Skip to content

Latest commit

 

History

History
145 lines (108 loc) · 5.14 KB

spec.md

File metadata and controls

145 lines (108 loc) · 5.14 KB
Raw

Specification

Requirements

Coffee is a beverage that CloudStack needs to support, a typical coffee life cycle involves it to be created, updated, listed, removed etc. Coffee can be brewed by hypervisors however a framework should allow developers to write custom coffee/brewing plugins. The coffee should have a maximum time to live, a garbage-collection thread to remove stale coffees. The feature should be available through UI.

Functional Spec

Your task now is to read and understand the above requirements and then proceed with the implementation based on a functional specification.

References:

FS references and examples:

The feature can be implemented as a CloudStack plugin at plugins/hackerbook/feature which should also implement a framework and two coffee machine plugins:

  • default: the in-built cloudstack brewer machine, implemented at plugins/hackerbook/default
  • kvm: the brewer machine that brews coffee on a KVM host, implemented at plugins/hackerbook/kvm

A Coffee should follow a finite-state-machine (FSM) with following states:

  • Created: Coffee order created
  • Brewing: Coffee is order is sent
  • Brewed: Coffee is brewed
  • Discarded: Coffee is put in trash

The size of coffee can be as follows:

  • SMALL
  • MEDIUM
  • LARGE

The following coffee offering should be supported:

  • Espresso
  • Cappuccino
  • Mocha
  • Latte

The following sections document high-level implementation details.

APIs

API Arguments Description
createCoffee name, offering, size, details Creates coffee
listCoffees id, offering, size Lists coffee by id, type and/or size
updateCoffee id, size, details Updates coffee order that is not ready yet
removeCoffee id, ids Removes coffee by id or ids

The details are string->string map of any arbitrary key/value to note any details of each coffee order.

DB Schema

Following tables should be added with appropriate DAO and VO classes:

  • coffee:

    • id: auto_increment ID for the table
    • uuid: the UUID for a coffee order
    • name: name on the coffee order
    • offering: the offering/type of the coffee
    • size: size of coffee
    • state: the state of Coffee
    • account_id: the user account ID
    • domain_id: the domain ID
    • created: the creation timestamp
    • removed: the removal timestamp

  • coffee_details:

    • id: auto_increment ID for the table
    • coffee_id: the coffee ID
    • name: the string key
    • value: the string value
    • display: whether to show/hide the detail in the API response

Service Changes

The feature would require a new service/business layer class that handles various API requests along with exports APIs, global settings and schedules background task(s). This may also implement the framework and support discovery and usage of plugins.

Global settings:

  • coffee.ttl.interval: The max time in seconds after which coffee becomes stale, default 600 seconds.
  • coffee.gc.interval: The interval in seconds for the GC thread to run, default 10s.
  • coffee.machine.plugin: The name of coffee machine plugin, default is inbuilt.

Background task: Run based on GC interval and scan coffees across all accounts and remove coffee that have exceeded the coffee TTL interval.

Pluggable CoffeeMaker

Implement a framework that allows plugin creation based on a CoffeeMaker interface that implements a brew() method. Implement two plugins, an inbuilt default coffeemaker that performs no-operation (maybe just log) and a separate maven project based lazy coffeemaker that randomly sleeps in brew().

Message Passing: IPC and RPC

On creation of any new CloudStack account, a fresh default coffee must be created and on removal of an account all coffees of that account must be purged. On coffee creation and removal, publish event on the message bus. On coffee discarded by GC thread, send an alert.

Create a CoffeeMachine plugin that brews coffee via a random agent such as the CPVM/SSVM agent or the KVM agent.

Coffee Usage Records

Introduce a new coffee usage record type and implement usage parser for the coffee resource that generates usage records aggregated per account by the number of coffees ordered.

UI

The feature should be implemented as a UI plugin which shows up as a new coffee tab in the UI. In the tab, users should see a table of coffee orders with name, offering, size, state and account. On clicking a coffee, it should show a details tab that lists the attributes of the coffee. The UI should provide appropriate buttons to create, update and remove coffee.

Testing

Create Marvin test for the API and features.

Packaging

Make the coffee feature and any of its plugins packaged separately as a separate rpm/deb package.