Jump to: navigation, search

Cue/api

< Cue
Revision as of 21:04, 25 November 2014 by Dagnello (talk | contribs) (Cue API Design)

Cue API Design

Acronyms

Acronym Definition
SSL Secure Sockets Layer
REST Representational State Transfer
URI Uniform Resource Identifier
UUID Universally Unique Identifier
AMQP Advanced Messaging Queuing Protocol

Introduction

Messaging systems enable software applications to communicate and scale. Many of today's enterprise systems make use of messaging systems as the glue between the different components or applications. These components depend on messaging to interact with each other to provide end-to-end functionality. This Openstack stackforge project is intended to solve the problem in the provisioning, deployment and management of messaging systems in a seamless automated fashion for applications deployed on the Openstack cloud operating system.

There are several off-the-shelf messaging systems that implement messaging and queuing semantics such as RabbitMQ, Qpid and ActiveMQ. These proven systems have wide acceptance, usage and testing. The goal of this project is not to implement another messaging system for the Openstack platform, but to implement an automated messaging provisioning, deployment and management system. This system will simplify the application development life-cycle to allow the developer to focus on their application instead of the back-end storage services.

Goals

The focus of the project is to provide high-availability, auto-healing and tenant level isolation to the end user. The initial goal for version 1.0 is to use RabbitMQ as the supported messaging system. The overall goal for this effort is to provide a messaging queueing provisioning service for OpenStack built with community support.

Requirements

The requirements for the MSGaaS Queue v1.0 project are listed in priority order below.

   * Community Building (Rackspace involvement)
   * Control plane
   * Rabbit Management Console
   * Keystone integration
   * Command-Line – Create/Delete Cluster
   * Horizon Integration - Create/Delete Cluster
   * Development Platform Installer Integration
   * Queue Control Plane Patching
   * RabbitMQ Instance Patching
   * Cluster Management – Scale up/down
   * Devstack integration
   * Gate Tests
   * Queue Control Plane Upgrade
   * RabbitMQ Instance Upgrade

System Context Diagram

SCD.jpg

Component Description
User Direct customer of MSGaaS.
Horizon MSGaaS functionality will be added to Horizon, which will provide a web-based portal for MSGaaS control.
CLI Command Line Interface to MSGaaS, provides user access to provisioning and deploying messaging clusters.
REST_API Provides user access to provisioning and deploying messaging clusters through REST interface. This is a light-weight interface, provisioning and configuration of clusters/nodes is delegated to the TaskWorker process.
TaskScheduler Used to synchronize work tasks between the REST_API and TaskWorker processes.
TaskWorker Carries out work associated with all provisioning, configuration and management of RabbitMQ clusters and nodes. Makes use of heat for initial provisioning/deployment.
DB Database to store information on clusters and nodes. Example, in cluster creation, when request is initially received through REST_API, this DB is updated accordingly and the work is delegated to TaskWorker. TaskWorker then updates this DB as provisioning and configuration takes place. Subsequent calls to check on status of cluster creation, will return updated information from this DB.
Heat Used cloud instance orchestration in deploying RabbitMQ cluster images.

REST API

General requirement, REST API must respond within 500ms.

List Clusters

GET /v1/clusters

This operation synchronously returns all clusters provisioned within the associated project id.

Request Parameters N/A

Response Code(s)
Normal ok (200)
Error unauthorized (401)

Response Parameters

Parameter Type Description
cluster(list) string List of clusters, detailing respective cluster id, name, status, created time stamp and updated time stamp.


JSON Request

N/A

JSON Response

{

   "clusters": [
       {
           "cluster_id": "b51948c9-1ac5-4c28-a580-6f7c500d82f8",
           "name": "Message Cluster 1",
           "status": "ACTIVE",
           "created": "2014-11-11T01:02:03Z",
           "updated": "2014-11-11T01:02:03Z"
       },
       {
           "cluster_id": "13c456c9-bbfc-4c31-b26d-3ae5c3cd7a77",
           "name": "Message Cluster 2",
           "status": "ACTIVE",
           "created": "2014-11-12T13:23:54Z",
           "updated": "2014-11-13T19:55:01Z"
       }
   ]

}

Create Cluster

POST /v1/clusters

This operation asynchronously creates a new cluster of Nova instances provisioned with the required message brokers in a central tenant.

Request Parameters

Parameter Type Description
nic UUID Network Identification for a Neutron network where cluster will be created in.
name string Name of cluster.
volume_size int Optional parameter to indicate size of volume for node instance. If volumes are supported, then this parameter will be used. If ephmeral disk are not supported, volume support will be required.
flavor (list) string List of node flavors, which specify VM type in terms of CPU/memory/disk resources (e.g. small, medium, large). Size of list denotes the number of nodes in cluster.

small: 1 GHz dual-core CPU; 512 MB memory; 250 GB disk

medium: 2.8 GHz dual-core CPU; 4 GB memory; 1 TB disk

large: 3.6 GHz quad-core CPU; 32 GB memory; 5 TB disk

Response Codes

Response Code(s)
Normal 202 (accepted)
Error badRequest (400), unauthorized (401), itemNotFound (404)

Response Parameters

Parameter Type Description
cluster_id UUID ID of cluster to be created.
name string Name of cluster (same as provided name in request parameters).
status string Current status of cluster.

BUILDING: Cluster is in progress of being provisioned.

ACTIVE: Cluster is running.

ERROR: Provisioning error(s) encountered.

DELETED: Cluster has been deleted.

JSON Request

{

   "nic": "d32019d3-bc6e-4319-9c1d-6722fc136a22",
   "name": "Message Cluster 1",
   "volume_size": 100
   "nodes": [
       {
           "flavor": "large"
       },
       {
           "flavor": "large"
       },
       {
           "flavor": "large"
       }
   ]

}

JSON Response

{

   "cluster_id": "dd745f4a-9333-417e-bb89-9c989c84c068",
   "name": "Message Cluster 1",
   "status": "BUILDING",
   "created": "2014-11-11T01:02:03Z",
   "updated": "2014-11-11T01:02:03Z",
   "nodes": [
       {
           "node_id": "616fb98f-46ca-475e-917e-2563e5a8cd19",
           "status": "BUILDING",
           "flavor": "large",
       },
       {
           "node_id": "e90c9d13-c4b8-4a08-992a-dad6109b8ac2",
           "status": "BUILDING",
           "flavor": "large",
       },
       {
           "node_id": "372f8f47-6818-4d83-aa42-8744c0e689b8",
           "status": "BUILDING",            
           "flavor": "large",
       }

}

Show Cluster

GET /v1/clusters/{cluster_id}

This operation synchronously returns the status and information on the specified cluster within the provided tenant.

Request Parameters

Parameter Type Description
cluster_id (URI) UUID Cluster ID. This value is returned when a new cluster is created.

Response Codes

Response Code(s)
Normal ok (200)
Error unauthorized (401), itemNotFound (404)

Response Parameters

Parameter Type Description
cluster_id UUID ID of requested cluster.
name string Name of cluster.
status string Current status of cluster.

BUILDING: Cluster is in progress of being provisioned.

ACTIVE: Cluster is running.

ERROR: Provisioning error(s) encountered.

DELETED: Cluster has been deleted.

created string Created time stamp in format: yyyy-mm-ddThh:mm:ssZ
updated string Last updated time stamp in format: yyyy-mm-ddThh:mm:ssZ
nodes(list) node List of nodes, which includes node id, instance state, and flavor.

JSON Request

N/A

JSON Response

{

   "cluster_id": "dd745f4a-9333-417e-bb89-9c989c84c068",
   "name": "Message Cluster 1",
   "status": "ACTIVE",
   "created": "2014-11-11T01:02:03Z",
   "updated": "2014-11-11T01:02:03Z",
   "nodes": [
       {
           "node_id": "616fb98f-46ca-475e-917e-2563e5a8cd19",
           "status": "ACTIVE",
           "flavor": "large",
           "endpoints": [
               {
                   "type": "AMQP",
                   "value": "amqp://10.20.30.40:10000"
               },
               {
                   "type": "console",
                   "value": "http://10.20.30.40:5672"
               }
           ],
           "created": "2014-11-11T01:06:03Z",
           "updated": "2014-11-11T01:06:03Z"
       },
       {
           "node_id": "e90c9d13-c4b8-4a08-992a-dad6109b8ac2",
           "status": "ACTIVE",
           "flavor": "large",
           "endpoints": [
               {
                   "type": "AMQP",
                   "value": "amqp://10.20.30.41:10000"
               },
               {
                   "type": "console",
                   "value": "http://10.20.30.41:5672"
               }
           ],
           "created": "2014-11-11T01:08:03Z",
           "updated": "2014-11-11T01:08:03Z"
       },
       {
           "node_id": "372f8f47-6818-4d83-aa42-8744c0e689b8",
           "status": "ACTIVE",
           "flavor": "large",
           "endpoints": [
               {
                   "type": "AMQP",
                   "value": "amqp://10.20.30.42:10000"
               },
               {
                   "type": "console",
                   "value": "http://10.20.30.42:5672"
               }
           ],
           "created": "2014-11-11T01:12:03Z",
           "updated": "2014-11-11T01:12:03Z"
       }
   ]

}

Delete Cluster

DELETE /v1/clusters/{cluster_id}''

This operation will asynchronously delete the indicated cluster within the provided tenant.

Request Parameters

Parameter Type Description
cluster_id (URI) UUID Cluster ID. This value is returned when a new cluster is created.

Response Codes

Response Code(s)
Normal accepted (202)
Error unauthorized (401), itemNotFound (404)

Response Parameters

N/A

JSON Request

N/A

JSON Response

N/A

Testing

Cue API testing will verify the expected functionality of the Cue user interface with both positive/negative tests. The overall scope will cover testing from the HTTP REST request to required database interactions and work-flow task submission for RPC workers.

Unit Tests

Unit tests will verify the resulting function calls for each REST-ful URI and action(s). The Python Mock library will be used to replace external system dependencies with placeholder objects.

Function Tests Input Data Expected Result(s)
List Clusters
  1. Create 'n' cluster(s), then call list clusters and verify expected return object
  2. Call list clusters when no clusters exist within tenant and verify expected return object
  • n = # of clusters to create
  1. List of 'n' clusters is returned with appropriate fields, HTTP Ok (200)
  2. Empty list of clusters is returned, HTTP Ok (200)
Create Cluster
  1. Create standard cluster 'c' and verify expected return object
  2. Create cluster with invalid flavor and verify expected return value
  3. Create cluster with invalid volume size and verify expected return value
  • n = cluster size (number of nodes)
  • m = cluster name
  • f = flavor
  • s = volume size
  • nic = Network UUID
  1. Cluster ID, name and status for newly created cluster is returned, HTTP Accepted (202)
  2. HTTP Bad request is returned (400)
  3. HTTP Bad request is returned (400)
Show Cluster
  1. Create cluster, then call get cluster and verify expected return object
  2. Get cluster with invalid cluster_id
  • cluster_id = cluster ID
  1. Cluster ID, name, status, created/updated date stamp and list of nodes is returned, HTTP Ok (200).
  2. HTTP Bad request is returned (400)
Delete Cluster
  1. Create a cluster, then call delete cluster and verify expected return value
  2. Delete cluster with invalid cluster_id
  • cluster_id = cluster ID
  1. HTTP accepted is returned (202)
  2. HTTP Bad request is returned (400)

Functional Tests

The functional tests will verify the HTTP REST URI request lifecycle from controller routing to HTTP response. These tests will make use of Pecan's testing utility; pecan.testing module.

Operation URI Tests Input Data Expected Data (JSON)
GET /v1/clusters
  1. Call when clusters exist
  2. Call when no clusters exist
  • n/a
  1. List of 'n' clusters is returned with appropriate fields, HTTP Ok (200)
  2. Empty list of clusters is returned, HTTP Ok (200)
POST /v1/clusters
  1. Create standard cluster 'c' and verify expected return object
  2. Create cluster with invalid flavor and verify expected return value
  3. Create cluster with invalid volume size and verify expected return value
  • n = cluster size (number of nodes)
  • m = cluster name
  • f = flavor
  • s = volume size
  • nic = Network UUID
  1. Cluster ID, name and status for newly created cluster is returned, HTTP Accepted (202)
  2. HTTP Bad request is returned (400)
  3. HTTP Bad request is returned (400)
GET /v1/clusters/{cluster_id}
  1. Create cluster, then call get cluster and verify expected return object
  2. Get cluster with invalid cluster_id
  • cluster_id = cluster
  1. Cluster ID, name, status, created/updated date stamp and list of nodes is returned, HTTP Ok (200).
  2. HTTP Bad request is returned (400)
DELETE /v1/clusters/{cluster_id}
  1. Create a cluster, then call delete cluster and verify expected return value
  2. Delete cluster with invalid cluster_id
  • cluster_id = cluster ID
  1. HTTP accepted is returned (202)
  2. HTTP Bad request is returned (400)

Integration Tests

The integration tests will cover API functionality from HTTP request to database access and task submission for RPC workers. The Pecan test utility will be used to route test request URI to appropriate controllers, then database record(s) will be verified for applicable change. Finally the creation of task objections will be verified to ensure valid task flows.