
k8s-qyyenxoh-slave-0   Ready     16m       v1.8.7
k8s-qyyenxoh-slave-1   Ready     16m       v1.8.7
k8s-qyyenxoh-slave-2   Ready     16m       v1.8.7

If you don't see all nodes there or some of them are in NotReady state, it means that cluster is not up yet. You should repeat kubectl get nodes again untill you see all nodes in Ready state.

Install Spinnaker with Helm

Tiller (Helm server component) is already installed on the cluster, and client is configured. Now you can install Spinnaker from Helm chart:

$ helm install -f spinnaker.yaml --name spinnaker spinnaker-3.1.0.tgz --wait

It will run for some time and then output information about created release object as well as all resources created in Kubernetes for it.

After it finishes, you can determine NodePort assigned to its UI using:

$ kubectl get --namespace default -o jsonpath="{.spec.ports[0].nodePort}" services spinnaker-spinnaker-deck
31860

Now you can use it with one of your floating IPs to see Spinnaker UI in your browser like this: http://172.17.50.93:31860

Jenkins deployed together with Spinnaker could be accessed using NodePort as well:

$ kubectl get --namespace spinnaker -o jsonpath="{.spec.ports[0].nodePort}" services jenkins-spinnaker
30991

foexle / heat-kubeadm