2 The AutoYaST control file #
2.1 Introduction #
A control file, also known as a profile, is a configuration description for a single system. It consists of sets of resources with properties including support for complex structures such as lists, records, trees and large embedded or referenced objects.
2.2 Format #
The XML configuration format provides a consistent file structure, which is easy to learn and to remember when attempting to configure a new system.
The AutoYaST control file uses XML to describe the system installation and
configuration. XML is a commonly used markup, and many users are familiar
with the concepts of the language and the tools used to process XML files.
If you edit an existing control file, or create a new control file, it
is strongly recommended to validate the control
file. This can be done using a validating XML parser such as
xmllint or jing, for example (see
Section 3.2, “Creating/editing a control file manually”).
The following example shows a control file in XML format:
<?xml version="1.0"?>
<!DOCTYPE profile>
<profile
xmlns="http://www.suse.com/1.0/yast2ns"
xmlns:config="http://www.suse.com/1.0/configns">
<partitioning config:type="list">
<drive>
<device>/dev/sda</device>
<partitions config:type="list">
<partition>
<filesystem config:type="symbol">btrfs</filesystem>
<size>10G</size>
<mount>/</mount>
</partition>
<partition>
<filesystem config:type="symbol">xfs</filesystem>
<size>120G</size>
<mount>/data</mount>
</partition>
</partitions>
</drive>
</partitioning>
<scripts>
<pre-scripts>
<script>
<interpreter>shell</interpreter>
<filename>start.sh</filename>
<source>
<![CDATA[
#!/bin/sh
echo "Starting installation"
exit 0
]]>
</source>
</script>
</pre-scripts>
</scripts>
</profile>2.3 Structure #
Below is an example of a basic control file container, the actual content of which is explained later on in this chapter.
<?xml version="1.0"?> <!DOCTYPE profile> <profile xmlns="http://www.suse.com/1.0/yast2ns" xmlns:config="http://www.suse.com/1.0/configns"> <!-- RESOURCES --> </profile>
The <profile> element (root node) contains one or
more distinct resource elements. The permissible resource elements are
specified in the schema files
2.3.1 Resources and properties #
A resource element either contains multiple and distinct property and resource elements, or multiple instances of the same resource element, or it is empty. The permissible content of a resource element is specified in the schema files.
A property element is either empty or contains a literal value. The permissible property elements and values in each resource element are specified in the schema files
An element can be either a container of other elements (a resource) or it has a literal value (a property); it can never be both. This restriction is specified in the schema files. A configuration component with more than one value must either be represented as an embedded list in a property value or as a nested resource.
An empty element, such as <foo></foo> or
<bar/>, will not be in the
parsed data model. Usually this is interpreted as wanting a sensible
default value. In cases where you need an explicitly empty string instead,
use a CDATA section:
<foo><![CDATA[]]></foo>.
2.3.2 Nested resources #
Nested resource elements allow a tree-like structure of configuration components to be built to any level.
There are two kinds of nested resources: maps and lists. Maps, also known as associative arrays, hashes, or dictionaries, contain mixed contents, identified by their tag names. Lists, or arrays, have all items of the same type.
...
<drive>
<device>/dev/sda</device>
<partitions config:type="list">
<partition>
<size>10G</size>
<mount>/</mount>
</partition>
<partition>
<size>1G</size>
<mount>/tmp</mount>
</partition>
</partitions>
</drive>
....
In the example above, the drive resource is a map
consisting of a device property and a
partitions resource. The partitions
resource is a list containing multiple instances of the
partition resource. Each partition
resource is a map containing a size and
mount property.
The default type of a nested resource is map, although you can specify it
as you want. Lists must be marked as such using the
config:type="list" attribute.
Starting with SUSE Linux Enterprise Micro
5.0, it is
possible to use the attribute t instead of
config:type to specify the element type.
<mode t="boolean">true</mode>
2.3.3 Attributes #
Global attributes are used to define metadata on resources and properties. Attributes are used to define context switching. They are also used for naming and typing properties as shown in the previous sections. Attributes are in a separate namespace so they do not need to be treated as reserved words in the default namespace.
The config:type attribute determines the type of the
resource or property in the parsed data model. For resources, lists need a
list type whereas a map is the default type that does
not need an attribute. There is one exception. When the map is empty, it
needs to be marked as a map so it does not get confused with a simple
string value.
<general t="map" />
For properties, boolean, symbol, and
integer can be used, the default being a string.
Except for map and string values, as explained before, attributes are not
optional. It may appear that attributes are optional, because various parts
of the schema are not very consistent in their usage of data types. In some
places an enumeration is represented by a symbol, elsewhere a string is
required. One resource needs config:type="integer",
another will parse the number from a string property. Some resources use
config:type="boolean", others want
yes or even 1. If in doubt, consult
the schema file.