[MESOS-343] Expose TASK_FAILED reason to Frameworks.

• Type: Story
• Status: Resolved
• Priority: Minor
• Resolution: Fixed
• Fix Version/s: 0.21.0
• Sprint: Twitter Mesos Q4 Sprint 3
• Story Points: 8
• Assignee: Dominic Hamon
• Reporter: Benjamin Mahler
• Created: 04/Feb/13 18:27
• Updated: 20/Mar/15 07:24
• Resolved: 04/Nov/14 22:20

Description

We now have a message string inside TaskStatus that provides human readable information about TASK_FAILED.

It would be good to add some structure to the failure reasons, for framework schedulers to act on programmatically.

E.g.

enum TaskFailure {
EXECUTOR_OOM;
EXECUTOR_OUT_OF_DISK;
EXECUTOR_TERMINATED;
SLAVE_LOST;
etc..
}

Issue Links

• is related to: MESOS-1143 Add a TASK_ERROR task status.
• relates to: MESOS-1830 Expose master stats differentiating between master-generated and slave-generated LOST tasks
• relates to: MESOS-1930 Expose TASK_KILLED reason.
• relates to: MESOS-2035 Add reason to containerizer proto Termination
• relates to: MESOS-2522 Add reason field for framework errors