Kolla Kubernetes Development Environment

Overview

The kolla-kubernetes development environment is intended to run on a dedicated development machine such as a workstation or laptop. This development environment is not intended to run on a virtual machine although that is feasible. Following this guide will have a minimal impact to the host operating system. Some software and libraries will be installed and some configuration changes will be required.

Install Vagrant and Ansible

You can use Halcyon-Vagrant-Kubernetes with the VirtualBox, Libvirt or OpenStack vagrant providers. The documentation here describes the Libvirt provider for Linux hosts, but VirtualBox is perfectly acceptable as well if preferred. For more information about Halcyon-Kubernetes, please refer to the Github repositories:

Note

Currently, the following versions are tested and required:
  • ansible >= 2.2.0
  • helm >= 2.1.3
  • kubernetes >= 1.5.2
  • vagrant <1.9.0

Note

The official Ubuntu image is currently incompatible with the vagrant-libvirt provider, but works without issues using either the VirtualBox or OpenStack providers.

CentOS 7.2 with Libvirt

Firstly install Vagrant:

sudo yum install -y \
     https://releases.hashicorp.com/vagrant/1.8.1/vagrant_1.8.1_x86_64.rpm

Then install the deps for vagrant libvirt and ensure git-review is present:

sudo yum install -y libvirt \
                    libxslt-devel \
                    libxml2-devel \
                    libvirt-devel \
                    libguestfs-tools-c \
                    ruby-devel \
                    gcc \
                    git \
                    git-review \
                    gcc-c++

Now we can install the libvirt plugin itself:

vagrant plugin install vagrant-libvirt

Now you can setup Libvirt for use without requiring root privileges:

sudo bash -c 'cat << EOF > /etc/polkit-1/rules.d/80-libvirt-manage.rules
polkit.addRule(function(action, subject) {
if (action.id == "org.libvirt.unix.manage" && subject.local && subject.active && subject.isInGroup("wheel")) {
  return polkit.Result.YES;
}
});
EOF'

sudo usermod -aG libvirt $USER

Once both Libvirt and Vagrant have been prepared, you should now start and enable Libvirt:

sudo systemctl start libvirtd
sudo systemctl enable libvirtd

Finally install Ansible to allow Halcyon Kubernetes to provision the cluster:

sudo yum install -y epel-release
sudo yum install -y ansible

Before continuing, log out and back in again for your session to have the correct permissions applied.

Ubuntu 16.04 with Libvirt

Firstly install Vagrant:

sudo apt-get update
# Note that theres is a packaging bug in ubuntu so the upstream package must
# be used: https://github.com/vagrant-libvirt/vagrant-libvirt/issues/575
curl -L https://releases.hashicorp.com/vagrant/1.8.1/vagrant_1.8.1_x86_64.deb > /tmp/vagrant_1.8.1_x86_64.deb
sudo apt-get -y install /tmp/vagrant_1.8.1_x86_64.deb

Then install the dependencies for vagrant-libvirt and ensure git-review is present:

sudo sed -i 's/^# deb-src/deb-src/g' /etc/apt/sources.list
sudo apt-get update
sudo apt-get -y build-dep vagrant ruby-libvirt
sudo apt-get install -y \
             qemu-kvm \
             libvirt-bin \
             ebtables \
             dnsmasq \
             libxslt-dev \
             libxml2-dev \
             libvirt-dev \
             zlib1g-dev \
             ruby-dev \
             git \
             git-review \
             g++ \
             qemu-utils

Now we can install the libvirt plugin itself:

vagrant plugin install vagrant-libvirt

Now you can setup Libvirt for use without requiring root privileges:

sudo adduser $USER libvirtd

Finally, install Ansible to allow Halcyon Kubernetes to provision the cluster:

sudo apt-get install -y software-properties-common
sudo apt-add-repository -y ppa:ansible/ansible
sudo apt-get update
sudo apt-get install -y ansible

Before continuing, log out and back in again for your session to have the correct permissions applied.

MacOS

Install the CLI Developer tools by opening a terminal and running:

xcode-select --install
Download and install VirtualBox from:
Download and install vagrant using the following url to obtain the package:

There is a bug in Vagrant 1.8.7’s embedded curl that prevents boxes being downloaded, as described in: https://github.com/mitchellh/vagrant/issues/7997. This can be resolved by running the following command:

sudo rm -f /opt/vagrant/embedded/bin/curl

If your version of MacOS doesn’t not include git in the CLI Developer tools installed above, you can download and install git from:

Now we can install Ansible:

easy_install --user pip
printf 'if [ -f ~/.bashrc ]; then\n  source ~/.bashrc\nfi\n' >> $HOME/.profile
printf 'export PATH=$PATH:$HOME/Library/Python/2.7/bin\n' >> $HOME/.bashrc
source $HOME/.profile
pip install --user --upgrade ansible
sudo mkdir /etc/ansible
sudo curl -L https://raw.githubusercontent.com/ansible/ansible/devel/examples/ansible.cfg -o /etc/ansible/ansible.cfg

Note

Under MacOS, you may encounter an error during vagrant up, complaining that too many files are open. This is as recent versions of MacOS limit the number of file descriptors per application to 200. A simple way to resolve this is by running ulimit -n 4048 from the CLI before bringing the environment up.

Set environment variable versions for easy copy and pasting

The contents of this document are designed to be easily copy and pasted into a terminal window. To facilitate the maintenance of this document, we use environment variables to specify the helm and kubernetes versions under use.

export HELM_VERSION=2.1.3
export KUBERNETES_VERSION=1.5.1

Install Kubernetes and Helm clients

To complete the development environment setup, it is mandatory to have both a kubernetes client (kubectl) and a helm client (helm) installed on the host operating system.

Installing Clients on CentOS or Ubuntu

To install the kubernetes clients:

curl -L https://dl.k8s.io/v$KUBERNETES_VERSION/kubernetes-client-linux-amd64.tar.gz | tar -xzv
sudo cp kubernetes/client/bin/* /usr/local/bin
sudo chmod 755 /usr/local/bin/kubefed /usr/local/bin/kubectl
sudo chown root: /usr/local/bin/kubefed /usr/local/bin/kubectl

To install the helm client:

curl -L https://storage.googleapis.com/kubernetes-helm/helm-v$HELM_VERSION-linux-amd64.tar.gz | tar -xzv
sudo cp linux-amd64/helm /usr/local/bin/helm
sudo chmod 755 /usr/local/bin/helm
sudo chown root: /usr/local/bin/helm

Installing Clients on MacOS

To install the kubernetes clients:

curl -L https://dl.k8s.io/v$KUBERNETES_VERSION/kubernetes-client-darwin-amd64.tar.gz | tar -xzv
sudo cp kubernetes/client/bin/* /usr/local/bin
sudo chmod 755 /usr/local/bin/kubefed /usr/local/bin/kubectl
sudo chown root: /usr/local/bin/kubefed /usr/local/bin/kubectl

To install the helm client:

curl -L https://storage.googleapis.com/kubernetes-helm/helm-v$HELM_VERSION-darwin-amd64.tar.gz | tar -xzv
sudo cp darwin-amd64/helm /usr/local/bin/helm
sudo chmod 755 /usr/local/bin/helm
sudo chown root: /usr/local/bin/helm

Setup environment

Clone the repo containing the dev environment:

git clone https://github.com/att-comdev/halcyon-vagrant-kubernetes

Initialize the `halcyon-vagrant-kubernetes` repository:

cd halcyon-vagrant-kubernetes
git submodule init
git submodule update

You can then setup Halcyon Vagrant for Kolla. You can select either centos or ubuntu as a guest operating system though currently Ubuntu is only supported by the Vagrant VirtualBox and OpenStack providers.

./setup-halcyon.sh \
    --k8s-config kolla \
    --k8s-version v$KUBERNETES_VERSION \
    --guest-os centos

Note

If you need to use a proxy then you should also edit the config.rb file as follows:

  • Set proxy_enable = true
  • Set proxy_http and proxy_https values for your proxy
  • Configure proxy_no as appropriate. proxy_no should also include the ip’s of all kube cluster members. (i.e. 172.16.35.11,172.16.35.12,172.16.35.13,172.16.35.14)
  • Edit the no_proxy environment variable on your host to include the kube master IP (172.16.35.11)

Managing and interacting with the environment

The kube2 system in your halcyon-vagrant environment should have a minimum of 4gb of ram and all others should be set to 2gb of ram. In your config.rb script kube_vcpus should be set to 2 and kube_count should be set to 4.

Once the environment’s dependencies have been resolved and configuration completed, you can run the following commands to interact with it:

 vagrant up         # To create and start your halcyon-kubernetes cluster.
                    # You can also use --provider=libvirt

 ./get-k8s-creds.sh # To get the k8s credentials for the cluster and setup
                    # kubectl on your host to access it, if you have the helm
                    # client installed on your host this script will also set
                    # up the client to enable you to perform all development
                    # outside of the cluster.

vagrant ssh kube1   # To ssh into the master node.

vagrant destroy     # To make it all go away.

Note that it will take a few minutes for everything to be operational, typically between 2-5 mins after vagrant/ansible has finished for all services to be online for my machine (Xeon E3-1240 v3, 32GB, SSD), primarily dependent on network performance. This is as it takes time for the images to be pulled, and CNI networking to come up, DNS being usually the last service to become active.

Testing the deployed environment

Test everything works by starting a container with an interactive terminal:

kubectl run -i -t $(uuidgen) --image=busybox --restart=Never

Once that pod has started and your terminal has connected to it, you can then test the Kubernetes DNS service (and by extension the CNI SDN layer) by running:

nslookup kubernetes

To test that helm is working you can run the following:

helm init --client-only
helm repo update
helm install stable/memcached --name helm-test
# check the deployment has succeded
helm ls
# and to check via kubectl
kubectl get all
# and finally remove the test memcached chart
helm delete helm-test --purge

Note

If you receive the error `Error: could not find a ready tiller pod` helm is likely pulling the image to the kubernetes cluster. This error may also be returned if you have a proxy server environment and the development environment is not setup properly for the proxy server.

Containerized development environment requirements and usage

Make sure to run the ./get-k8s-creds.sh script or the development environment container will not be able to connect to the vagrant kubernetes cluster.

The kolla-kubernetes and kolla-ansible project should be checked out into the same base directory as halcyon-vagrant-kubernetes. The default assumed in kolla-kubernetes/tools/Dockerfile is ~/devel. If that is not the case in your environment then change that value in Dockerfile.

git clone https://github.com/openstack/kolla-kubernetes.git
git clone https://github.com/openstack/kolla-ansible.git

# Edit kolla-kubernetes/tools/Dockerfile to match development base dir

kolla-kubernetes/tools/build_dev_image.sh
kolla-kubernetes/tools/run_dev_image.sh