Details
-
Task
-
Resolution: Fixed
-
Critical
-
None
-
None
-
4: Deliver Training, More 2.2
-
5
Description
Now we have auto allocation of resource requests based on memory quotas, there are a few improvements that should be made:
- Opt in (needs a configuration switch)
- Updates (changes the requests when the memory quotas change)
- Overhead specification (+25% is/could be a ridiculous waste)
- Query memory specification (you cannot specify query memory, so a query-only pod will allocate nothing, this scares me due to OOM killing)
Attachments
Issue Links
- relates to
-
K8S-2048 Documentation for minor enhancements/changes in CAO 2.2
- In Progress