-
Epic
-
Resolution: Done
-
Critical
-
None
-
cnv-better-eviction
-
False
-
-
False
-
- MUST Descheduler handles VMs correctly
- MUST the problem discovered by Robert K does not appear anymore
-
Green
-
To Do
-
CNV-8280 - VM balancing
-
0% To Do, 0% In Progress, 100% Done
-
dev-ready, doc-ready, po-ready, qe-ready, ux-ready
-
-
---
-
---
Goal
Today eviction is called on a PDB protected launcher, "rejected", but kubevirt is starting the live migration nevertheless (CNV-28280), this is confusing i.e. the descheduler (OCPBUGS-11891).
This epic is about exploring a cooperative effort between the descheduer and CNV to mitigate this. CNV's responsibilites are outlined here: https://issues.redhat.com/browse/CNV-28280#comment-23825049
User Stories
- As a cluster administrator, I want to leverage the descheduler to balance workloads without threatening the stability of the cluster so that services can be provided to tenants and customers without disruption from routine activites such as upgrade or load balancing.
Non-Requirements
- List of things not included in this epic, to alleviate any doubt raised during the grooming process.
Notes
- An eviction level solution is being explored in CNV-35196
Done Checklist
| Who | What | Reference |
|---|---|---|
| DEV | Upstream roadmap issue (or individual upstream PRs) | https://github.com/kubevirt/kubevirt/pull/12025 |
| DEV | Upstream documentation merged | |
| DEV | gap doc updated | |
| DEV | Upgrade consideration | |
| DEV | CEE/PX summary presentation | |
| QE | Test plans in Polarion | https://docs.google.com/document/d/12R0zQeWolYOMVQ3yVaDsBrDFPzkxSJ6VOmStDTbEVRE/edit#heading=h.ikw26xd2gfs8 |
| QE | Automated tests merged | |
| DOC | Downstream documentation merged | Tracked in |
- blocks
-
-
- In Progress
-
- depends on
-
-
- Closed
-
- is depended on by
-
CNV-50724 [enterprise-4.17] Remove TP note Issue in file virt/virtual_machines/advanced_vm_management/virt-enabling-descheduler-evictions.adoc
-
- Closed
-
- is documented by
-
-
- Closed
-
-
-
- Closed
-
- is related to
-
CNV-30684 [2219785] With cluster-level evictionStrategy:LiveMigrate set in HCO CR certain VMs would fail to restart and get stuck during node drain
-
- Closed
-
- relates to
-
RFE-5564 descheduler: GA DevPreviewLongLifecycle
-
- Accepted
-
-
-
- Closed
-
- links to