Details
-
Improvement
-
Status: Resolved
-
Major
-
Resolution: Duplicate
-
None
-
None
-
None
-
None
-
Mesosphere Sprint 38
-
5
Description
Right now, a task is called orphaned if an agent re-registers with it but the corresponding framework information is not known to the master. This happens immediately after a master failover.
It would great if the master knows the information about the framework even after a failover, irrespective of whether a framework re-registers, so that we don't have orphan tasks. Getting rid of orphan tasks will make the task authorization story easy (see MESOS-5757).
Attachments
Issue Links
- duplicates
-
MESOS-4659 Avoid leaving orphan task after framework failure + master failover
- Accepted
- is related to
-
MESOS-1719 Master should persist framework information
- Accepted
- relates to
-
MESOS-5757 Authorize orphaned tasks
- Resolved