fiaas-deploy-daemon

---

You need Python 2.7 and pip(7.x.x or higher) on your PATH to work with fiaas-deploy-daemon.

Supported use-cases

This application is written to support three separate use-cases:

• Normal: Running inside a cluster, deploying applications in the same cluster
• Bootstrap: Running outside a cluster, deploying itself into a specified cluster
• Development: Running outside a cluster, not actually deploying anything (dry-run)

A combination of command-line arguments and environment-variables control which use-case is in effect, and how the application should act. There is no global flag to select a use-case, instead the application will look at available information at each step and determine its course of action.

See the config-module for more information.

Getting started with developing

Use $ pip install -r requirements.txt to install dependencies

• Make changes to code Run tests with tox:
• $ tox -e codestyle checks code style, indentations etc.
• $ tox -e test runs unit tests
• $ tox -e integration_test runs end-to-end/integration tests. These tests require a working minikube with the one of the hyperkit, kvm, kvm2, xhyve or virtualbox drivers installed.

Useful resources:

• http://docs.python-guide.org/
• http://pytest.org/
• http://www.voidspace.org.uk/python/mock
• http://flask.pocoo.org/

Running fiaas-deploy-daemon with minikube

To run fiaas-deploy-daemon locally and connect it to a minikube cluster, do the following.

• Set up development environment and install fiaas-deploy-daemon ($ pip install -r requirements.txt)
• Start minikube: $ minikube start
• Run $ bin/run_fdd_against_minikube

There should be a bunch of logging while fiaas-deploy-daemon starts and initializes the required ThirdPartyResources and/or CustomResourceDefinitions. This is normal.

If you need to test some behavior manually you can deploy applications into minikube via fiaas-deploy-daemon in a few ways:

Deploying an application via ThirdPartyResource

In Kubernetes 1.6 and 1.7 you can deploy applications by creating a PaasbetaApplication ThirdPartyResource.

An example PaasbetaApplication:

apiVersion: schibsted.io/v1beta
kind: PaasbetaApplication
metadata:
  labels:
    app: example
    fiaas/deployment_id: test
  name: example
  namespace: default
spec:
  application: example
  image: nginx:1.13.0
  config:
    version: 2
    host: example.com
    prometheus:
      enabled: false
    resources:
      limits:
        memory: 128M
        cpu: 200m
      requests:
        memory: 64M
        cpu: 100m

Create the resource by saving this in a file like e.g. example.yml and then run $ kubectl --context minikube create -f example.yml.

Deploying an application via CustomResourceDefinition

In Kubernetes 1.7 and later you can deploy applications by creating a Application CustomResource

apiVersion: fiaas.schibsted.io/v1
kind: Application
metadata:
  labels:
    app: example
    fiaas/deployment_id: test
  name: example
  namespace: default
spec:
  application: example
  image: nginx:1.13.0
  config:
    version: 2
    host: example.com
    prometheus:
      enabled: false
    resources:
      limits:
        memory: 128M
        cpu: 200m
      requests:
        memory: 64M
        cpu: 100m

Create the resource by saving this in a file like e.g. example.yml and then run $ kubectl --context minikube create -f example.yml.

IntelliJ runconfigs

Running fiaas-deploy-daemon

Use this configuration both for debugging and for manual bootstrapping into a cluster.

• Create a Python configuration with a suitable name: fiaas-deploy-daemon (dev)
• Script: Find the fiaas-deploy-daemon executable in the virtualenvs bin directory
  • If using bash, try which fiaas-deploy-daemon inside the virtualenv
• Script parameters:
  • --debug
  • If you want to deploy to a kubernetes cluster use --api-server, --api-token and --api-cert as suitable
• Python Interpreter: Make sure to add the virtualenv as an SDK, and use that interpreter

Tests

• Create a Python tests -> py.test configuration with a suitable name (name of test-file)
• Target: The specific test-file you wish to run
• Python Interpreter: Make sure to add the virtualenv as an SDK, and use that interpreter

fiaas-config

fiaas-deploy-daemon will read a fiaas-config to determine how to deploy your application. This configuration is a YAML file. If any field is missing, a default value will be used. The default values, and explanation of their meaning are available at /defaults on any running instance.

Release process

When changes are merged to master the master branch is built using Semaphore CI. The build generates a docker image that is published to the fiaas/fiaas-deploy-daemon repository on Docker Hub and is publicly available.

Additionally as part of the master build process jobs for updating release channels for fiaas-deploy-daemon are executed. Release channels are used by Skipper to manage FIAAS in a given cluster and it uses metadata from the release channels to determine which version of fiaas-deploy-daemon to install and when to upgrade as new versions become available.

Release channels are available via Github pages and metadata is source controlled in a Github repository.

When the master branch is built successfully the latest release channel is updated with references to the docker image, build and commit for tracability. The job for updating the latest release channel will persist the release metadata to the releases repository.

Similarly a job for updating the stable release channel is now pending but requires manual judgement and execution by a maintainer of the FIAAS organization for the release to be promoted to stable and the stable release channel to be updated.