mrlesmithjr / ansible-k8s

ansible-k8s

An Ansible role to deploy a Kubernetes - K8s Cluster

• Weave-Net is used for the network overlay currently

Requirements

Install additional required Ansible roles:

sudo ansible-galaxy install -r requirements.yml

Role Variables

defaults/main.yml

Dependencies

None

Example Playbook

Vagrant/playbook.yml

Vagrant

To spin up a test cluster using Vagrant. Check out the README.

License

MIT

Author Information

Larry Smith Jr.

• EverythingShouldBeVirtual
• @mrlesmithjr
• mailto:[email protected]

ansible-k8s's Issues

Add ability to pin a specific version of K8s

We need to add the ability to pin a specific version of each kubeadm, kubectl, and kubelet in order to install specific versions.

Add ability to add add'l worker nodes after cluster is provisioned

Currently there is not a way to add add'l worker nodes to the cluster after being provisioned. This needs to be a thing.

Convert Ansible roles for Vagrant to submodules

Need to convert all Ansible roles used for Vagrant testing to submodules. This will bring a more efficient way of handling the required roles going forward.

Joining cluster provisioning issue

Currently, there is an issue when provisioning the cluster. We need to investigate this.

TASK [ansible-k8s : join_cluster | Joining Additional Nodes To K8s Cluster] ****
fatal: [node1]: FAILED! => {"changed": true, "cmd": ["kubeadm", "join", "--token", "ff377f.b1385ee1e877c22e", "192.168.250.10:6443"], "delta": "0:00:00.239864", "end": "2017-12-31 14:04:57.355427", "msg": "non-zero return code", "rc": 3, "start": "2017-12-31 14:04:57.115563", "stderr": "\t[WARNING FileExisting-crictl]: crictl not found in system path\ndiscovery: Invalid value: \"\": using token-based discovery without DiscoveryTokenCACertHashes can be unsafe. set --discovery-token-unsafe-skip-ca-verification to continue", "stderr_lines": ["\t[WARNING FileExisting-crictl]: crictl not found in system path", "discovery: Invalid value: \"\": using token-based discovery without DiscoveryTokenCACertHashes can be unsafe. set --discovery-token-unsafe-skip-ca-verification to continue"], "stdout": "[preflight] Running pre-flight checks.", "stdout_lines": ["[preflight] Running pre-flight checks."]}
fatal: [node2]: FAILED! => {"changed": true, "cmd": ["kubeadm", "join", "--token", "ff377f.b1385ee1e877c22e", "192.168.250.10:6443"], "delta": "0:00:00.238416", "end": "2017-12-31 14:04:57.341109", "msg": "non-zero return code", "rc": 3, "start": "2017-12-31 14:04:57.102693", "stderr": "\t[WARNING FileExisting-crictl]: crictl not found in system path\ndiscovery: Invalid value: \"\": using token-based discovery without DiscoveryTokenCACertHashes can be unsafe. set --discovery-token-unsafe-skip-ca-verification to continue", "stderr_lines": ["\t[WARNING FileExisting-crictl]: crictl not found in system path", "discovery: Invalid value: \"\": using token-based discovery without DiscoveryTokenCACertHashes can be unsafe. set --discovery-token-unsafe-skip-ca-verification to continue"], "stdout": "[preflight] Running pre-flight checks.", "stdout_lines": ["[preflight] Running pre-flight checks."]}

Allow DNS for control plane endpoint

We need to allow specifying a DNS name for the control plane endpoint. This SHOULD resolve issues where controller node IP addresses might change.

https://kubernetes.io/docs/setup/production-environment/tools/kubeadm/create-cluster-kubeadm/#considerations-about-apiserver-advertise-address-and-controlplaneendpoint