When receiving a drain request the agent will attempt to kill all tasks. It should also prevent launching new tasks to deal with possible (future) race scenarios where a task launch arrives after a drain request.
It seems on place one could insert such a check is Slave::__run.
When receiving a drain request the agent will attempt to kill all tasks. It should also prevent launching new tasks to deal with possible (future) race scenarios where a task launch arrives after a drain request.
It seems on place one could insert such a check is Slave::__run .