openstack-seeder

README

Openstack Seeder

Seed your openstack content with a kubernetes operator.

• introduces a new kubernetes CustomResourceDefinition openstackseed
  • kubectl get openstackseeds lists all deployed openstack seeds
  • kubectl apply/delete can be used to maintain openstack seed specs
• the k8s openstack-seeder watches the lifecycle events of these specs
• openstack seed specs can depend on another (hierarchy)
• on a lifecycle event of a seed spec (only create/update), the operator resolves the dependencies of the specs (merges them) and invokes the seeding of the merged seed spec

Seeding currently only supports creating or updating of entities (upserts).

Supported entities

• regions

• roles

• role_inferences

• services

  • endpoints

• flavors

  • extra-specs

• share_types

  • is_public
  • specs
    • driver_handles_share_servers
    • snapshot_support
  • extra-specs

• domains

  • configuration
  • domain-role-assignments
  • projects
    • project-role-assignments
    • project-endpoints
    • network quotas
    • address scopes
      • subnet pools
    • subnet pools
    • networks
      • tags
      • subnets
    • routers
      • interfaces
    • swift
      • account
      • containers
    • dns_quota
    • dns_zones
      • recordsets
    • dns_tsigkeys
    • ec2_creds
    • flavors
    • share_types
  • groups
    • group-role-assignments
  • users
    • user-role-assignments
  • roles

Spec format

The seeding content can be provided in the usual kubernets spec yaml format.

The exact specfification of the seed format can be found in the pkg/seeder/apis/v1/types go doc.

Example seed spec of a keystone seed to be deployed via helm:

apiVersion: "openstack.stable.sap.cc/v1"
kind: "OpenstackSeed"
metadata:
  name: keystone-seed
  labels:
    app: {{ template "fullname" . }}
    chart: "{{ .Chart.Name }}-{{ .Chart.Version }}"
    release: "{{ .Release.Name }}"
    heritage: "{{ .Release.Service }}"
spec:
  roles:
  - name: admin
    description: 'Keystone Administration'
  - name: member
    description: 'Keystone Member'
  - name: reader
    description: 'Keystone Read-Only'
  - name: service
    description: 'Keystone Service'

  role_inferences:
  - prior_role: admin
    implied_role: member
    
  regions:
  - id: eu
    description: 'Europe'
  - id: staging
    description: 'Staging'
    parent_region_id: eu
  - id: qa
    description: 'QA'
    parent_region_id: eu
  - id: local
    description: 'Local Development'

  services:
  - name: keystone
    type: identity
    description: Openstack Identity
    endpoints:
    - region: local
      interface: public
      url: {{ .Value.keystoneUrl }}:5000/v3
      enabled: true
    - region: local
      interface: admin
      url: {{ .Value.keystoneUrl }}:35357/v3
      enabled: true
    - region: local
      interface: internal
      url: http://keystone.{{.Release.Namespace}}.svc.kubernetes.{{.Values.region}}.{{.Values.tld}}:5000/v3'
      enabled: false

  domains:
  - name: Default
    id: default
    description: Openstack Internal Domain
    enabled: true
    users:
    - name: admin
      description: Openstack Cloud Administrator
      enabled: true
      password: secret123
      role_assignments:
      - domain: Default
        role: admin
      - project: admin
        role: admin
      - project: service
        role: admin

    groups:
    - name: administrators
      description: Administrators
      role_assignments:
      - domain: Default
        role: admin
      - project: admin
        role: admin
      - project: service
        role: admin
      users:
      - admin
    - name: members
      description: Members
      role_assignments:
      - domain: Default
        role: member
    projects:
    - name: admin
      description: Administrator Project
    - name: service
      description: Services Project    

why did you not use gophercloud as a go openstack client?

When we started the implementation of the operator, the gophercloud api coverage was far from complete. Hence we decided to spawn a python seeder that does the actual seeding via the python opensatck api clients.