Bug 1217097

The recent VM Retirement enhancements could cause issues with existing customers using the old style automate model who wish to use the old style retirement statemachine. The following directions are intended to resolve this issue.

Backward Compatibility:

With the exception of policy, modifications to the automate model would make it possible to use the old style retirement state machine:

The previous VM retirement functionality:
* Set the VM retired flag 
* Set the VM retire_date value
* Stopped a running VM
* Called into the retirement state machine

Because a major feature of the new retirement is that it exposes the functionality to automate, backward compatibility can be achieved by adding an automate method to the customers existing state machine that would accomplish the tasks previously done in code. 

This same functionality can be attained by the following:

Modify the state machine to use the old retirement state machine

For the following directions, the domain references of “CUSTOMER” means the customer domain containing the old automate model which has a higher priority than the ManageIQ shipped domain. 

1. Add new event instance, “request_vm_retire” event to /system/event class in the CUSTOMER domain.
  a. Set logical event to “vm_retired”
  b. Set rel5 value to point to the same value that is set in the /system/event/vm_retired rel5 relationship in the CUSTOMER domain 
  *Note - value is most likely to be: /Automation/VMLifecycle/Retirement?vm_id=${process#vm_id}

2. Change schema for CUSTOMER domain:
  /Factory/StateMachines/VMRetirement class to add a new entry
  name:  preretirement
  type:    state
  datatype: string
  subst flag = true
  message: create
  on_entry: update_retirement_status(status => ’Starting preretirement’)
  on_exit: update_retirement_status(status => ‘Ended preretirement’)
  on_error: update_retirement_status(status => 'Error in preretirement’)

3. Change the schema sequence, if necessary. The new preretirement state should be the first state in the state machine.

4. Change /Factory/StateMachines/VMRetirement/Default instance:
   add preretirement value of: /Factory/VM/Preretirement

5. Add instance:
   /Factory/VM/Preretirement
   set execute attribute value to: preretirement

6. Add method:
   /Factory/VM/preretirement
   with the following content:
   vm = $evm.root[‘vm’]
   vm.finish_retirement if vm

Note* It appears backwards that we're calling the finish_retirement method in the preretirement method.  The reason for this is that the previous retirement process immediately set the retired flag and retired_on date.  The enhanced retirement process waits until the retirement is finished before it sets the retired and retires_on attributes.  By using the finish_retirement method, we reverting back to the old behavior so that the customers existing retirement state machine can proceed as before.