Use a selector to choose the server for a project or step.
A selector describes the kind of server that is appropriate for the project
or step. It can specify a server directly (by name) or indirectly (by an attribute
collected by a collector and stored in the manifest).
The following are common best practices for setting up selectors:
- By Name: create selectors named after each of your servers which specify
the server by its BF_NAME (the unique name used in the Management Console).
Choose these selectors when you want to be specific about your choice.
- By Server Pool: if you organize servers into named pools, create a collector
for each pool that assigns the pool name as a property. Then create selectors
for each pool name.
- By Server Attributes: you can choose servers based on their real functional
properties, such as available hard-disk space, operating system, or number
of CPUs. To implement dynamic selection, first create collectors that collect
and assign appropriate properties. Assign the collectors to the appropriate
servers. Finally, create a selector for each property or set of properties
that might influence your choice of machine.
- A selector for machines whose operating system name includes "Windows®"
- A selector for machines with more than one CPU
- A selector for machines running at less than a specified load
Each selector chooses from all servers that match its criteria.
- Nested collectors: you can use .include properties to nest collectors
within one another. You may want to create individual collectors for each
server, for example, so that each server can have some unique properties that
you specify, while using .include to include some utility collectors. For
example, you might create a collector called Version that issues commands
to check the version numbers for some key tools in your environment, such
as Perl and Java, and then include that collector in all of your server-specific
collectors.