Details
-
New Feature
-
Status: Open
-
Major
-
Resolution: Unresolved
-
None
-
None
-
None
Description
Currently one query that spills excessively to disk can use up all cluster space and cause other queries/workloads to fail. I understand that a global maximum would be difficult to track, but some limit per mapper/reducer would help keep the spill under control.
Attachments
Issue Links
- Is contained by
-
HIVE-17906 use kill query mechanics to kill queries in WM
-
- Closed
-