Jesse Keating 8e8dcf9360 Don't hard set keystone endpoint

Keystone client utilizes the auth_url provided to it in order to
discover the appropriate end point for the action. It will either use
the publicURL or the adminURL it gets in the service catalog for the
identity service. UNLESS caller to the keystone client hard sets the
endpoint key, in which case keystone will blindly use it instead of
looking in the catalog. Because rally was setting it, rally also had to
set it differently when doing admin level stuff, but rally doesn't have
to do that at all. Rally can rely on the service catalog and let
keystone sort it out.

This change removes manually setting endpoint and just passes along the
auth_url to keystone. This obviates the need to define an admin_port in
a deployment configuration, and reduces the number of tests that need to
be ran.

A warning will be issued if a deployment defines an admin_port, however
the admin_port itself will be ignored. This provides some backwards
compatibility with existing deployments.

Change-Id: I917412e954e57ca5f03df0bb739b7806d38a12a5
Related-Bug: 1398375

2015-01-07 10:55:56 -08:00

4.1 KiB

Raw Blame History

Deploy engines

Introduction

One of the core entities in Rally architecture are the Deploy engines. The task of a deploy engine is to control the process of deploying some OpenStack distribution like DevStack or FUEL before any benchmarking procedures take place. Every deploy engine should implement the following fairly simple interface:

constuctor, which takes a deployment entity as its only parameter;

deploy(), which should deploy the appropriate OpenStack distribution given the cloud config from the deployment object the engine was initialized with (possibly using one of available server providers <server_providers>). The method should also return a dictionary with endpoints of the deployed OpenStack distribution;

cleanup(), which should clean up the OpenStack deployment (again, possibly using one of available server providers <server_providers>).

Below you will find a short but informative description of deploy engines implemented in Rally.

Available Deploy engines

ExistingCloud

Description

This engine in fact does not deploy anything, but uses an existing OpenStack installation. It may be useful in case you have a preconfigured OpenStack deployment ready to launch benchmark scenarios.

Configuration Example

{
    "type": "ExistingCloud",
    "auth_url": "http://192.168.122.22:5000/v2.0/",
    "endpoint_type": "public",
    "admin": {
         "username": "admin",
         "password": "password",
         "tenant_name": "admin",
    }
}

Or using keystone v3 API endpoint:

{
    "type": "ExistingCloud",
    "auth_url": "http://localhost:5000/v3/",
    "endpoint_type": "public",
    "admin": {
        "username": "engineer1",
        "user_domain_name": "qa",
        "project_name": "qa_admin_project",
        "project_domain_name": "qa",
        "password": "password",
        "region_name": "RegionOne",
    }
}

endpoint_type option will be used later for selecting access method to the cloud. Users can select from "public", "internal", "admin" access methods.

All you should specify in the config is the OpenStack cloud endpoint: the auth_url and also admin credentials, including tenant name. Rally will use the specified admin account to manage temporary non-admin tenants and users exploited while launching benchmark scenarios.

DevstackEngine

Description

This engine deploys a Devstack cloud using the given Devstack repository.

Configuration Example

{
    "type": "DevstackEngine",
    "localrc": {
        "ADMIN_PASSWORD": "secret",
        "NOVA_REPO": "git://example.com/nova/",
        ...
    },
    "devstack_repo": "git://example.com/devstack/",
    "type": {
        "name": "${PROVIDER_NAME}",
        ...
    }
}

The localrc field of the Devstack engine configuration will be used to initialize the Devstack's localrc file. As this deploy engine does not use an existing cloud, it also needs a concrete server provider <server_providers> specification: the type of the used provider (${PROVIDER_NAME}), followed by provider-specific fields configuration.

Note

More Deploy engines are to come in future releases, namely deploy engines for FUEL, Tripple-O etc. Stay tuned.

4.1 KiB Raw Blame History