anmoel / ansible-role-kubernetes

An ansible role to install an HA Kubernetes cluster

• 👋 Hi, my name is André Möller (@anmoel).
• I'm a cloud and devops consultant.
• 👀 I’m interested in cloud technologies.
• I have experiences in Azure and Google Cloud Platform.

ansible-role-kubernetes's Issues

ensure support for centos7

add update process

With this role, an existing cluster should be able to be updated.
This function must be implemented.

add tests for role

we need an testing pipeline to test all scenarios

how to skip etcd installation as a seprate server and make it to install on same master.

Hi Anmoel,

My request is with respect to the selection of configuration of servers. is it possible to make etcd install with api server rather than a separate server.

Note: I have not given a separate server for etcd instead , But rather i want to install on same master.

About "has no attribute" error

Hello,

I am up to 3 instance on gce using terraform. I'm trying to create a kubernetes cluster on these 3 nodes. Until ansible, everything works fine. Servers open, firewall rules, network is complete.

I used the example of allinone. I always get the following error. Where am I making a mistake?

cat inventory

#xxx=public ip adresses

[kubernetes:children]
etcd
k8s_masters
k8s_nodes

[kubernetes:vars]
docker_version="18.06.0-ce"

[etcd]
x.x.x.x ansible_ssh_user=testuser
x.x.x.x ansible_ssh_user=testuser
x.x.x.x ansible_ssh_user=testuser

[k8s_masters]
x.x.x.x ansible_ssh_user=testuser

[k8s_nodes]
x.x.x.x ansible_ssh_user=testuser
x.x.x.x ansible_ssh_user=testuser

cat playbook.yml

- hosts: kubernetes
#  sudo: yes
  become: true
  become_user: root
  gather_facts: false
  tasks:
  - name: install python 2
    raw: test -e /usr/bin/python || (apt -y update && apt install -y python-minimal)
    changed_when: false

- hosts: kubernetes
#  sudo: yes
  become: true
  become_user: root
  roles:
#    - geerlingguy.docker
    - anmoel.kubernetes

my command;
export ANSIBLE_HOST_KEY_CHECKING=False && time ansible-playbook -i inventory playbook.yml -v

error message;

TASK [anmoel.kubernetes : /etc/kubernetes/kubeadm/clusterconfig.yaml] **********************************************************************************************************************************************************************************************************
fatal: [x.x.x.x]: FAILED! => {"changed": false, "msg": "AnsibleUndefinedVariable: 'ansible.vars.hostvars.HostVarsVars object' has no attribute 'ansible_host'"}
fatal: [x.x.x.x]: FAILED! => {"changed": false, "msg": "AnsibleUndefinedVariable: 'ansible.vars.hostvars.HostVarsVars object' has no attribute 'ansible_host'"}
fatal: [x.x.x.x]: FAILED! => {"changed": false, "msg": "AnsibleUndefinedVariable: 'ansible.vars.hostvars.HostVarsVars object' has no attribute 'ansible_host'"}
	to retry, use: --limit @.../playbook.retry


PLAY RECAP *********************************************************************************************************************************************************************************************************************************************************************
x.x.x.x             : ok=30   changed=14    unreachable=0    failed=1
x.x.x.x             : ok=30   changed=14    unreachable=0    failed=1
x.x.x.x             : ok=30   changed=14    unreachable=0    failed=1

Add documentation

use preemptible vms

Unable to connect to the server: x509: certificate is valid for [fixed SAN ips]

After the installation was successful, the allowed host SANs are set to kubernetes_etcd_*_cert_sans.
1.)
This prevents us from setting further SAN entries (e.g for other remote control hosts)
2.)
Maybe we should split etcd and kubeadm anyway, as we might have seperate certs for etcd later on.

add seperated etcd_nodes to the cluster

if the etcd nodes are seperated nodes, the kubernetes cluster don't see this nodes.
this nodes must be join into the cluster with an taint.

Ensure swap is disable on all nodes

Swap should be disabled when running kubernetes.
The play should check if swap is enabled and disable it if necessary.

Otherwise using your role might result in:

F1129 19:04:46.741210 8662 server.go:262] failed to run Kubelet: Running with swap on is not supported, please disable swap! or set --fail-swap-on flag to false.

ensure support for debian9