As defined in the point (c) and (d) of bug #616504, the Provisioning Agent must observe the Machine Agent status, and in case it disconnects from Ensemble and does not reconnect under a specified amount of time, the machine it is responsible for should be shut down (if needed; the machine might have died already), and then a new machine should be started to handle the workload.
This work depends on the observing extensions of the MachineStateManager which are defined in bug #616559.
As defined in the point (c) and (d) of bug #616504, the Provisioning Agent must observe the Machine Agent status, and in case it disconnects from Ensemble and does not reconnect under a specified amount of time, the machine it is responsible for should be shut down (if needed; the machine might have died already), and then a new machine should be started to handle the workload.
This work depends on the observing extensions of the MachineStateManager which are defined in bug #616559.