The agent would be run on a set of existing log files but would not generate any items. Once it exits, it will resume creating items.

--dry-run doesn't save the state of the agent. Maybe we should add an option to dry-run which does?

Requested by Jay