Commands that target multiple agents/hosts must specify the behavior when the command fail for some of those hosts (ex: due to state transitions like shutting down, etc)
The options are:
- continue (default): proceed with the command and log warnings for any hosts where the command did not succeed
- abort: abort the command as soon as a command fails on any host
- retry-until-success [future]: keep retrying until all hosts have succeeded