agnostics

== Agnostic Scheduler

image:https://github.com/redhat-gpe/agnostics/workflows/Go/badge.svg[Go] image:https://quay.io/repository/redhat-gpte/scheduler/status["Docker Repository on Quay", link="https://quay.io/repository/redhat-gpte/scheduler"]

image::https://gpte-public.s3.amazonaws.com/agnostics/scheduler.png[width=100%]

=== link:https://redhat-gpe.github.io/agnostics/api-reference[API Reference]

=== Usage (server)

[source,subs="+quotes,verbatim,macros"]
----
Usage of ./scheduler:
  -api-addr string
        The address API listens to.
        Environment variable: *API_ADDR*
         (default ":8080")
  -api-auth
        Enable authentication for the API.
        Environment variable: *API_AUTH*  ('true' or 'false')
         (default true)
  -api-htpasswd string
        The path of the htpasswd file to use for authentication for the API.
        Environment variable: *API_HTPASSWD*
         (default "api-htpasswd")
  -console-addr string
        The address the Console listens to.
        Environment variable: *CONSOLE_ADDR*
         (default ":8081")
  -debug
        Debug mode.
        Environment variable: *DEBUG*

  -git-ssh-private-key string
        The path of the SSH private key used to authenticate to the git repository. Used only when 'git-url' is an SSH URL.
        Environment variable: *GIT_SSH_PRIVATE_KEY*

  -git-url string
        The URL of the git repository where the scheduler will find its configuration. SSH is assumed, unless the URL starts with 'http'.
        Environment variable: *GIT_URL*
         (default "\git@github.com:redhat-gpe/scheduler-config.git")
  -redis-url string
        The URL to access redis. The format is described by the IANA specification for the scheme, see https://www.iana.org/assignments/uri-schemes/prov/redis
        Environment variable: *REDIS_URL*
         (default "redis://localhost:6379")
  -template-dir string
        The directory containing the golang templates for the Console.
        Environment variable: *TEMPLATE_DIR*
         (default "templates")
----

== Config Git repository ==

The Git repository must contain the following:

- `/policy.yaml` - describing the policy.
- `/clouds` - directory containting the definition of the resources (clouds) to be scheduled.

.example `policy.yaml`
[source,yaml]
----
---
predicates:
  - name: LabelPredicates
  - name: TaintPredicates

priorities:
  - name: LabelPriorities
    weight: 1 # <1>
  - name: TaintPriorities
    weight: 1 # <2>
----
<1> The weight that will be used to increase the priority of Clouds that match the preferences provided in the Schedule request
<2> The weight that will be used to decrease the priority of Clouds if they have Taints of type "PreferNoSchedule"

.example `clouds/openstack-blue.yml`
[source,yaml]
----
---
name: openstack-blue
labels:
  type: osp
  region: na
  datacenter: wdc
  purpose: ilt
----

== Example using the scheduler (client)

Here is an example how the scheduler can be used from ansible.

.Input variables
[source,yaml]
----
agnosticv_meta:
  scheduler:
    enable: true
    data:
      cloud_preference:
        purpose: development
        region: '{{ region }}'
      uuid: '{{ uuid }}'
    endpoint: /api/v1/schedule
    url: https://scheduler.example.com

----

.Playbook to schedule or retrieve a placement using UUID
[source,yaml]
----
- name: Schedule or retrieve a placement using UUID
  hosts: localhost
  gather_facts: false
  vars:
    output: /tmp/placement.json
  tasks:
    - name: Schedule a placement
      uri:
        url: "{{ agnosticv_meta.scheduler.url + agnosticv_meta.scheduler.endpoint }}"
        validate_certs: "{{ agnosticv_meta.scheduler.validate_certs | default(false) }}"
        return_content: true
        method: POST
        body_format: json
        body: "{{ agnosticv_meta.scheduler.data }}"
        status_code: [200, 400]
        dest: "{{ output }}"
      register: r_placement
      retries: 10
      delay: 30
      until: r_placement is succeeded

    - when: >-
        r_placement.status == 400
        and 'service uuid already has a placement' in r_placement.json.message
      name: Get placement using uuid
      uri:
        url: "{{ agnosticv_meta.scheduler.url }}/api/v1/placements/{{ uuid }}"
        validate_certs: "{{ agnosticv_meta.scheduler.validate_certs | default(false) }}"
        return_content: true
        method: GET
        dest: "{{ output }}"
      register: g_placement
      retries: 10
      delay: 30
      until: g_placement is succeeded

    - debug:
        msg: >-
          {% if r_placement.json.cloud is defined %}
          {{ r_placement.json.cloud.name }}
          {% else %}
          {{ r_placement.json.message }}: {{ g_placement.json.cloud.name }}
          {% endif %}
----

. Playbook to delete placement using UUID
[source,yaml]
----
- name: Delete placement using UUID
  hosts: localhost
  gather_facts: false
  tasks:
    - name: Delete placement using uuid
      uri:
        url: "{{ agnosticv_meta.scheduler.url }}/api/v1/placements/{{ uuid }}"
        validate_certs: "{{ agnosticv_meta.scheduler.validate_certs | default(false) }}"
        return_content: true
        method: DELETE
      register: r_placement
      retries: 10
      delay: 30
      until: r_placement is succeeded
----

== License

The scripts and documentation in this project are released under the
link:LICENSE[MIT License]

== Contributions

Contributions are welcome! See link:docs/contributors.md[Contributor’s
Guide]

=== Code of Conduct

Be nice :wave:

link:http://docs.ansible.com/ansible/community.html#community-code-of-conduct[Ansible Code of Conduct] applies here.


=== See Also

- https://github.com/redhat-cop/agnosticd
- https://github.com/redhat-cop/agnosticv
- https://github.com/redhat-cop/babylon