c12s-kubespray/docs/getting-started.md
Matthew Mosesohn a44235d11b Refactor remove node to allow removing dead nodes and etcd members (#5009)
Change-Id: I1c59249f08f16d0f6fd60df6ab61f17a0a7df189
2019-08-07 04:46:50 -07:00

6.5 KiB

Getting started

Building your own inventory

Ansible inventory can be stored in 3 formats: YAML, JSON, or INI-like. There is an example inventory located here.

You can use an inventory generator to create or modify an Ansible inventory. Currently, it is limited in functionality and is only used for configuring a basic Kubespray cluster inventory, but it does support creating inventory file for large clusters as well. It now supports separated ETCD and Kubernetes master roles from node role if the size exceeds a certain threshold. Run python3 contrib/inventory_builder/inventory.py help help for more information.

Example inventory generator usage:

cp -r inventory/sample inventory/mycluster
declare -a IPS=(10.10.1.3 10.10.1.4 10.10.1.5)
CONFIG_FILE=inventory/mycluster/hosts.yml python3 contrib/inventory_builder/inventory.py ${IPS[@]}

Then use inventory/mycluster/hosts.yml as inventory file.

Starting custom deployment

Once you have an inventory, you may want to customize deployment data vars and start the deployment:

IMPORTANT: Edit my_inventory/groups_vars/*.yaml to override data vars:

ansible-playbook -i inventory/mycluster/hosts.yml cluster.yml -b -v \
  --private-key=~/.ssh/private_key

See more details in the ansible guide.

Adding nodes

You may want to add worker, master or etcd nodes to your existing cluster. This can be done by re-running the cluster.yml playbook, or you can target the bare minimum needed to get kubelet installed on the worker and talking to your masters. This is especially helpful when doing something like autoscaling your clusters.

  • Add the new worker node to your inventory in the appropriate group (or utilize a dynamic inventory).

  • Run the ansible-playbook command, substituting cluster.yml for scale.yml:

    ansible-playbook -i inventory/mycluster/hosts.yml scale.yml -b -v \
      --private-key=~/.ssh/private_key
    

Remove nodes

You may want to remove master, worker, or etcd nodes from your existing cluster. This can be done by re-running the remove-node.yml playbook. First, all specified nodes will be drained, then stop some kubernetes services and delete some certificates, and finally execute the kubectl command to delete these nodes. This can be combined with the add node function. This is generally helpful when doing something like autoscaling your clusters. Of course, if a node is not working, you can remove the node and install it again.

Use --extra-vars "node=<nodename>,<nodename2>" to select the node(s) you want to delete.

ansible-playbook -i inventory/mycluster/hosts.yml remove-node.yml -b -v \
  --private-key=~/.ssh/private_key \
  --extra-vars "node=nodename,nodename2"

If a node is completely unreachable by ssh, add --extra-vars reset_nodes=no to skip the node reset step. If one node is unavailable, but others you wish to remove are able to connect via SSH, you could set reset_nodes=no as a host var in inventory.

Connecting to Kubernetes

By default, Kubespray configures kube-master hosts with insecure access to kube-apiserver via port 8080. A kubeconfig file is not necessary in this case, because kubectl will use http://localhost:8080 to connect. The kubeconfig files generated will point to localhost (on kube-masters) and kube-node hosts will connect either to a localhost nginx proxy or to a loadbalancer if configured. More details on this process are in the HA guide.

Kubespray permits connecting to the cluster remotely on any IP of any kube-master host on port 6443 by default. However, this requires authentication. One could generate a kubeconfig based on one installed kube-master hosts (needs improvement) or connect with a username and password. By default, a user with admin rights is created, named kube. The password can be viewed after deployment by looking at the file {{ credentials_dir }}/kube_user.creds (credentials_dir is set to {{ inventory_dir }}/credentials by default). This contains a randomly generated password. If you wish to set your own password, just precreate/modify this file yourself.

For more information on kubeconfig and accessing a Kubernetes cluster, refer to the Kubernetes documentation.

Accessing Kubernetes Dashboard

As of kubernetes-dashboard v1.7.x:

  • New login options that use apiserver auth proxying of token/basic/kubeconfig by default
  • Requires RBAC in authorization_modes
  • Only serves over https
  • No longer available at https://first_master:6443/ui until apiserver is updated with the https proxy URL

If the variable dashboard_enabled is set (default is true), then you can access the Kubernetes Dashboard at the following URL, You will be prompted for credentials: https://first_master:6443/api/v1/namespaces/kube-system/services/https:kubernetes-dashboard:/proxy/#!/login

Or you can run 'kubectl proxy' from your local machine to access dashboard in your browser from: http://localhost:8001/api/v1/namespaces/kube-system/services/https:kubernetes-dashboard:/proxy/#!/login

It is recommended to access dashboard from behind a gateway (like Ingress Controller) that enforces an authentication token. Details and other access options here: https://github.com/kubernetes/dashboard/wiki/Accessing-Dashboard---1.7.X-and-above

Accessing Kubernetes API

The main client of Kubernetes is kubectl. It is installed on each kube-master host and can optionally be configured on your ansible host by setting kubectl_localhost: true and kubeconfig_localhost: true in the configuration:

  • If kubectl_localhost enabled, kubectl will download onto /usr/local/bin/ and setup with bash completion. A helper script inventory/mycluster/artifacts/kubectl.sh also created for setup with below admin.conf.
  • If kubeconfig_localhost enabled admin.conf will appear in the inventory/mycluster/artifacts/ directory after deployment.

You can see a list of nodes by running the following commands:

cd inventory/mycluster/artifacts
./kubectl.sh get nodes

If desired, copy admin.conf to ~/.kube/config.