Currently the execjob_prologue runs on mother superior and then launches the job. When a job does a tm_spawn, tm_attach, or a pbs_attach the the attach will run first and then the execjob_prologue. If the node encounters an issue with the execjob_prologue all of the work the job has done up until that point is wasted.
The behavior should be changed such that the execjob_prologue is run on every node before the job is launched, similar to the execjob_begin behavior.
execjob_prologue gets run after execjob_begin, but dues not wait for actual job processes to be spawned on secondary hosts