While a job is running, the machine goes to an offline / unavailable status but, the scheduler still tries to execute killjob and max_run_events against the job. The killjob fails as the machine is unreachable. The user would like autosys to delay processing the events or retry them once the machine comes back online.
As of AutoSys 12.1 sp1 there are no settings that would cause the killjob or max_run_alarm to be delayed or re-queued as a result of a machine being offline.