Hi all, Currently, Mesos Docker executor treats Docker task as TASK_FAILED on Docker daemon restart. It causes problems for operators with cluster maintenance. Starting with Docker 1.12, one can configure the daemon so that containers remain running if the daemon becomes unavailable.
We're proposing an improvement for the Mesos Docker executor to address this problem. The current design doc is: https://docs.google.com/document/d/1JeLTr9L31S8eIg-6xpjedIUKvnfNake0kPTzxEwdUdI/
