When we use the Automate capability of CloudForms, we write scripts in the Ruby language and use objects that the CloudForms Automation Engine makes available to us. The CloudForms Web User Interface (WebUI) allows us to access the Automate functionality via the Automate top-level menu (see Figure 2-1).
The first menu item that we see takes us to the Explorer. This is our visual interface into the Automate Datastore, and it contains the various kinds of Automate objects that we’ll use throughout this book (see Figure 2-2).
Before we start our journey into learning CloudForms Automate, we’ll take a tour of the Automate Datastore to familiarize ourselves with the objects that we’ll find there.
The Automate Datastore has a directory-like structure, consisting of several types of organizational units arranged in a hierarchy (see Figure 2-3).
Next, we’ll look at each type of object in more detail.
A domain is a collection of namespaces, classes, instances, and methods. The ManageIQ project provides a single ManageIQ domain for all supplied automation code, while Red Hat adds the supplemental RedHat domain containing added-value code for the CloudForms product. Both the ManageIQ and RedHat domains are locked, indicating their read-only nature, but we can create new domains for our own custom automation code. Figure 2-2 shows the two default domains and two custom domains: Bit63 and ACME.
Organizing our own code into custom domains greatly simplifies the task of exporting and importing code (simplifying code portability and reuse). It also leaves ManageIQ or Red Hat free to update the locked domains through minor releases without fear of overwriting our customizations.
User-added domains can be individually enabled or disabled and can be ordered by priority such that if code exists in the same path in multiple domains (for example, /Cloud/VM/Provisioning/StateMachines), the code in the highest-priority enabled domain will be executed. We can change the priority order of our user-added domains using the Configuration → Edit Priority Order of Domains menu (see Figure 2-4).
We can export domains using rake
from the command line and import them either using rake
or from the WebUI. (Using rake
enables us to specify more import and export options.) A typical rake
import line is as follows:
bin/rake evm:automate:import YAML_FILE=bit63.yaml IMPORT_AS=Bit63 SYSTEM=false ENABLED=true DOMAIN=Export PREVIEW=false
We frequently need to customize code in the locked RedHat or ManageIQ domains—for example, when implementing our own custom VM placement method. Fortunately, we can easily copy any object from the locked domains into our own, using Configuration → Copy this (see Figure 2-5).
When we copy an object such as a class, we are prompted for the From and To domains. We can optionally deselect “Copy to same path” and specify our own destination path for the object (see Figure 2-6).
Domains were a new feature of the Automate Datastore in CloudForms 3.1. Prior to this release, all factory-supplied and user-created automation code was contained in a common structure, which made updates difficult when any user-added code was introduced (the user-supplied modifications needed exporting and reimporting/merging whenever an automation update was released).
To import a Datastore backup from a CloudForms 3.0 and prior format Datastore, we must convert it to the new Datastore format first, like so:
cd /var/www/miq/vmdb bin/rake evm:automate:convert FILE=database.xml DOMAIN=SAMPLE ZIP_FILE=/tmp/sample_converted.zip
A namespace is a folder-like container for classes, instances, and methods, and is used purely for organizational purposes. We create namespaces to arrange our code logically, and namespaces often contain other namespaces (see Figure 2-7).
A class is similar to a template: it contains a generic definition for a set of automation operations. Each class has a schema that defines the variables, states, relationships, or methods that instances of the class will use.
The Automate Datastore uses object-oriented terminology for these objects. A class is a generic definition for a set of automation operations, and these classes are instantiated as specific instances. The classes that we work with in the Automate Datastore are not the same as Ruby classes that we work with in our automation scripts.
A schema is made up of a number of elements, or fields, that describe the properties of the class. A schema often has just one entry—to run a single method—but in many cases it has several components. Figure 2-8 shows the schema for a placement class, which has several different field types.
We add or edit each schema field in the schema editor by specifying the field type from a drop-down list (see Figure 2-9).
Each field type has an associated data type, which is also selectable from a drop-down list (see Figure 2-10).
We can define default values for fields in a class schema. These will be inherited by all instances created from the class but can be optionally overridden in the schema of any particular instance.
One of the schema field types is a relationship, which links to other instances elsewhere in the Automate Datastore. We often use relationships as a way of chaining instances together, and relationship values can accept variable substitutions for flexibility (see Figure 2-11).
An instance is a specific instantiation or “clone” of the generic class and is the entity run by the Automation Engine. An instance contains a copy of the class schema but with actual values of the fields filled in (see Figure 2-12).
A method is a self-contained block of Ruby code that gets executed when we run any automation operation. A typical method looks like this:
#
# Description: This method checks to see if the VM has been powered off or
# suspended
#
# Get vm from root object
vm
=
$evm
.
root
[
'vm'
]
if
vm
power_state
=
vm
.
attributes
[
'power_state'
]
ems
=
vm
.
ext_management_system
$evm
.
log
(
'info'
,
"VM:<
#{
vm
.
name
}
> on provider:<
#{
ems
.
try
(
:name
)
}
has Power
State:<
#{
power_state
}
>"
)
# If VM is powered off or suspended exit
if
%w(off suspended)
.
include?
(
power_state
)
# Bump State
$evm
.
root
[
'ae_result'
]
=
'ok'
elsif
power_state
==
"never"
# If never then this VM is a template so exit the retirement state machine
$evm
.
root
[
'ae_result'
]
=
'error'
else
$evm
.
root
[
'ae_result'
]
=
'retry'
$evm
.
root
[
'ae_retry_interval'
]
=
'60.seconds'
end
end
Methods can have one of three location values: inline, built-in, or URI. In practice most of the methods that we create are inline methods, which means they run as a separate Ruby process outside of Rails.
3.145.172.146