Need an assist on serverspec

[jaypipes]

Hi Chris,

OK, so I'm very happy with Vagabond so far. It's been immensely useful for me to test a fix for CHEF 2100 (http://tickets.opscode.com/browse/COOK-2100). I have a master and slave recipe in my local cookbook and two nodes in a cluster assigned respectively. Here's my Vagabondfile:

{
    :nodes => {
        :metering_master => {
            :template => "ubuntu_1204",
            :ipaddress => "10.0.3.30",
            :environment => "testing",
            :run_list => [
                "recipe[infra-db::metering-master-db]"
            ]
        },
        :metering_slave => {
            :template => "ubuntu_1204",
            :ipaddress => "10.0.3.31",
            :environment => "testing",
            :run_list => [
                "recipe[infra-db::metering-slave-db]"
            ]
        }
    },
    :clusters => {
        "metering-db" => [ "metering_master", "metering_slave" ]
    },
    :local_chef_server => {
        :zero => false,
        :berkshelf => true,
        :librarian => false,
        :enabled => true,
        :auto_upload => true
    },
    :sudo => true
}

I run bundle exec vagabond cluster metering_db and the cluster nodes come up just fine. All good in the hood.

So, I decided to move on to writing some serverspec tests. I did a bundle exec vagabond spec init, edited the Layout file to look like this:

{
  :defaults => {
    :platform => 'ubuntu_1204',
    :environment => 'testing'
  },
  :definitions => {
    :metering_master => {
        :template => "ubuntu_1204",
        :ipaddress => "10.0.3.30",
        :environment => "testing",
        :run_list => [
            "recipe[infra-db::metering-master-db]"
        ]
    },
    :metering_slave => {
        :template => "ubuntu_1204",
        :ipaddress => "10.0.3.31",
        :environment => "testing",
        :run_list => [
            "recipe[infra-db::metering-slave-db]"
        ]
    }
  },
  :clusters => {
    :metering_db => {
      :nodes => [
        'metering_master',
        'metering_slave'
      ]
    }
  }
}

I then created two identical serverspec tests:

require 'serverspec_helper'

describe 'mysqld' do
  it { should be_enabled }
  it { should be_running }
end

and placed the above into files called spec/metering_master.rb and spec/metering_slave.rb.

After that, I fired off the spec, using:

bundle exec vagabond spec start metering_db

Both the master and slave nodes came up, and at the end of the run, I got:

  -> PROVISIONED
Destroying cluster: metering_db
Vagabond: Destroying node: s-metering_master-0-PDPVBPZvMD
Waiting for graceful shutdown and cleanup...
  -> DESTROYED
Vagabond: Destroying node: s-metering_slave-1-PDPVBPZvMD
Waiting for graceful shutdown and cleanup...
  -> DESTROYED
 --> Cluster metering_db DESTROYED
--> Specs for cluster metering_db: PASSED

Now, my question is... did the tests actually get run? I don't think they actually did :)

Best,
-jay

[chrisroberts]

spec/recipes/infra-db/metering-slave/mysqld.rb and spec/recipes/infra-db/metering-master/mysqld.rb

The spec stuff is based off the run list of the node, not the name. This allows for a flow of a "node that has a runlist of x should behave like y". It will map to a directory, and use all files within it, so you can have one, or a bunch to help organize them. There is some support for a custom_specs key within the configurations, but it's not fully fleshed out yet. Feel free to have a look in spec.rb and give it a try if they work :)

[jaypipes]

Chris,

I did what you suggested but didn't have much luck... I destroyed the Chef server and all nodes and then ran vagabond spec start metering-db but this time, it didn't really do anything. The Chef server got deployed and then the cluster gets immediately destroyed and the tests fail:

jpipes@uberbox:~/repos/att-cloud/cookbook-infra-db$ bundle exec vagabond spec start metering-db
Vagabond: Creating server
  -> Chef Server container created!
  -> Chef Server CREATED!
  -> Bootstrapping erchef...
Auto uploading all assets to local Chef server...
Local chef server: Uploading roles
  -> UPLOADED ROLES
Local chef server: Uploading data bags
  -> UPLOADED DATA BAGS
Local chef server: Uploading environments
  -> UPLOADED ENVIRONMENTS
Local chef server: Uploading cookbooks
Cookbooks being uploaded via berks
  -> UPLOADED COOKBOOKS
  -> All assets uploaded!
  -> CREATED!
Destroying cluster: metering-db
 --> Cluster metering-db DESTROYED
--> Specs for cluster metering-db: FAILED

Not sure what happened, but all I did was mv the two files as you suggested and re-start the cluster. Passing --debug to the command did not produce any more output:

jpipes@uberbox:~/repos/att-cloud/cookbook-infra-db$ bundle exec vagabond spec start metering-db --debug
[sudo] password for jpipes: 
ERROR: Node already exists and is running: server
Vagabond: Starting node: server
  -> STARTED
Destroying cluster: metering-db
 --> Cluster metering-db DESTROYED
--> Specs for cluster metering-db: FAILED

Any ideas?

Best,
-jay

[jaypipes]

Doh, ignore the above :( I was using metering-db and not metering_db. Probably would be good to have a better error message in case of mistyped cluster names! :) I'll work on that and push a patch.

[jaypipes]

OK, well the cluster completed again, and I got all tests passed:

10.0.3.31 [2013-07-16T12:30:56+00:00] WARN: Changing master to 10.0.3.30 at log position 190 in log file mysql-bin.000002
10.0.3.31 [2013-07-16T12:30:56+00:00] WARN: Starting slave ubuntu1204-ynDjVVlnvaYn
10.0.3.31 
10.0.3.31     - execute the ruby block start-slave
10.0.3.31 
10.0.3.31 
10.0.3.31 Chef Client finished, 19 resources updated
  -> PROVISIONED
Destroying cluster: metering_db
Vagabond: Destroying node: s-metering_master-0-PDPVBPZvMD
Waiting for graceful shutdown and cleanup...
  -> DESTROYED
Vagabond: Destroying node: s-metering_slave-1-PDPVBPZvMD
Waiting for graceful shutdown and cleanup...
  -> DESTROYED
 --> Cluster metering_db DESTROYED
--> Specs for cluster metering_db: PASSED

It's still not clear however that the tests ran.. any way I can tell ?