[MESOS-5322] Add "reason" field to FAILURE event in HTTP scheduler API

• Type: Improvement
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Component/s: HTTP API
• Labels: mesosphere
• Epic Link: Scheduler API v1 improvements
• Assignee: Unassigned
• Reporter: Neil Conway
• Created: 03/May/16 20:36
• Updated: 03/May/16 20:53

Description

Right now, the FAILURE event doesn't provide much context about why the slave was removed and/or why the executor terminated. A reason field, similar to TaskStatus.reason, would help users differentiate (e.g., between slaves that are removed for maintenance versus those that are removed because of a network partition; or between the various different executor termination scenarios).

Issue Links

• is related to: MESOS-4548 Errors communicated to the scheduler should be associated with stable error codes.