promenade/docs/source/getting-started.rst

4.3 KiB

Getting Started

Running Tests

Initial Setup of Virsh Environment

To setup a local functional testing environment on your Ubuntu 16.04 machine, run:

./tools/setup_gate.sh

Running Functional Tests

To run complete functional tests locally:

./tools/gate.sh

For more verbose output, try:

PROMENADE_DEBUG=1 ./tools/gate.sh

The gate leaves its test VMs running for convenience. To shut everything down:

./tools/stop_gate.sh

To run a particular set of functional tests, you can specify the set on the command line:

./tools/gate.sh <SUITE>

Valid functional test suites are defined by JSON files that live in tools/g2/manifests.

Utilities

There are a couple of helper utilities available for interacting with gate VMs. These can be found in tools/g2/bin. The most important is certainly ssh.sh, which allows you to connect easily to test VMs:

./tools/g2/bin/ssh.sh n0

Development

Using a Local Registry

Repeatedly downloading multiple copies images during development can be quite slow. To avoid this issue, you can run a docker registry on the development host:

./tools/registry/start.sh
./tools/registry/update_cache.sh

Then, the images used by the example can be updated using:

./tools/registry/update_example.sh

That change can be undone via:

./tools/registry/revert_example.sh

The registry can be stopped with:

./tools/registry/stop.sh

Deployment using Vagrant

Initial Setup of Vagrant

Deployment using Vagrant uses KVM instead of Virtualbox due to better performance of disk and networking, which both have significant impact on the stability of the etcd clusters.

Make sure you have [Vagrant](https://vagrantup.com) installed, then run ./tools/vagrant/full-vagrant-setup.sh, which will do the following:

  • Install Vagrant libvirt plugin and its dependencies
  • Install NFS dependencies for Vagrant volume sharing
  • Install [packer](https://packer.io) and build a KVM image for Ubuntu 16.04

Deployment

A complete set of configuration that works with the Vagrantfile in the top-level directory is provided in the example directory.

To exercise that example, first generate certs and combine the configuration into usable parts:

./tools/build-example.sh

Start the VMs:

vagrant up --parallel

Then bring up the genesis node:

vagrant ssh n0 -c 'sudo /vagrant/example/scripts/genesis.sh'

Join additional master nodes:

vagrant ssh n1 -c 'sudo /vagrant/example/scripts/join-n1.sh'
vagrant ssh n2 -c 'sudo /vagrant/example/scripts/join-n2.sh'

Re-provision the genesis node as a normal master:

vagrant ssh n0 -c 'sudo promenade-teardown'
vagrant ssh n1 -c 'sudo kubectl delete node n0'
vagrant destroy -f n0
vagrant up n0
vagrant ssh n0 -c 'sudo /vagrant/example/scripts/join-n0.sh'

Join the remaining worker:

vagrant ssh n3 -c 'sudo /vagrant/example/scripts/join-n3.sh'

Building the image

To build the image directly, you can use the standard Docker build command:

docker build -t promenade:local .

To build the image from behind a proxy, you can:

export http_proxy=...
export no_proxy=...
docker build --build-arg http_proxy=$http_proxy --build-arg https_proxy=$http_proxy --build-arg no_proxy=$no_proxy  -t promenade:local .

For convenience, there is a script which builds an image from the current code, then uses it to construct scripts for the example:

./tools/dev-build.sh

NOTE the dev-build.sh script puts Promenade in debug mode, which will instruct it to use Vagrant's shared directory to source local charts.

Using Promenade Behind a Proxy

To use Promenade from behind a proxy, use the proxy settings see configuration/kubernetes-network.