Within the PCP Execution Protocol layer, controllers invoke agent capabilities by requesting actions. This is done by sending request messages to one or more agents. Such messages specify the action name, module and input parameters.
Actions related to a given application are grouped together in modules; for
instance, a puppet module may provide the run and status actions.
Action requests can be blocking or non-blocking; that changes the way the action outcome is referred and stored. Please refer to the Request/Response and Transaction Status sections for more details.
To inform the requester controller about the outcome of the requested action, the agent replies with a response message.