Satori

OpenStack Resource State Discovery

The mission of the OpenStack Resource State Discovery project is to discover the state and current configuration of different resource types in an openstack system. For example, given a URL and some credentials, discover which server(s) the URL is hosted on, the topology of the infrastructure and what software is running on each server, as well as, for example, how close it is to reaching its max Apache clients, or the frequency of PHP errors in the recent past.

Use Cases

We expect that the output of a configuration discovery tool could be used for:

• Troubleshooting unfamiliar environments
• Configuration analysis (compare the discovered state vs a library of best practices)
• System State Verification
• Detecting configuration changes over time
• Heat Template generation
• Solum Application creation/import
• Creation of Chef recipes/cookbooks, Puppet modules, Ansible playbooks, setup scripts, etc.

Getting Started

The project will begin with a simple command-line tool so that new users can discover a configuration without needing to setup additional infrastructure. A mockup for the output is listed below:

$ waldo discover www.foo.com
Domain: foo.com
   Registered at TUCOWS DOMAINS INC.
   Expires in 475 days.
   Name servers:
       DNS1.STABLETRANSIT.COM
       DNS2.STABLETRANSIT.COM    
Address:
   www.foo.com resolves to IPv4 address 4.4.4.4
Host:
   4.4.4.4 (www.foo.com) is hosted on a Nova Instance
   Instance Information:
       URI: https://nova.api.somecloud.com/v2/111222/servers/d9119040-f767-414
            1-95a4-d4dbf452363a
       Name: sampleserver01.foo.com
       ID: d9119040-f767-4141-95a4-d4dbf452363a
   ip-addresses:
       public:
           ::ffff:404:404
           4.4.4.4
       private:
           10.1.1.156
   System Information:
       Ubuntu 12.04 installed
       Server was rebooted 11 days, 22 hours ago
       /dev/xvda1 is using 9% of its inodes.
   Running Services:
       httpd on 127.0.0.1:8080
       varnishd on 0.0.0.0:80
       sshd on 0.0.0.0:22
   httpd:
       Using 7 of 100 MaxClients

Initial Design Ideas

Though not yet proposed in Blueprints, these are some of the initial thoughts for the project to help guide discussion:

• A pluggable implementation to allow discovering:
  • OpenStack infrastructure (instances, networks, et al.)
  • Non-OpenStack Infrastructure (APIs for other clouds, nodes in a Chef server, vendor management interfaces)
  • External infrastructure: DNS ownership and configuration, SSL Certificate validity, etc.
  • Operating system and application configuration (Chef Ohai, Puppet Facter, Devstructure Blueprint, etc)
  • Runtime information such as active processes and connections between systems.
• An API service for tenants to make discovery requests.
• A backend datastore to record the results of discoveries.