Alex Withrow created MESOS-6381:

             Summary: Support for Docker's --live-restore setting
                 Key: MESOS-6381
             Project: Mesos
          Issue Type: Wish
          Components: docker
            Reporter: Alex Withrow

Docker after 1.12.0 has a --live-restore which allows for containers to 
continue to run when the docker daemon itself goes down (such as upgrades). It 
would be great if mesos supported this feature when running tasks in docker. 
Currently if the docker daemon goes down, all tasks running in docker 
containers are reported FAILED and are rescheduled.

Perhaps the addition of a timeout setting before ending the tasks would allow 
this to work? Having this would allow for upgrades of the docker daemon without 
having to reschedule the current tasks similar to how mesos agent upgrades work.

This message was sent by Atlassian JIRA

Reply via email to