Details
-
Bug
-
Resolution: Fixed
-
Critical
-
6.0.0
-
Untriaged
-
Unknown
-
CX Sprint 118
Description
To avoid running out of memory and resources contention, we have to make sure that the cluster maximum capacity is never exceeded. Currently it is possible for multiple code paths to trigger a job clean up that will result on returning the capacity of an aborted job multiple times. We should make sure that the capacity of any terminated job is released only once.
Attachments
Issue Links
- links to